Peter Wang: Python and the Source Code of Humans, Computers, and Reality

00:00:00.000 | The following is a conversation with Peter Wang,

00:00:02.400 | one of the most impactful leaders and developers

00:00:04.640 | in the Python community,

00:00:06.320 | former physicist, current philosopher,

00:00:09.040 | and someone who many people told me about

00:00:11.640 | and praised as a truly special mind

00:00:14.180 | that I absolutely should talk to.

00:00:16.480 | Recommendations ranging from Travis Oliphant

00:00:19.440 | to Eric Weinstein.

00:00:20.920 | So here we are.

00:00:23.320 | This is the Lex Friedman Podcast.

00:00:25.600 | To support it, please check out our sponsors

00:00:27.680 | in the description.

00:00:28.800 | And now, here's my conversation with Peter Wang.

00:00:32.400 | You're one of the most impactful humans

00:00:35.640 | in the Python ecosystem.

00:00:37.120 | (laughs)

00:00:38.240 | So you're an engineer, leader of engineers,

00:00:40.840 | but you're also a philosopher.

00:00:42.880 | So let's talk both in this conversation

00:00:45.120 | about programming and philosophy.

00:00:47.280 | First, programming.

00:00:48.960 | What to you is the best

00:00:51.120 | or maybe the most beautiful feature of Python,

00:00:54.040 | or maybe the thing that made you fall in love

00:00:56.160 | or stay in love with Python?

00:00:59.000 | - Well, those are three different things.

00:01:00.680 | What I think is the most beautiful,

00:01:01.960 | what made me fall in love with me, stay in love.

00:01:03.960 | When I first started using it

00:01:05.760 | was when I was a C++ computer graphics performance nerd.

00:01:10.040 | - In the '90s?

00:01:10.880 | - Yeah, in late '90s.

00:01:12.060 | And that was my first job out of college.

00:01:14.120 | And we kept trying to do more and more like abstract

00:01:18.680 | and higher order programming in C++,

00:01:20.520 | which at the time was quite difficult with templates.

00:01:24.160 | The compiler support wasn't great, et cetera.

00:01:26.560 | So when I started playing around with Python,

00:01:28.720 | that was my first time encountering

00:01:30.480 | really first-class support for types,

00:01:32.440 | for functions and things like that.

00:01:34.320 | And it felt so incredibly expressive.

00:01:37.200 | So that was what kind of made me fall in love

00:01:39.120 | with it a little bit.

00:01:39.960 | And also, once you spend a lot of time

00:01:42.280 | in a C++ dev environment,

00:01:44.160 | the ability to just whip something together

00:01:46.200 | that basically runs and works the first time is amazing.

00:01:49.680 | So really productive scripting language.

00:01:51.960 | I mean, I knew Perl, I knew Bash.

00:01:53.680 | I was decent at both, but Python just made everything,

00:01:57.160 | it made the whole world accessible, right?

00:01:59.680 | I could script this and that and the other network things,

00:02:02.600 | you know, little hard drive utilities.

00:02:04.000 | I could write all of these things

00:02:05.080 | in the space of an afternoon.

00:02:06.560 | And that was really, really cool.

00:02:07.640 | So that's what made me fall in love.

00:02:08.640 | - Is there something specific you could put your finger on

00:02:11.560 | that you're not programming in Perl today?

00:02:14.080 | Like why Python for scripting?

00:02:17.120 | - I think there's not a specific thing

00:02:19.640 | as much as the design motif of both the creator

00:02:22.800 | of the language and the core group of people

00:02:25.320 | that built the standard library around him.

00:02:27.480 | There was definitely, there was a taste to it.

00:02:32.200 | I mean, Steve Jobs, you know, used that term,

00:02:34.120 | you know, in somewhat of a arrogant way,

00:02:35.720 | but I think it's a real thing that it was designed to fit.

00:02:39.240 | A friend of mine actually expressed this really well.

00:02:40.920 | He said, "Python just fits in my head."

00:02:43.000 | And there's nothing better to say than that.

00:02:45.240 | Now, people might argue modern Python,

00:02:47.920 | there's a lot more complexity,

00:02:49.280 | but certainly as version 152,

00:02:51.800 | I think was my first version,

00:02:53.360 | that fit in my head very easily.

00:02:54.800 | So that's what made me fall in love with it.

00:02:56.560 | - Okay, so the most beautiful feature of Python

00:03:01.400 | that made you stay in love.

00:03:03.040 | It's like over the years, what has like, you know,

00:03:06.560 | you do a double take and you return too often

00:03:09.520 | as a thing that just brings you a smile.

00:03:11.560 | - I really still like the ability to play with metaclasses

00:03:16.560 | and express higher order of things.

00:03:19.000 | When I have to create some new object model

00:03:22.000 | to model something, right?

00:03:23.360 | It's easy for me,

00:03:24.400 | 'cause I'm pretty expert as a Python programmer.

00:03:27.080 | I can easily put all sorts of lovely things together

00:03:29.800 | and use properties and decorators

00:03:31.600 | and other kinds of things

00:03:32.920 | and create something that feels very nice.

00:03:34.680 | So that to me, I would say that's tied with the NumPy

00:03:38.560 | and vectorization capabilities.

00:03:40.640 | I love thinking in terms of the matrices and the vectors

00:03:43.800 | and these kind of data structures.

00:03:46.080 | So I would say those two are kind of tied for me.

00:03:49.400 | - So the elegance of the NumPy data structure,

00:03:52.720 | like slicing through the different multi-dimensional--

00:03:54.800 | - Yeah, there's just enough things there.

00:03:56.200 | It's like a very, it's a very simple, comfortable tool.

00:04:00.040 | Just it's easy to reason about what it does

00:04:02.800 | when you don't stray too far afield.

00:04:05.000 | - Can you put your finger on how to design a language

00:04:10.000 | such that it fits in your head?

00:04:11.880 | Certain things like the colon

00:04:14.040 | or the certain notation aspects of Python

00:04:17.200 | that just kind of work.

00:04:18.680 | Is it something you have to kind of write out on paper,

00:04:22.240 | look and say, it's just right?

00:04:24.680 | Is it a taste thing or is there a systematic process?

00:04:27.640 | What's your sense?

00:04:28.840 | - I think it's more of a taste thing.

00:04:31.560 | But one thing that should be said

00:04:33.600 | is that you have to pick your audience, right?

00:04:36.400 | So the better defined the user audience is

00:04:39.200 | or the users are,

00:04:40.440 | the easier it is to build something

00:04:42.320 | that fits in their minds

00:04:43.840 | because their needs will be more compact and coherent.

00:04:47.280 | It is possible to find a projection, right?

00:04:49.200 | A compact projection for their needs.

00:04:50.960 | The more diverse the user base, the harder that is.

00:04:54.520 | And so as Python has grown in popularity,

00:04:57.200 | that's also naturally created more complexity

00:05:00.160 | as people try to design any given thing.

00:05:01.840 | There'll be multiple valid opinions

00:05:04.200 | about a particular design approach.

00:05:06.320 | And so I do think that's the downside of popularity.

00:05:10.280 | It's almost an intrinsic aspect

00:05:11.520 | of the complexity of the problem.

00:05:13.120 | - Well, at the very beginning,

00:05:14.480 | aren't you an audience of one?

00:05:16.240 | Isn't ultimately,

00:05:17.480 | aren't all the greatest projects in history

00:05:19.520 | were just solving a problem that you yourself had?

00:05:21.880 | - Well, so Clay Shirky in his book on crowdsourcing

00:05:25.440 | or in his kind of thoughts on crowdsourcing,

00:05:27.560 | he identifies the first step of crowdsourcing

00:05:29.600 | is me first collaboration.

00:05:31.520 | You first have to make something

00:05:32.600 | that works well for yourself.

00:05:34.320 | It's very telling that when you look at all of the impactful

00:05:37.760 | big project, well, they're fundamental projects now

00:05:40.040 | in the SciPy and PyData ecosystem,

00:05:42.600 | they all started with the people in the domain

00:05:46.760 | trying to scratch their own itch.

00:05:48.240 | And the whole idea of scratching your own itch

00:05:49.760 | is something that the open source

00:05:51.360 | or the free software world has known for a long time.

00:05:53.680 | But in the scientific computing areas,

00:05:56.840 | these are assistant professors

00:05:58.320 | or electrical engineering grad students.

00:06:00.560 | They didn't have really a lot of programming skill

00:06:03.040 | necessarily, but Python was just good enough

00:06:05.560 | for them to put something together

00:06:06.800 | that fit in their domain, right?

00:06:09.440 | So it's almost like it's a necessity

00:06:12.440 | as a mother of invention aspect.

00:06:13.960 | And also it was a really harsh filter

00:06:16.880 | for utility and compactness and expressiveness.

00:06:20.880 | Like it was too hard to use,

00:06:22.400 | then they wouldn't have built it

00:06:23.360 | 'cause there was just too much trouble, right?

00:06:24.960 | It was a side project for them.

00:06:26.160 | - And also necessity creates a kind of deadline.

00:06:28.120 | It seems like a lot of these projects

00:06:29.440 | are quickly thrown together in the first step.

00:06:32.320 | And that, even though it's flawed,

00:06:34.520 | that just seems to work well for software projects.

00:06:38.840 | - Well, it does work well for software projects in general.

00:06:41.280 | And in this particular space,

00:06:43.280 | one of my colleagues, Stan Sieber, identified this,

00:06:46.320 | that all the projects in the SciPy ecosystem,

00:06:49.040 | if we just rattle them off, there's NumPy, there's SciPy,

00:06:53.040 | built by different collaborations of people,

00:06:55.000 | although Travis is the heart of both of them.

00:06:57.680 | But NumPy coming from Numeric and NumRay,

00:06:59.360 | these are different people.

00:07:00.640 | And then you've got Pandas, you've got Jupyter or IPython.

00:07:04.480 | There's Matplotlib.

00:07:06.880 | There's just so many others, I'm not gonna do justice

00:07:09.680 | if I try to name them all,

00:07:10.720 | but all of them are actually different people.

00:07:12.800 | And as they rolled out their projects,

00:07:15.280 | the fact that they had limited resources

00:07:17.560 | meant that they were humble about scope.

00:07:19.560 | A great famous hacker, Jamie Zawiski,

00:07:23.560 | once said that every geek's dream

00:07:26.040 | is to build the ultimate middleware, right?

00:07:29.280 | And the thing is with these scientists turned programmers,

00:07:32.280 | they had no such thing.

00:07:33.120 | They were just trying to write something

00:07:34.800 | that was a little bit better for what they needed,

00:07:36.560 | the MATLAB, and they were gonna leverage

00:07:38.560 | what everyone else had built.

00:07:39.960 | So naturally, almost in kind of this annealing process

00:07:42.680 | or whatever, we built a very modular cover

00:07:46.520 | of the basic needs of a scientific computing library.

00:07:50.320 | - If you look at the whole human story,

00:07:51.960 | how much of a leap is it?

00:07:53.720 | We've developed all kinds of languages,

00:07:55.520 | all kinds of methodologies for communication,

00:07:57.760 | and it just kind of like grew this collective intelligence,

00:08:00.600 | the civilization grew, it expanded,

00:08:02.880 | wrote a bunch of books, and now we tweet.

00:08:06.460 | How big of a leap is programming

00:08:08.660 | if programming is yet another language?

00:08:10.820 | Is it just a nice little trick

00:08:12.860 | that's temporary in our human history,

00:08:14.980 | or is it like a big leap in the,

00:08:18.780 | almost us becoming another organism

00:08:23.100 | at a higher level of abstraction, something else?

00:08:26.180 | - I think the act of programming

00:08:28.260 | or using grammatical constructions

00:08:32.380 | of some underlying primitives,

00:08:34.940 | that is something that humans do learn,

00:08:37.520 | but every human learns this.

00:08:38.860 | Anyone who can speak learns how to do this.

00:08:41.160 | What makes programming different

00:08:42.420 | has been that up to this point,

00:08:44.860 | when we try to give instructions to computing systems,

00:08:49.000 | all of our computers, well, actually, this is not quite true,

00:08:51.580 | but I'll first say it,

00:08:53.080 | and then I'll tell you why it's not true.

00:08:54.980 | But for the most part, we can think of computers

00:08:56.800 | as being these iterated systems.

00:08:58.900 | So when we program, we're giving very precise instructions

00:09:02.740 | to iterated systems that then run at incomprehensible speed

00:09:07.380 | and run those instructions.

00:09:08.820 | In my experience, some people are just better equipped

00:09:12.700 | to model systematic iterated systems,

00:09:16.860 | well, whatever, iterated systems in their head.

00:09:19.220 | Some people are really good at that,

00:09:21.740 | and other people are not.

00:09:23.780 | And so when you have, like, for instance,

00:09:26.060 | sometimes people have tried to build systems

00:09:27.640 | that make programming easier

00:09:29.300 | by making it visual drag and drop.

00:09:31.260 | And the issue is you can have a drag and drop thing,

00:09:33.700 | but once you start having to iterate the system

00:09:35.220 | with conditional logic, handling case statements

00:09:37.140 | and branch statements and all these other things,

00:09:39.540 | the visual drag and drop part doesn't save you anything.

00:09:42.080 | You still have to reason about this giant iterated system

00:09:44.720 | with all these different conditions around it.

00:09:46.420 | That's the hard part, right?

00:09:48.220 | So handling iterated logic, that's the hard part.

00:09:52.300 | The languages we use then emerge to give us ability

00:09:55.360 | and capability over these things.

00:09:57.300 | Now, the one exception to this rule, of course,

00:09:58.900 | is the most popular programming system in the world,

00:10:00.700 | which is Excel, which is a data flow

00:10:03.300 | and a data-driven, immediate mode,

00:10:05.420 | data transformation-oriented programming system.

00:10:08.380 | And it's actually not an accident

00:10:10.400 | that that system is the most popular programming system

00:10:12.840 | because it's so accessible

00:10:14.420 | to a much broader group of people.

00:10:16.940 | I do think as we build future computing systems,

00:10:21.260 | you're actually already seeing this a little bit,

00:10:22.900 | it's much more about composition of modular blocks.

00:10:25.660 | They themselves actually maintain all their internal state,

00:10:29.620 | and the interfaces between them

00:10:31.060 | are well-defined data schemas.

00:10:32.980 | And so to stitch these things together using like IFTTT

00:10:35.900 | or Zapier or any of these kind of,

00:10:38.780 | I would say compositional scripting kinds of things,

00:10:41.980 | I mean, HyperCard was also a little bit in this vein.

00:10:45.020 | That's much more accessible to most people.

00:10:47.620 | It's really that implicit state

00:10:49.980 | that's so hard for people to track.

00:10:52.060 | - Yeah, okay, so that's modular stuff,

00:10:53.460 | but there's also an aspect

00:10:54.460 | where you're standing on the shoulders of giants.

00:10:55.940 | So you're building like higher and higher levels

00:10:58.740 | of abstraction, but you do that a little bit with language.

00:11:02.180 | So with language, you develop sort of ideas,

00:11:05.180 | philosophies from Plato and so on.

00:11:07.580 | And then you kind of leverage those philosophies

00:11:09.540 | as you try to have deeper and deeper conversations.

00:11:12.660 | But with programming,

00:11:13.540 | it seems like you can build much more complicated systems.

00:11:17.060 | Like without knowing how everything works,

00:11:18.940 | you can build on top of the work of others.

00:11:21.340 | And it seems like you're developing

00:11:22.860 | more and more sophisticated expressions,

00:11:27.860 | ability to express ideas in a computational space.

00:11:31.300 | - I think it's worth pondering the difference here

00:11:35.020 | between complexity and complication.

00:11:38.620 | Okay, right? - Back to Excel.

00:11:42.220 | - Well, not quite back to Excel,

00:11:43.260 | but the idea is when we have a human conversation,

00:11:47.540 | all languages for humans emerged

00:11:51.220 | to support human relational communications,

00:11:55.540 | which is that the person we're communicating with

00:11:57.700 | is a person and they would communicate back to us.

00:12:01.220 | And so we sort of hit a resonance point, right?

00:12:05.620 | When we actually agree on some concepts.

00:12:07.460 | So there's a messiness to it and there's a fluidity to it.

00:12:10.260 | With computing systems,

00:12:11.860 | when we express something to the computer and it's wrong,

00:12:14.220 | we just try again.

00:12:15.380 | So we can basically live many virtual worlds

00:12:17.500 | of having failed at expressing ourselves to the computer

00:12:20.180 | until the one time we expressed ourselves right.

00:12:22.820 | Then we kind of put in production

00:12:23.940 | and then discover that it's still wrong,

00:12:25.900 | a few days down the road.

00:12:27.140 | So I think the sophistication of the things

00:12:30.460 | that we build with computing,

00:12:32.460 | one has to really pay attention to the difference

00:12:35.580 | between when an end user is expressing something

00:12:38.180 | onto a system that exists

00:12:39.940 | versus when they're extending the system

00:12:42.460 | to increase the system's capability

00:12:45.420 | for someone else to then interface with.

00:12:47.420 | And we happen to use the same language

00:12:48.620 | for both of those things.

00:12:49.500 | And in most cases, but it doesn't have to be that.

00:12:52.060 | And Excel is actually a great example of this,

00:12:54.340 | of kind of a counterpoint to that.

00:12:56.220 | - Okay, so what about the idea of, you said messiness.

00:13:01.220 | Wouldn't you put the software 2.0 idea,

00:13:06.220 | this idea of machine learning

00:13:07.740 | into the further and further steps

00:13:12.500 | into the world of messiness.

00:13:14.500 | The same kind of beautiful messiness

00:13:16.060 | of human communication.

00:13:17.580 | Isn't that what machine learning is?

00:13:19.100 | Is building on levels of abstraction

00:13:23.540 | that don't have messiness in them,

00:13:25.500 | that at the operating system level,

00:13:27.380 | then there's Python, the programming languages

00:13:29.380 | that have more and more power.

00:13:30.860 | But then finally there's neural networks

00:13:34.580 | that ultimately work with data.

00:13:38.140 | And so the programming is almost in the space of data

00:13:40.540 | and the data is allowed to be messy.

00:13:42.420 | Isn't that a kind of program?

00:13:43.860 | So the idea of software 2.0

00:13:45.900 | is a lot of the programming happens in the space of data.

00:13:50.900 | So back to Excel, all roads lead back to Excel.

00:13:54.060 | In the space of data and also the hyperparameters

00:13:56.540 | of the neural networks.

00:13:57.380 | And all of those allow the same kind of messiness

00:14:02.220 | that human communication allows.

00:14:04.380 | - It does, but my background is a physics.

00:14:07.780 | I took like two CS courses in college.

00:14:09.940 | So I don't have, now I did cram a bunch of CS in prep

00:14:13.820 | when I applied for grad school,

00:14:15.500 | but still I don't have a formal background

00:14:18.140 | in computer science.

00:14:19.740 | But what I have observed in studying programming languages

00:14:22.500 | and programming systems and things like that

00:14:24.980 | is that there seems to be this triangle.

00:14:27.340 | It's one of these beautiful little iron triangles

00:14:30.420 | that you find in life sometimes.

00:14:32.060 | And it's the connection between the code correctness

00:14:35.620 | and kind of expressiveness of code,

00:14:37.420 | the semantics of the data,

00:14:39.900 | and then the kind of correctness or parameters

00:14:42.500 | of the underlying hardware compute system.

00:14:44.940 | So there's the algorithms that you wanna apply.

00:14:48.460 | There's what the bits that are stored

00:14:51.420 | on whatever media actually represent.

00:14:53.340 | So the semantics of the data within the representation,

00:14:56.940 | and then there's what the computer can actually do.

00:14:59.740 | And every programming system, every information system

00:15:02.860 | ultimately find some spot

00:15:05.020 | in the middle of this little triangle.

00:15:07.420 | Sometimes some systems collapse them into just one edge.

00:15:11.100 | - Are we including humans as a system in this?

00:15:13.500 | - No, no, I'm just thinking about computing systems here.

00:15:15.540 | - Okay. - Okay?

00:15:16.380 | And the reason I bring this up is because

00:15:17.820 | I believe there's no free lunch around this stuff.

00:15:20.100 | So if we build machine learning systems

00:15:22.980 | to sort of write the correct code

00:15:25.500 | that is at a certain level of performance,

00:15:27.140 | so it'll sort of select, right?

00:15:28.900 | With the hyperparameters, we can tune

00:15:31.100 | kind of how we want the performance boundary

00:15:32.860 | and SLA to look like

00:15:34.980 | for transforming some set of inputs

00:15:37.300 | into certain kinds of outputs.

00:15:39.580 | That training process itself is intrinsically sensitive

00:15:43.660 | to the kinds of inputs we put into it.

00:15:45.580 | It's quite sensitive to the boundary conditions

00:15:47.940 | we put around the performance.

00:15:49.340 | So I think even as we move to using automated systems

00:15:52.180 | to build this transformation,

00:15:53.460 | as opposed to humans explicitly

00:15:55.460 | from a top-down perspective,

00:15:56.980 | figuring out, well, this schema and this database

00:15:59.300 | and these columns get selected for this algorithm.

00:16:01.660 | And here we put a Fibonacci heap for some other thing.

00:16:04.860 | Human design or computer design,

00:16:06.780 | ultimately what we hit,

00:16:08.140 | the boundaries that we hit with these information systems

00:16:10.620 | is when the representation of the data hits the real world

00:16:14.220 | is where there's a lot of slop and a lot of interpretation.

00:16:17.540 | And that's where actually I think

00:16:18.940 | a lot of the work will go in the future

00:16:20.860 | is actually understanding kind of how to better

00:16:23.340 | in the view of these live data systems,

00:16:26.220 | how to better encode the semantics of the world

00:16:29.300 | for those things.

00:16:30.140 | There'll be less of the details

00:16:31.060 | of how we write a particular SQL query.

00:16:33.460 | - Okay, but given the semantics of the real world

00:16:35.500 | and the messiness of that,

00:16:36.900 | what does the word correctness mean

00:16:38.620 | when you're talking about code?

00:16:40.780 | - There's a lot of dimensions to correctness.

00:16:42.940 | Historically, and this is one of the reasons I say

00:16:45.140 | that we're coming to the end of the era of software,

00:16:47.620 | 'cause for the last 40 years or so,

00:16:49.820 | software correctness was really defined

00:16:52.860 | about functional correctness.

00:16:54.820 | I write a function, it's got some inputs,

00:16:56.340 | does it produce the right outputs?

00:16:57.820 | If so, then I can turn it on,

00:16:59.220 | hook it up to the live database and it goes.

00:17:01.460 | And more and more now we have,

00:17:03.180 | I mean, in fact, I think the bright line in the sand

00:17:05.100 | between machine learning systems

00:17:06.740 | or modern data-driven systems

00:17:08.140 | versus classical software systems

00:17:10.780 | is that the values of the input

00:17:14.260 | actually have to be considered with the function together

00:17:17.380 | to say this whole thing is correct or not.

00:17:19.420 | And usually there's a performance SLA as well.

00:17:21.780 | Like, did it actually finish making this-

00:17:23.340 | - What's SLA?

00:17:24.180 | - Sorry, service level agreement.

00:17:25.380 | So it has to return within some time.

00:17:27.100 | You have a 10 millisecond time budget

00:17:29.180 | to return a prediction of this level of accuracy, right?

00:17:32.700 | So these are things that were not traditionally

00:17:35.020 | in most business computing systems for the last 20 years

00:17:37.580 | at all, people didn't think about it.

00:17:39.420 | But now we have value dependence on functional correctness.

00:17:42.740 | So that question of correctness

00:17:44.140 | is becoming a bigger and bigger question.

00:17:45.740 | - What does that map to the end of software?

00:17:48.180 | - We've thought about software as just this thing

00:17:50.500 | that you can do in isolation with some test trial inputs

00:17:54.780 | and in a very sort of sandboxed environment.

00:17:58.660 | And we can quantify how does it scale?

00:18:00.660 | How does it perform?

00:18:02.060 | How many nodes do we need to allocate

00:18:03.260 | if we wanna scale this many inputs?

00:18:05.140 | When we start turning this stuff into prediction systems,

00:18:08.380 | real cybernetic systems,

00:18:10.140 | you're going to find scenarios where you get inputs

00:18:12.340 | that you're gonna wanna spend

00:18:13.180 | a little more time thinking about.

00:18:14.580 | You're gonna find inputs that are not,

00:18:15.860 | it's not clear what you should do, right?

00:18:17.460 | So then the software has a varying amount of runtime

00:18:20.180 | and correctness with regard to input.

00:18:22.260 | And that is a different kind of system altogether.

00:18:23.980 | Now it's a full-on cybernetic system.

00:18:25.980 | It's a next generation information system

00:18:27.620 | that is not like traditional software systems.

00:18:30.180 | - Can you maybe describe what is a cybernetic system?

00:18:33.140 | Do you include humans in that picture?

00:18:35.060 | So is it human in the loop kind of complex mess

00:18:38.900 | of the whole kind of interactivity of software

00:18:41.500 | with the real world, or is it something more concrete?

00:18:44.780 | - Well, when I say cybernetic,

00:18:45.740 | I really do mean that the software itself

00:18:47.700 | is closing the observe, orient, decide, act loop by itself.

00:18:51.620 | So humans being out of the loop is the fact

00:18:54.260 | what for me makes it a cybernetic system.

00:18:58.740 | - Humans are out of that loop.

00:19:00.900 | - When humans are out of the loop,

00:19:01.940 | when the machine is actually sort of deciding on its own

00:19:05.220 | what it should do next to get more information,

00:19:07.660 | that makes it a cybernetic system.

00:19:09.740 | So we're just at the dawn of this, right?

00:19:11.380 | I think everyone talking about MLAI, it's great,

00:19:15.380 | but really the thing we should be talking about

00:19:16.860 | is when we really enter the cybernetic era

00:19:20.380 | and all of the questions of ethics and governance

00:19:22.540 | and correctness and all these things,

00:19:24.620 | they really are the most important questions.

00:19:27.180 | - Okay, can we just linger on this?

00:19:28.580 | What does it mean for the human to be out of the loop

00:19:30.620 | in a cybernetic system?

00:19:31.780 | Because isn't the cybernetic system

00:19:34.100 | that's ultimately accomplishing some kind of purpose

00:19:37.420 | that at the bottom, the turtles all the way down,

00:19:41.820 | at the bottom, turtle's a human.

00:19:44.140 | - Well, the human may have set some criteria,

00:19:45.900 | but the human wasn't precise.

00:19:47.260 | So for instance, I just read the other day

00:19:49.260 | that earlier this year,

00:19:51.340 | or maybe it was last year at some point,

00:19:52.500 | the Libyan army, I think,

00:19:55.220 | sent out some automated killer drones with explosives.

00:19:58.220 | And there was no human in the loop at that point.

00:19:59.980 | They basically put them in a geo-fenced area,

00:20:02.220 | said find any moving target,

00:20:03.420 | like a truck or vehicle that looks like this, and boom.

00:20:06.980 | That's not a human in the loop, right?

00:20:09.300 | - So increasingly, the less human there is in the loop,

00:20:12.700 | the more concerned you are about these kinds of systems,

00:20:15.460 | because there's unintended consequences,

00:20:18.340 | like less the original designer and engineer of the system

00:20:21.940 | is able to predict, even one with good intent

00:20:24.980 | is able to predict the consequences of such a system.

00:20:27.340 | Is that- - That's right.

00:20:28.620 | There are some software systems, right,

00:20:30.140 | that run without humans in the loop that are quite complex.

00:20:32.780 | And that's like the electronic markets.

00:20:34.300 | And we get flash crashes all the time.

00:20:35.900 | We get, you know, in the heyday of high-frequency trading,

00:20:40.460 | there was a lot of market microstructure,

00:20:41.780 | people doing all sorts of weird stuff

00:20:43.620 | that the market designers had never really thought about,

00:20:47.220 | contemplated, or intended.

00:20:48.820 | So when we run these full-on systems

00:20:50.660 | with these automated trading bots,

00:20:52.860 | now they become automated, you know, killer drones,

00:20:55.420 | and then all sorts of other stuff,

00:20:57.260 | we are, that's what I mean by

00:20:58.980 | we're at the dawn of the cybernetic era

00:21:00.580 | and the end of the era of just pure software.

00:21:02.880 | - Are you more concerned,

00:21:05.980 | if you're thinking about cybernetic systems

00:21:08.220 | or even like self-replicating systems,

00:21:09.980 | so systems that aren't just doing a particular task,

00:21:13.540 | but are able to sort of multiply and scale

00:21:15.700 | in some dimension in the digital

00:21:17.980 | or even the physical world,

00:21:20.340 | are you more concerned about like the lobster being boiled,

00:21:24.780 | so a gradual, with us not noticing,

00:21:27.360 | collapse of civilization, or a big explosion,

00:21:33.260 | like oops, kind of a big thing where everyone notices,

00:21:38.660 | but it's too late?

00:21:40.100 | - I think that it will be a different experience

00:21:44.020 | for different people.

00:21:45.060 | I do share a common point of view

00:21:49.780 | with some of the climate, you know,

00:21:52.180 | people who are concerned about climate change

00:21:53.740 | and just the big existential risks that we have,

00:21:58.740 | but unlike a lot of people who share my level of concern,

00:22:02.540 | I think the collapse will not be quite so dramatic

00:22:06.180 | as some of them think.

00:22:07.540 | And what I mean is that I think that for certain tiers

00:22:10.780 | of let's say economic class or certain locations

00:22:13.580 | in the world,

00:22:14.620 | people will experience dramatic collapse scenarios.

00:22:17.740 | But for a lot of people, especially in the developed world,

00:22:20.300 | the realities of collapse will be managed.

00:22:24.020 | There'll be narrative management around it

00:22:26.600 | so that they essentially insulate,

00:22:29.120 | the middle class will be used to insulate the upper class

00:22:31.300 | from the pitchforks and the flaming torches and everything.

00:22:35.860 | - It's interesting 'cause,

00:22:36.940 | so my specific question wasn't as general,

00:22:39.540 | my question was more about cybernetic systems or software.

00:22:42.460 | - Okay.

00:22:43.580 | - It's interesting,

00:22:44.420 | but it would nevertheless perhaps be about class.

00:22:46.660 | So the effect of algorithms might affect certain classes

00:22:49.460 | more than others.

00:22:50.300 | - Absolutely.

00:22:51.120 | - I was more thinking about whether it's social media

00:22:53.700 | algorithms or actual robots,

00:22:56.140 | is there going to be a gradual effect on us

00:22:59.940 | where we wake up one day

00:23:02.580 | and don't recognize the humans we are?

00:23:05.140 | Or is it something truly dramatic where there's,

00:23:08.420 | like a meltdown of a nuclear reactor kind of thing,

00:23:12.140 | Chernobyl, like catastrophic events

00:23:15.220 | that are almost bugs in a program

00:23:19.100 | that scaled itself too quickly?

00:23:21.100 | - Yeah, I'm not as concerned about the visible stuff.

00:23:26.100 | And the reason is because the big visible explosions,

00:23:29.540 | I mean, this is the thing I said about social media

00:23:31.180 | is that, at least with nuclear weapons,

00:23:33.180 | when a nuke goes off, you can see it and you're like,

00:23:35.100 | "Well, that's really, wow, that's kind of bad."

00:23:37.780 | I mean, Oppenheimer was reciting the Bhagavad Gita,

00:23:40.620 | when he saw one of those things go off.

00:23:42.180 | So we can see nukes are really bad.

00:23:45.140 | - He's not reciting anything about Twitter.

00:23:47.980 | - Well, but right.

00:23:49.340 | But then when you have social media,

00:23:51.160 | when you have all these different things that conspire

00:23:54.380 | to create a layer of virtual experience for people

00:23:57.100 | that alienates them from reality and from each other,

00:24:00.940 | that's very pernicious.

00:24:02.380 | It's impossible to see, right?

00:24:03.820 | And it kind of slowly gets in there.

00:24:06.220 | So-

00:24:07.180 | - You've written about this idea of virtuality

00:24:09.580 | on this topic, which you define as the subjective phenomenon

00:24:14.100 | of knowingly engaging with virtual sensation and perception

00:24:17.820 | and suspending or forgetting the context

00:24:19.900 | that it's a simulacrum.

00:24:22.040 | So let me ask, what is real?

00:24:26.460 | Is there a hard line between reality and virtuality?

00:24:30.440 | Like perception drifts from some kind of physical reality.

00:24:33.460 | We have to kind of have a sense of what is the line

00:24:36.120 | that's we've gone too far.

00:24:37.660 | - Right, right.

00:24:38.760 | For me, it's not about any hard line about physical reality,

00:24:42.620 | as much as a simple question of,

00:24:46.180 | does the particular technology help people connect

00:24:51.560 | in a more integral way with other people,

00:24:54.380 | with their environment,

00:24:55.980 | with all of the full spectrum of things around them?

00:24:58.540 | So it's less about, oh, this is a virtual thing

00:25:00.860 | and this is a hard real thing,

00:25:02.980 | more about when we create virtual representations

00:25:05.740 | of the real things,

00:25:06.700 | always some things are lost in translation.

00:25:10.740 | Usually many, many dimensions are lost in translation.

00:25:14.340 | We're now coming to almost two years of COVID,

00:25:16.860 | people on Zoom all the time.

00:25:17.940 | You know it's different when you meet somebody in person

00:25:19.740 | than when you see them on,

00:25:20.580 | I've seen you on YouTube lots, right?

00:25:22.580 | But then seeing a person is very different.

00:25:24.020 | And so I think when we engage in virtual experiences

00:25:29.020 | all the time, and we only do that,

00:25:31.700 | there is absolutely a level of embodiment.

00:25:34.260 | There's a level of embodied experience

00:25:36.540 | and participatory interaction that is lost.

00:25:40.060 | And it's very hard to put your finger

00:25:41.620 | on exactly what it is.

00:25:42.500 | It's hard to say, oh, we're gonna spend $100 million

00:25:44.540 | building a new system that captures this 5% better

00:25:49.460 | higher fidelity human expression.

00:25:51.220 | No one's gonna pay for that, right?

00:25:52.580 | So when we rush madly into a world of simulacrum

00:25:57.420 | and virtuality, the things that are lost are,

00:26:02.420 | it's difficult, once everyone moves there,

00:26:05.420 | it can be hard to look back and see what we've lost.

00:26:08.100 | - So is it irrecoverably lost,

00:26:10.380 | or rather when you put it all on the table,

00:26:14.220 | is it possible for more to be gained than is lost?

00:26:17.180 | If you look at video games,

00:26:18.700 | they create virtual experiences that are surreal

00:26:22.700 | and can bring joy to a lot of people,

00:26:24.460 | can connect a lot of people,

00:26:26.700 | and can get people to talk a lot of trash.

00:26:29.900 | So it can bring out the best and the worst in people.

00:26:32.620 | So is it possible to have a future world

00:26:35.720 | where the pros outweigh the cons?

00:26:38.560 | - It is.

00:26:39.400 | I mean, it's possible to have that in the current world,

00:26:41.560 | but when literally trillions of dollars of capital

00:26:46.560 | are tied to using those things

00:26:48.920 | to groom the worst of our inclinations

00:26:52.840 | and to attack our weaknesses in the limbic system

00:26:56.080 | to create these things into id machines

00:26:57.640 | versus connection machines,

00:26:59.120 | then those good things don't stand a chance.

00:27:03.200 | - Can you make a lot of money

00:27:04.400 | by building connection machines?

00:27:06.160 | Is it possible, do you think,

00:27:08.360 | to bring out the best in human nature

00:27:10.920 | to create fulfilling connections and relationships

00:27:13.720 | in the digital world and make a shit ton of money?

00:27:17.020 | - If I figure it out, I'll let you know.

00:27:21.080 | - But what's your intuition

00:27:22.000 | without concretely knowing what's the solution?

00:27:24.680 | - My intuition is that a lot of our digital technologies

00:27:27.720 | give us the ability to have synthetic connections

00:27:30.800 | or to experience virtuality.

00:27:33.020 | They have co-evolved with sort of the human expectations.

00:27:38.020 | It's sort of like sugary drinks.

00:27:40.840 | As people have more sugary drinks,

00:27:42.320 | they need more sugary drinks to get that same hit, right?

00:27:45.860 | So with these virtual things and with TV and fast cuts

00:27:50.400 | and TikToks and all these different kinds of things,

00:27:52.840 | we're co-creating essentially humanity

00:27:55.040 | that sort of asks and needs those things.

00:27:57.320 | And now it becomes very difficult

00:27:58.880 | to get people to slow down.

00:28:00.340 | It gets difficult for people to hold their attention

00:28:03.480 | on slow things and actually feel

00:28:05.720 | that embodied experience, right?

00:28:07.300 | So mindfulness now more than ever is so important

00:28:10.340 | in schools and as a therapy technique for people

00:28:13.520 | because our environment has been accelerated.

00:28:15.640 | And McLuhan actually talks about this

00:28:17.340 | in the electric environment of the television.

00:28:19.480 | And that was before TikTok and before front-facing cameras.

00:28:22.400 | So I think for me, the concern is that

00:28:25.320 | it's not like we can ever switch to doing something better,

00:28:28.100 | but more of the humans and technology,

00:28:32.040 | they're not independent of each other.

00:28:33.520 | The technology that we use kind of molds what we need

00:28:37.720 | for the next generation of technology.

00:28:39.080 | - Yeah, but humans are intelligent

00:28:40.680 | and they're introspective and they can reflect

00:28:44.280 | on the experiences of their life.

00:28:45.800 | So for example, there's been many years in my life

00:28:47.860 | where I ate an excessive amount of sugar.

00:28:50.900 | And then a certain moment I woke up and said,

00:28:53.620 | "Why do I keep doing this?

00:28:55.880 | This doesn't feel good."

00:28:57.640 | Like long-term.

00:28:59.000 | And I think, so going through the TikTok process

00:29:02.300 | of realizing, okay, when I shorten my attention span,

00:29:06.280 | actually that does not make me feel good longer term.

00:29:10.240 | And realizing that, and then going to platforms,

00:29:13.160 | going to places that are away from the sugar.

00:29:18.160 | So in so doing, you can create platforms

00:29:21.080 | that can make a lot of money to help people wake up

00:29:24.000 | to what actually makes them feel good long-term.

00:29:26.280 | Develop, grow as human beings.

00:29:28.280 | And it just feels like humans are more intelligent

00:29:31.040 | than mice looking for cheese.

00:29:35.120 | They're able to sort of think, I mean,

00:29:36.960 | we can contemplate our own mortality.

00:29:39.440 | - Right. - We can contemplate

00:29:40.720 | things like long-term love and we can have a long-term fear

00:29:45.720 | of certain things like mortality.

00:29:48.080 | We can contemplate whether the experiences,

00:29:51.240 | the sort of the drugs of daily life

00:29:53.680 | that we've been partaking in

00:29:55.600 | is making us happier, better people.

00:29:58.320 | And then once we contemplate that,

00:30:00.200 | we can make financial decisions in using services

00:30:03.800 | and paying for services that are making us better people.

00:30:06.880 | So it just seems that we're in the very first stages

00:30:11.440 | of social networks that just were able

00:30:14.680 | to make a lot of money really quickly.

00:30:16.520 | But in bringing out sometimes the bad parts of human nature,

00:30:21.520 | they didn't destroy humans.

00:30:23.200 | They just fed everybody a lot of sugar.

00:30:26.040 | And now everyone's gonna wake up and say,

00:30:28.520 | "Hey, we're gonna start having like sugar-free social media."

00:30:31.760 | - Right, right.

00:30:33.280 | Well, there's a lot to unpack there.

00:30:34.800 | I think some people certainly have the capacity for that.

00:30:37.560 | And I certainly think, I mean, it's very interesting

00:30:39.680 | even the way you said it.

00:30:40.520 | You woke up one day and you thought,

00:30:42.040 | "Well, this doesn't feel very good."

00:30:44.200 | Well, that's still your limbic system saying,

00:30:45.680 | "This doesn't feel very good."

00:30:47.280 | Right, you have a cat brain's worth of neurons

00:30:49.160 | around your gut, right?

00:30:50.720 | And so maybe that saturated and that was telling you,

00:30:53.560 | "Hey, this isn't good."

00:30:55.040 | Humans are more than just mice looking for cheese

00:30:58.320 | or monkeys looking for sex and power, right?

00:31:00.800 | So- - Let's slow down.

00:31:02.640 | Now a lot of people would argue with you on that one.

00:31:05.960 | But yes. - Well, we're more

00:31:06.800 | than just that, but we're at least that.

00:31:08.480 | And we're very, very seldom not that.

00:31:11.800 | So I don't actually disagree with you

00:31:15.080 | that we could be better and that better platforms exist.

00:31:18.240 | And people are voluntarily noping out

00:31:20.000 | of things like Facebook and noping out.

00:31:21.720 | - That's an awesome verb. - It's a great term.

00:31:23.680 | Yeah, I love it.

00:31:24.520 | I use it all the time.

00:31:25.720 | You're welcome, Mike. - I'm gonna have to

00:31:26.560 | nope out of that. - I'm gonna nope out of that.

00:31:28.280 | Right, it's gonna be a hard pass.

00:31:29.760 | And that's great.

00:31:32.840 | But that's again, to your point,

00:31:34.240 | that's the first generation of front-facing cameras,

00:31:37.240 | of social pressures.

00:31:38.680 | And you as a self-starter, self-aware adult

00:31:43.560 | have the capacity to say, "Yeah, I'm not gonna do that.

00:31:46.320 | "I'm gonna go and spend time on long-form reads.

00:31:48.440 | "I'm gonna spend time managing my attention.

00:31:50.200 | "I'm gonna do some yoga."

00:31:52.080 | If you're a 15-year-old in high school

00:31:55.000 | and your entire social environment

00:31:57.040 | is everyone doing these things,

00:31:58.280 | guess what you're gonna do?

00:31:59.560 | You're gonna kind of have to do that

00:32:00.680 | because your limbic system says,

00:32:01.680 | "Hey, I need to get the guy or the girl or the whatever,

00:32:04.440 | "and that's what I'm gonna do."

00:32:05.520 | And so one of the things that we have to reason about here

00:32:07.800 | is the social media systems or social media, I think,

00:32:11.080 | is our first encounter with a technological system

00:32:16.440 | that runs a bit of a loop

00:32:18.280 | around our own cognition and attention.

00:32:21.840 | It's not the last.

00:32:23.600 | It's far from the last.

00:32:25.760 | And it gets to the heart

00:32:26.760 | of some of the philosophical Achilles heel

00:32:29.640 | of the Western philosophical system,

00:32:31.880 | which is each person gets to make their own determination.

00:32:34.320 | Each person is an individual that's sacrosanct

00:32:37.200 | in their agency and their sovereignty and all these things.

00:32:39.960 | The problem with these systems is they come down

00:32:42.560 | and they are able to manage their attention

00:32:44.360 | they come down and they are able to manage everyone en masse.

00:32:47.960 | And so every person is making their own decision,

00:32:50.160 | but together the bigger system is causing them to act

00:32:53.600 | with a group dynamic that's very profitable for people.

00:32:58.600 | So this is the issue that we have

00:33:00.960 | is that our philosophies are actually not geared

00:33:03.580 | to understand what is it for a person

00:33:06.480 | to have a high trust connection as part of a collective

00:33:11.400 | and for that collective to have its right

00:33:13.400 | to coherency and agency.

00:33:16.160 | That's something like when a social media app

00:33:19.280 | causes a family to break apart,

00:33:21.600 | it's done harm to more than just individuals.

00:33:24.520 | So that concept is not something we really talk about

00:33:27.200 | or think about very much,

00:33:28.040 | but that's actually the problem

00:33:30.040 | is that we're vaporizing molecules into atomic units

00:33:32.960 | and then we're hitting all the atoms with certain things

00:33:35.040 | that's like, yeah, well, that person chose to look at my app.

00:33:38.240 | - So our understanding of human nature

00:33:40.480 | is at the individual level,

00:33:42.360 | it emphasizes the individual too much

00:33:44.680 | because ultimately society operates at the collective level.

00:33:47.440 | - And these apps do as well.

00:33:48.600 | - And the apps do as well.

00:33:49.840 | So for us to understand the progression

00:33:52.680 | and the development of this organism

00:33:54.600 | we call human civilization,

00:33:55.940 | we have to think of the collective level too.

00:33:57.960 | - I would say multi-tiered.

00:33:59.240 | - Multi-tiered.

00:34:00.080 | - Multi-tiered.

00:34:00.920 | - So individual as well.

00:34:01.740 | - Individuals, family units, social collectives,

00:34:05.040 | and on the way up too.

00:34:07.120 | - So you've said that individual humans are multi-layered.

00:34:10.320 | Susceptible to signals and waves and multiple strata.

00:34:13.600 | The physical, the biological, social, cultural,

00:34:16.120 | intellectual, so sort of going along these lines,

00:34:19.440 | can you describe the layers of the cake

00:34:22.840 | that is a human being

00:34:25.320 | and maybe the human collective, human society?

00:34:29.180 | - So I'm just stealing wholesale here from Robert Persig,

00:34:32.880 | who is the author of "Zen and the Art of Motorcycle

00:34:34.560 | Maintenance" and in his follow-on book

00:34:39.040 | there's a sequel to it called "Lila".

00:34:40.680 | He goes into this in a little more detail,

00:34:42.280 | but it's a crude approach to thinking about people,

00:34:46.940 | but I think it's still an advancement

00:34:48.840 | over traditional subject-object metaphysics,

00:34:51.200 | where we look at people as a dualist would say,

00:34:53.920 | well, is your mind, your consciousness,

00:34:57.140 | is that just merely the matter that's in your brain

00:35:01.120 | or is there something kind of more beyond that?

00:35:03.440 | And they would say, yes, there's a soul,

00:35:05.120 | sort of ineffable soul

00:35:06.960 | beyond just merely the physical body, right?

00:35:09.360 | And I'm not one of those people, right?

00:35:11.360 | I think that we don't have to draw a line

00:35:13.440 | between are things only this or only that.

00:35:16.820 | Collectives of things can emerge structures and patterns

00:35:19.760 | that are just as real as the underlying pieces,

00:35:22.040 | but they're transcendent,

00:35:24.160 | but they're still of the underlying pieces.

00:35:26.520 | So your body is this way.

00:35:28.400 | I mean, we just know physically you consist of atoms

00:35:31.000 | and whatnot, and then the atoms are arranged into molecules,

00:35:34.680 | which then arrange into certain kinds of structures

00:35:37.120 | that seem to have a homeostasis to them,

00:35:39.560 | we call them cells,

00:35:40.600 | and those cells form sort of biological structures.

00:35:44.520 | Those biological structures give your body

00:35:46.800 | its physical ability and the biological ability

00:35:49.000 | to consume energy and to maintain homeostasis,

00:35:51.800 | but humans are social animals.

00:35:54.000 | I mean, human by themselves is not very long for the world.

00:35:57.180 | So we also, part of our biology

00:35:59.500 | is why are two connect to other people?

00:36:02.200 | From the mirror neurons to our language centers

00:36:04.920 | and all these other things.

00:36:06.200 | So we are intrinsically, there's a layer,

00:36:09.200 | there's a part of us that wants to be part of a thing.

00:36:12.560 | If we're around other people, not saying a word,

00:36:14.680 | but they're just up and down jumping and dancing, laughing,

00:36:16.760 | we're gonna feel better, right?

00:36:18.580 | And there was no exchange of physical anything.

00:36:21.760 | They didn't give us like five atoms of happiness, right?

00:36:24.880 | But there's an induction in our own sense of self

00:36:27.520 | that is at that social level.

00:36:29.680 | And then beyond that,

00:36:31.000 | Pirsig puts the intellectual level

00:36:33.680 | kind of one level higher than social.

00:36:35.080 | I think they're actually more intertwined than that.

00:36:37.080 | But the intellectual level is the level of pure ideas

00:36:41.080 | that you are a vessel for memes,

00:36:42.800 | you're a vessel for philosophies.

00:36:45.000 | You will conduct yourself in a particular way.

00:36:47.720 | I mean, I think part of this is if we think about it

00:36:49.480 | from a physics perspective,

00:36:50.880 | there's the joke that physicists like to approximate things.

00:36:55.080 | And we'll say, well, approximate a spherical cow, right?

00:36:57.480 | You're not a spherical cow, you're not a spherical human,

00:36:59.600 | you're a messy human.

00:37:00.720 | And we can't even say what the dynamics of your emotion

00:37:03.960 | will be unless we analyze all four of these layers, right?

00:37:08.520 | If you're Muslim at a certain time of day, guess what?

00:37:11.720 | You're gonna be on the ground, kneeling and praying, right?

00:37:13.960 | And that has nothing to do with your biological need

00:37:15.920 | to get on the ground or physics of gravity.

00:37:18.480 | It is an intellectual drive that you have.

00:37:20.400 | It's a cultural phenomenon

00:37:21.960 | and an intellectual belief that you carry.

00:37:23.720 | So that's what the four layered stack is all about.

00:37:28.080 | It's that a person is not only one of these things,

00:37:30.320 | they're all of these things at the same time.

00:37:31.720 | It's a superposition of dynamics that run through us,

00:37:35.640 | that make us who we are.

00:37:37.320 | - So no layer is special.

00:37:39.460 | - Not so much no layer is special,

00:37:41.680 | each layer is just different.

00:37:43.180 | But we are-

00:37:45.800 | - Each layer gets the participation trophy.

00:37:48.280 | - Yeah, each layer is a part of what you are.

00:37:50.280 | You are a layer cake, right, of all these things.

00:37:52.040 | And if we try to deny, right?

00:37:54.440 | So many philosophies do try to deny

00:37:56.560 | the reality of some of these things, right?

00:37:58.960 | Some people say, "Well, we're only atoms."

00:38:01.440 | Well, we're not only atoms

00:38:02.480 | because there's a lot of other things that are only atoms.

00:38:04.080 | I can reduce a human being to a bunch of soup

00:38:07.000 | and they're not the same thing,

00:38:08.520 | even though it's the same atoms.

00:38:09.800 | So I think the order and the patterns

00:38:12.160 | that emerge within humans to understand,

00:38:15.960 | to really think about what a next generation

00:38:17.960 | of philosophy would look like,

00:38:19.060 | that would allow us to reason about

00:38:20.640 | extending humans into the digital realm

00:38:22.960 | or to interact with autonomous intelligences

00:38:25.560 | that are not biological in nature.

00:38:27.480 | We really need to appreciate these,

00:38:29.600 | that human, what human beings actually are,

00:38:32.240 | is the superposition of these different layers.

00:38:34.720 | - You mentioned consciousness.

00:38:36.600 | Are each of these layers of cake conscious?

00:38:39.760 | Is consciousness a particular quality of one of the layers?

00:38:43.720 | Is there like a spike,

00:38:45.000 | if you have a consciousness detector at these layers?

00:38:47.940 | Or is something that just permeates all of these layers

00:38:50.320 | and just takes different form?

00:38:51.920 | - I believe what humans experience as consciousness

00:38:54.400 | is something that sits on a gradient scale

00:38:57.640 | of a general principle in the universe

00:39:00.540 | that seems to look for order and reach for order

00:39:04.960 | when there's an excess of energy.

00:39:06.780 | You know, it would be odd to say a proton is alive, right?

00:39:09.400 | It'd be odd to say that this particular atom

00:39:12.040 | or molecule of hydrogen gas is alive.

00:39:15.840 | But there's certainly something we can make

00:39:20.040 | assemblages of these things

00:39:22.400 | that have autopoetic aspects to them,

00:39:24.400 | that will create structures,

00:39:25.500 | that will, you know, crystalline solids

00:39:27.460 | will form very interesting and beautiful structures.

00:39:29.960 | This gets kind of into weird mathematical territories.

00:39:33.000 | You start thinking about Penrose and Game of Life stuff

00:39:35.320 | about the generativity of math itself,

00:39:37.920 | like the hyperreal numbers, things like that.

00:39:39.480 | But without going down that rabbit hole,

00:39:42.340 | I would say that there seems to be a tendency in the world

00:39:46.560 | that when there is excess energy,

00:39:49.120 | things will structure and pattern themselves.

00:39:51.360 | And they will then actually furthermore try to create

00:39:53.760 | an environment that furthers their continued stability.

00:39:58.080 | It's the concept of externalized extended phenotype

00:40:00.840 | or niche construction.

00:40:02.280 | So this is ultimately what leads to certain kinds

00:40:06.280 | of amino acids forming certain kinds of structures

00:40:09.200 | and so on and so forth until you get the ladder of life.

00:40:11.160 | So what we experience as consciousness,

00:40:12.920 | no, I don't think cells are conscious of that level.

00:40:15.600 | But is there something beyond mere equilibrium state biology

00:40:19.600 | and chemistry and biochemistry

00:40:21.960 | that drives what makes things work?

00:40:25.420 | I think there is.

00:40:26.420 | So Adrian Bajan has this construct a lot.

00:40:29.600 | There's other things you can look at.

00:40:31.200 | When you look at the life sciences

00:40:32.440 | and you look at any kind of statistical physics

00:40:36.040 | and statistical mechanics,

00:40:37.440 | when you look at things far out of equilibrium,

00:40:39.740 | when you have excess energy, what happens then?

00:40:43.400 | Life doesn't just make a hot herb soup.

00:40:45.800 | It starts making structure.

00:40:47.440 | There's something there.

00:40:48.640 | - The poetry of reaches for order

00:40:50.880 | when there's an excess of energy

00:40:52.780 | because you brought up game of life.

00:40:56.060 | You did it, not me.

00:40:59.160 | I love cellular automata,

00:41:00.360 | so I have to sort of linger on that for a little bit.

00:41:04.420 | So cellular automata, I guess, or game of life

00:41:09.340 | is a very simple example of reaching for order

00:41:11.800 | when there's an excess of energy

00:41:14.040 | or reaching for order and somehow creating complexity.

00:41:18.480 | It's explosion of just turmoil

00:41:22.400 | somehow trying to construct structures.

00:41:25.520 | And so doing creates

00:41:28.720 | very elaborate organism-looking type things.

00:41:31.520 | What intuition do you draw from the simplest mechanism?

00:41:35.800 | - Well, I like to turn that around its head

00:41:38.000 | and look at it as what if every single one of the patterns

00:41:42.360 | created life or created, not life,

00:41:45.760 | but created interesting patterns?

00:41:47.360 | Because some of them don't.

00:41:48.640 | And sometimes you make cool gliders.

00:41:50.560 | And other times you start with certain things

00:41:52.240 | and you make gliders and other things

00:41:54.000 | that then construct like AND gates and NOT gates, right?

00:41:57.000 | And you build computers on them.

00:41:59.240 | All of these rules that create these patterns

00:42:00.800 | that we can see, those are just the patterns we can see.

00:42:04.600 | What if our subjectivity is actually limiting our ability

00:42:06.960 | to perceive the order in all of it?

00:42:09.680 | What if some of the things that we think are random

00:42:12.280 | are actually not that random?

00:42:13.260 | We're simply not integrating at a final level

00:42:16.100 | across a broad enough time horizon.

00:42:17.920 | And this is again, I said, we go down the rabbit holes

00:42:20.560 | and the Penrose stuff or like Wolfram's explorations

00:42:22.840 | on these things.

00:42:23.680 | There is something deep and beautiful

00:42:27.260 | in the mathematics of all this.

00:42:28.560 | That is, hopefully one day I'll have enough money

00:42:30.360 | to work and retire and just ponder those questions.

00:42:33.400 | But there's something there.

00:42:34.520 | - But you're saying there's a ceiling to

00:42:36.080 | when you have enough money and you retire and you ponder,

00:42:38.440 | there's a ceiling to how much you can truly ponder

00:42:40.720 | because there's cognitive limitations

00:42:43.020 | in what you're able to perceive as a pattern.

00:42:46.340 | - Yeah.

00:42:47.180 | - And maybe mathematics extends your perception capabilities

00:42:51.500 | but it's still finite.

00:42:53.820 | It's just like.

00:42:55.460 | - Yeah, the mathematics we use is the mathematics

00:42:57.640 | that can fit in our head.

00:42:58.980 | - Yeah.

00:42:59.820 | - Did God really create the integers

00:43:02.580 | or did God create all of it?

00:43:03.740 | And we just happen at this point in time

00:43:05.340 | to be able to perceive integers.

00:43:07.220 | - Well, he just did the positive in it.

00:43:09.380 | - She, I just think did she create all of it?

00:43:11.340 | And then we.

00:43:13.080 | - She just created the natural numbers

00:43:15.740 | and then we screwed all up with zero and then I guess, okay.

00:43:18.540 | - But we did, we created mathematical operations

00:43:21.600 | so that we can have iterated steps

00:43:23.700 | to approach bigger problems, right?

00:43:26.000 | I mean, the entire point of the Arabic numeral system

00:43:29.040 | and it's a rubric for mapping a certain set of operations

00:43:32.580 | of folding them into a simple little expression.

00:43:35.360 | But that's just the operations that we can fit in our heads.

00:43:38.740 | There are many other operations besides, right?

00:43:41.160 | - The thing that worries me the most about aliens and humans

00:43:46.020 | is that they're aliens are all around us

00:43:49.100 | and we're too dumb to see them.

00:43:51.700 | - Oh, certainly, yeah.

00:43:52.860 | - Or life, let's say just life,

00:43:54.820 | life of all kinds of forms or organisms.

00:43:58.100 | You know what?

00:43:58.920 | Just even the intelligence of organisms

00:44:01.940 | is imperceptible to us

00:44:04.160 | because we're too dumb and self-centered.

00:44:06.620 | That worries me.

00:44:07.460 | - Well, we're looking for a particular kind of thing.

00:44:10.340 | When I was at Cornell,

00:44:11.200 | I had a lovely professor of Asian religions,

00:44:13.880 | Jane Marie Law,

00:44:14.720 | and she would tell this story about a musical,

00:44:17.800 | a musician, a Western musician who went to Japan

00:44:20.120 | and he taught classical music

00:44:21.900 | and could play all sorts of instruments.

00:44:24.040 | He went to Japan and he would ask people,

00:44:27.520 | he would basically be looking for things in the style of

00:44:30.640 | Western chromatic scale and these kinds of things.

00:44:34.100 | And then finding none of it, he would say,

00:44:35.600 | "Well, there's really no music in Japan,"

00:44:37.560 | but they're using a different scale.

00:44:38.840 | They're playing different kinds of instruments.

00:44:40.480 | The same thing she was using as sort of a metaphor

00:44:42.640 | for religion as well.

00:44:43.680 | In the West, we center a lot of religion,

00:44:45.800 | certainly the religions of Abraham,

00:44:48.000 | we center them around belief.

00:44:50.080 | And in the East, it's more about practice,

00:44:52.480 | spirituality and practice rather than belief.

00:44:54.640 | So anyway, the point is here to your point,

00:44:57.400 | life, I think so many people are so fixated

00:45:00.840 | on certain aspects of self-replication

00:45:03.360 | or homeostasis or whatever.

00:45:06.160 | But if we kind of broaden and generalize this thing

00:45:08.880 | of things reaching for order,

00:45:10.880 | under which conditions can they then create an environment

00:45:13.600 | that sustains that order, that allows them,

00:45:17.240 | the invention of death is an interesting thing.

00:45:20.160 | There are some organisms on earth

00:45:21.520 | that are thousands of years old.

00:45:23.420 | And it's not like they're incredibly complex,

00:45:25.560 | they're actually simpler than the cells that comprise us,

00:45:28.560 | but they never die.

00:45:29.640 | So at some point, death was invented,

00:45:33.320 | somewhere along the eukaryotic scale,

00:45:34.720 | I mean, even the protists, right?

00:45:35.980 | There's death.

00:45:37.160 | And why is that along with the sexual reproduction, right?

00:45:41.440 | There is something about the renewal process,

00:45:44.960 | something about the ability to respond

00:45:46.480 | to a changing environment,

00:45:48.120 | where it just become, just killing off the old generation

00:45:51.480 | and letting new generations try,

00:45:54.200 | seems to be the best way to fit into the niche.

00:45:56.680 | - You know, human historians seems to write about wheels

00:45:59.280 | and fires, the greatest inventions,

00:46:01.600 | but it seems like death and sex are pretty good.

00:46:04.320 | And they're kind of essential inventions

00:46:06.540 | at the very beginning.

00:46:07.380 | - At the very beginning, yeah.

00:46:08.560 | Well, we didn't invent them, right?

00:46:10.500 | - Well, broad, we, you didn't invent them.

00:46:13.260 | I see us as one, you particular homo sapien

00:46:16.980 | did not invent them, but we together,

00:46:19.500 | it's a team project, just like you're saying.

00:46:21.860 | - I think the greatest homo sapien invention

00:46:24.260 | is collaboration.

00:46:25.620 | - So when you say collaboration,

00:46:27.220 | Peter, where do ideas come from?

00:46:32.180 | And how do they take hold in society?

00:46:35.240 | Is that the nature of collaboration?

00:46:36.880 | Is that the basic atom of collaboration is ideas?

00:46:40.400 | - It's not not ideas, but it's not only ideas.

00:46:43.160 | There's a book I just started reading

00:46:44.420 | called "Death From a Distance."

00:46:45.880 | Have you heard of this? - No.

00:46:46.980 | - It's a really fascinating thesis,

00:46:49.160 | which is that humans are the only conspecific,

00:46:52.880 | the only species that can kill other members

00:46:55.800 | of the species from range.

00:46:58.240 | And maybe there's a few exceptions,

00:46:59.600 | but if you look in the animal world,

00:47:01.040 | you see like pronghorns, butting heads, right?

00:47:03.000 | You see the alpha lion and the beta lion,

00:47:05.840 | and they take each other down.

00:47:07.180 | Humans, we develop the ability to chuck rocks at each other,

00:47:10.120 | well, at prey, but also at each other.

00:47:11.960 | And that means the beta male can chunk a rock

00:47:14.920 | at the alpha male and take them down.

00:47:17.440 | And he can throw a lot of rocks, actually,

00:47:20.040 | miss a bunch of times, but just hit once and be good.

00:47:22.400 | So this ability to actually kill members of our own species

00:47:26.600 | from range without a threat of harm to ourselves

00:47:29.960 | created essentially mutually assured destruction

00:47:32.360 | where we had to evolve cooperation.

00:47:34.000 | If we didn't, then if we just continue to try to do,

00:47:37.280 | like I'm the biggest monkey in the tribe,

00:47:39.480 | and I'm gonna own this tribe and you have to go,

00:47:43.240 | if we do it that way, then those tribes basically failed.

00:47:46.720 | And the tribes that persisted

00:47:48.440 | and that have now given rise to the modern Homo sapiens

00:47:51.440 | are the ones where respecting the fact

00:47:53.860 | that we can kill each other from range without harm,

00:47:56.920 | like there's an asymmetric ability

00:47:58.640 | to snipe the leader from range,

00:48:00.840 | that meant that we sort of had to learn

00:48:03.920 | how to cooperate with each other.

00:48:05.200 | Right, come back here, don't throw that rock at me.

00:48:06.600 | Let's talk our differences out.

00:48:07.960 | - So violence is also part of collaboration.

00:48:10.240 | - The threat of violence, let's say.

00:48:12.300 | Well, the recognition, maybe the better way to put it

00:48:15.700 | is the recognition that we have more to gain

00:48:17.480 | by working together than the prisoner's dilemma

00:48:21.140 | of both of us defecting.

00:48:22.580 | - So mutually assured destruction in all its forms

00:48:26.360 | is part of this idea of collaboration.

00:48:28.720 | - Well, and Eric Weinstein talks about our nuclear peace.

00:48:31.360 | I mean, it kind of sucks

00:48:32.440 | that we have thousands of warheads aimed at each other,

00:48:34.040 | we mean Russia and the US, but it's like,

00:48:36.320 | on the other hand, we only fought proxy wars.

00:48:39.880 | We did not have another World War III

00:48:41.360 | of like hundreds of millions of people

00:48:43.160 | dying to like machine gun fire and giant guided missiles.

00:48:47.840 | - So the original nuclear weapon is a rock

00:48:50.400 | that we learned how to throw, essentially?

00:48:52.800 | - Yeah, well, the original scope of the world

00:48:54.280 | for any human being was their little tribe.

00:48:56.440 | I would say it still is for the most part.

00:49:00.760 | (both laughing)

00:49:02.000 | - Eric Weinstein speaks very highly of you,

00:49:05.800 | which is very surprising to me at first

00:49:08.000 | 'cause I didn't know there's this depth to you

00:49:10.840 | 'cause I knew you as an amazing leader of engineers

00:49:15.600 | and engineer yourself and so on.

00:49:17.400 | So it's fascinating.

00:49:18.480 | Maybe just as a comment, a side tangent that we can take,

00:49:23.840 | what's the nature of your friendship with Eric Weinstein?

00:49:27.160 | How did such two interesting paths cross?

00:49:30.640 | Is it your origins in physics?

00:49:32.960 | Is it your interest in philosophy

00:49:35.640 | and the ideas of how the world works?

00:49:37.280 | What is it?

00:49:38.120 | - It's actually, it's very random.

00:49:39.240 | Eric found me.

00:49:40.920 | He actually found Travis and I.

00:49:43.840 | - Travis Oliphant. - Oliphant, yeah.

00:49:45.120 | We were both working at a company called Enthought

00:49:47.400 | back in the mid 2000s,

00:49:48.520 | and we were doing a lot of consulting

00:49:50.840 | around scientific Python.

00:49:52.840 | And we'd made some tools

00:49:54.400 | and Eric was trying to use some of these Python tools

00:49:57.400 | to visualize that he had a fiber bundle approach

00:50:00.200 | to modeling certain aspects of economics.

00:50:03.440 | He was doing this

00:50:04.280 | and that's how he kind of got in touch with us.

00:50:05.600 | And so-

00:50:07.200 | - This was in the early-

00:50:08.120 | - This was in the mid 2000s.

00:50:10.080 | '07 timeframe, '06, '07 timeframe.

00:50:13.680 | - Eric Weinstein trying to use Python

00:50:16.200 | to visualize fiber bundles. - Right, to visualize

00:50:17.200 | fiber bundles using some of the tools

00:50:19.720 | that we'd built in the open source.

00:50:21.280 | That's somehow entertaining to me.

00:50:23.000 | - It's pretty funny. - The thought of that.

00:50:24.120 | - It's pretty funny.

00:50:25.040 | But then, we met with him a couple of times,

00:50:27.120 | a really interesting guy.

00:50:27.960 | And then in the wake of the '07, '08

00:50:29.840 | kind of financial collapse,

00:50:31.320 | he helped organize with Lee Smolin

00:50:34.720 | a symposium at the Perimeter Institute

00:50:36.960 | about, okay, well, clearly,

00:50:39.880 | big finance can't be trusted,

00:50:41.160 | government's in its pockets with regulatory capture.

00:50:43.480 | What the F do we do?

00:50:45.280 | And all sorts of people, Nassim Tlaib was there

00:50:47.600 | and Andy Lo from MIT was there

00:50:49.440 | and Bill Janeway, I mean, just a lot of

00:50:52.880 | top billing people were there.

00:50:54.800 | And he invited me and Travis

00:50:56.880 | and another one of our coworkers, Robert Kern,

00:51:00.280 | who is anyone in the SciPy, NumPy community knows Robert,

00:51:03.880 | really great guy.

00:51:04.720 | So the three of us also got invited to go to this thing.

00:51:06.600 | And that's where I met Brett Weinstein

00:51:07.760 | for the first time as well.

00:51:09.480 | Yeah, I knew him before he got all famous

00:51:11.680 | for unfortunate reasons, I guess.

00:51:13.520 | But anyway, so we met then and kind of had a friendship.

00:51:19.600 | Throughout, since then.

00:51:21.400 | - You have a depth of thinking

00:51:24.000 | that kind of runs with Eric

00:51:26.800 | in terms of just thinking about the world deeply

00:51:28.640 | and thinking philosophically.

00:51:30.440 | And then there's Eric's interest in programming.

00:51:33.480 | I actually have never,

00:51:34.640 | he'll bring up programming to me quite a bit

00:51:38.960 | as a metaphor for stuff.

00:51:41.080 | But I never kind of pushed the point of like,

00:51:43.640 | what's the nature of your interest in programming?

00:51:46.800 | I think he saw it probably as a tool.

00:51:48.840 | - Yeah, absolutely.

00:51:49.680 | - That you visualize to explore mathematics

00:51:52.200 | and explore physics.

00:51:53.120 | But I was wondering like,

00:51:55.040 | what's his depth of interest and also his vision

00:51:59.360 | for what programming would look like in the future?

00:52:05.640 | Have you had interaction with him,

00:52:07.400 | like discussion in the space of Python programming?

00:52:09.840 | - Well, in the sense of sometimes he asks me,

00:52:11.920 | why is this stuff still so hard?

00:52:13.840 | (both laughing)

00:52:16.960 | Yeah, you know, everybody's a critic,

00:52:18.280 | but actually no, Eric--

00:52:20.320 | - Programming you mean like in general?

00:52:21.520 | - Yes, yes.

00:52:22.360 | Well, not programming in general,

00:52:23.320 | but certain things in the Python ecosystem.

00:52:25.440 | But he actually, I think what I find

00:52:29.040 | in listening to some of his stuff

00:52:30.120 | is that he does use programming metaphors a lot, right?

00:52:33.280 | He'll talk about APIs or object oriented

00:52:35.400 | and things like that.

00:52:36.400 | So I think that's a useful set of frames

00:52:39.240 | for him to draw upon for discourse.

00:52:42.880 | I haven't pair programmed with him in a very long time.

00:52:45.440 | - You've previously--

00:52:47.240 | - Well, I mean, I look at his code,

00:52:48.760 | trying to help like put together

00:52:50.200 | some of the visualizations around these things,

00:52:51.560 | but it's been a very, not really pair program,

00:52:54.040 | but like even looked at his code, right?

00:52:55.800 | I mean--

00:52:56.640 | - How legendary would be is that like,

00:52:58.960 | get repo with Peter Wang and Eric Weinstein.

00:53:02.680 | - Well, honestly, Robert Kern did all the heavy lifting.

00:53:05.480 | So I have to give credit where credit is due.

00:53:06.880 | Robert is the silent, but incredibly deep,

00:53:10.320 | quiet, not silent, but quiet,

00:53:11.640 | but incredibly deep individual

00:53:13.480 | at the heart of a lot of those things

00:53:14.720 | that Eric was trying to do.

00:53:16.680 | But we did have, you know, in the,

00:53:18.560 | as Travis and I were starting our company in 2012 timeframe,

00:53:23.520 | we went to New York.

00:53:24.640 | Eric was still in New York at the time.

00:53:26.120 | He hadn't moved to,

00:53:27.560 | this was before he joined Teal Capital.

00:53:29.160 | We just had like a steak dinner somewhere.

00:53:31.560 | Maybe it was Keen's, I don't know, somewhere in New York.

00:53:33.400 | So it was me, Travis, Eric,

00:53:35.080 | and then Wes McKinney, the creative pandas.

00:53:37.280 | And then Wes' then business partner, Adam,

00:53:40.280 | the five of us sat around having this,

00:53:42.000 | just a hilarious time, amazing dinner.

00:53:45.520 | I forget what all we talked about,

00:53:46.960 | but it was one of those conversations,

00:53:49.000 | which I wish as soon as COVID is over,

00:53:51.400 | maybe Eric and I can sit down.

00:53:53.080 | - Recreate.

00:53:53.920 | - Recreate it somewhere in LA, or maybe he comes here.

00:53:56.720 | 'Cause a lot of cool people are here in Austin, right?

00:53:58.200 | - Exactly.

00:53:59.040 | - Yeah, we're all here.

00:53:59.880 | He should come here. - Come here.

00:54:00.720 | - Yeah.

00:54:01.560 | - So he uses the metaphor of source code

00:54:03.600 | sometimes to talk about physics.

00:54:05.320 | We figure out our own source code.

00:54:07.160 | So you with the physics background,

00:54:10.160 | and somebody who's quite a bit of an expert in source code,

00:54:14.160 | do you think we'll ever figure out our own source code

00:54:17.760 | in the way that Eric means?

00:54:19.040 | Do you think we'll figure out the nature of it?

00:54:19.880 | - Well, I think we're constantly working on that problem.

00:54:21.760 | I mean, I think we'll make more and more progress.

00:54:24.440 | For me, there's some things I don't really doubt too much.

00:54:28.120 | Like I don't really doubt that one day we will create

00:54:31.600 | a synthetic, maybe not fully in silicon,

00:54:34.520 | but a synthetic approach to cognition

00:54:37.760 | that rivals the biological 20 watt computers in our heads.

00:54:44.760 | - What's cognition here?

00:54:46.160 | - Cognition.

00:54:47.000 | - Which aspect?

00:54:47.820 | - Perception, attention, memory, recall,

00:54:50.000 | asking better questions.

00:54:51.880 | That for me is a measure of intelligence.

00:54:53.240 | - Doesn't Roomba Vacuum Cleaner already do that?

00:54:55.440 | Or do you mean, oh, it doesn't ask questions.

00:54:57.120 | - I mean, no.

00:54:58.040 | It's, I mean, I have a Roomba, but it's--

00:55:00.920 | - Well, it asks questions.

00:55:01.760 | - It's not even as smart as my cat, right?

00:55:02.840 | So.

00:55:03.680 | - It doesn't ask questions about what is this wall?

00:55:05.360 | It now, new feature asks, is this poop or not, apparently.

00:55:08.960 | - Yes, a lot of our current cybernetic system,

00:55:11.360 | it's a cybernetic system.

00:55:12.680 | It will go and it will happily vacuum up some poop, right?

00:55:15.020 | The older generations would.

00:55:16.720 | - The new one, just released, does not vacuum up the poop.

00:55:19.960 | This is a commercial for Roomba.

00:55:20.800 | - I wonder if it still gets stuck

00:55:21.720 | under my first rung of my stair.

00:55:23.760 | In any case, these cybernetic systems we have,

00:55:27.120 | they are mold, they're designed to be sent off

00:55:32.040 | into a relatively static environment.

00:55:34.040 | And whatever dynamic things happen in the environment,

00:55:36.360 | they have a very limited capacity to respond to.

00:55:38.840 | A human baby, a human toddler of 18 months of age,

00:55:43.040 | has more capacity to manage its own attention

00:55:45.800 | and its own capacity to make better sense of the world

00:55:49.140 | than the most advanced robots today.

00:55:51.680 | So again, my cat, I think, can do a better job of my two,

00:55:55.080 | and they're both pretty clever.

00:55:56.360 | So I do think though, back to my kind of original point,

00:55:59.440 | I think that it's not, for me, it's not a question at all

00:56:02.680 | that we will be able to create synthetic systems

00:56:05.960 | that are able to do this better than the human,

00:56:09.200 | at an equal level or better than the human mind.

00:56:11.680 | It's also, for me, not a question

00:56:14.840 | that we will be able to put them alongside humans

00:56:19.840 | so that they capture the full broad spectrum

00:56:23.200 | of what we are seeing as well,

00:56:25.360 | and also looking at our responses, listening to our responses,

00:56:28.880 | even maybe measuring certain vital signs about us.

00:56:32.040 | So in this kind of sidecar mode,

00:56:34.440 | a greater intelligence could use us

00:56:37.600 | in our whatever, 80 years of life, to train itself up,

00:56:42.200 | and then be a very good simulacrum of us moving forward.

00:56:45.040 | - So who is in the sidecar in that picture of the future,

00:56:49.920 | exactly, is it the human?

00:56:51.200 | - The baby version of our immortal selves.

00:56:53.000 | - Okay, so once the baby grows up,

00:56:56.200 | is there any use for humans?

00:56:58.480 | - I think so.

00:56:59.960 | I think that out of epistemic humility,

00:57:03.240 | we need to keep humans around for a long time.

00:57:05.600 | And I would hope that anyone making those systems

00:57:07.960 | would believe that to be true.

00:57:10.040 | - Out of epistemic humility.

00:57:11.640 | What's the nature of the humility that-

00:57:13.480 | - That we don't know what we don't know.

00:57:15.520 | - So we don't-

00:57:17.280 | - Right?

00:57:19.800 | - So we don't know-

00:57:20.640 | - First, I mean, first we have to build systems

00:57:21.680 | that help us do the things that we do know about,

00:57:24.400 | that can then probe the unknowns that we know about,

00:57:26.740 | but the unknown unknowns, we don't know.

00:57:28.840 | Nature is the one thing that is infinitely

00:57:32.400 | able to surprise us.

00:57:33.780 | So we should keep biological humans around

00:57:35.880 | for a very, very, very long time,

00:57:37.560 | even after our immortal selves have transcended

00:57:40.440 | and have gone off to explore other worlds,

00:57:42.880 | gone to go communicate with the lifeforms

00:57:44.520 | living in the sun or whatever else.

00:57:45.880 | So I think that's, for me,

00:57:49.520 | these seem like things that are going to happen.

00:57:53.000 | Like I don't really question that,

00:57:54.480 | that they're gonna happen.

00:57:55.720 | Assuming we don't completely destroy ourselves.

00:57:58.220 | - Is it possible to create an AI system

00:58:02.480 | that you fall in love with and it falls in love with you

00:58:06.140 | and you have a romantic relationship with it

00:58:08.460 | or a deep friendship, let's say?

00:58:10.740 | - I would hope that that is the design criteria

00:58:12.660 | for any of these systems.

00:58:14.500 | If we cannot have a meaningful relationship with it,

00:58:18.460 | then it's still just a chunk of silicon.

00:58:20.300 | - So then what is meaningful?

00:58:21.640 | Because back to sugar.

00:58:23.780 | - Well, sugar doesn't love you back, right?

00:58:25.360 | So the computer has to love you back.

00:58:26.820 | And what does love mean?

00:58:28.180 | Well, in this context, for me, love,

00:58:30.140 | I'm gonna take a page from Alain de Botton,

00:58:31.980 | love means that it wants to help us

00:58:34.360 | become the best version of ourselves.

00:58:36.620 | - Yes, that's beautiful.

00:58:39.420 | That's a beautiful definition of love.

00:58:40.700 | So what role does love play in the human condition

00:58:45.340 | at the individual level and at the group level?

00:58:48.680 | 'Cause you were kind of saying that humans,

00:58:51.060 | we should really consider humans

00:58:52.260 | both at the individual and the group and the societal level.

00:58:55.260 | What's the role of love in this whole thing?

00:58:56.920 | We talked about sex, we talked about death,

00:58:59.620 | thanks to the bacteria that invented it.

00:59:02.500 | At which point did we invent love, by the way?

00:59:04.300 | I mean, is that also-

00:59:05.500 | - No, I think love is the start of it all.

00:59:08.900 | And the feelings of, and this gets,

00:59:11.100 | this is sort of beyond just romantic, sensual,

00:59:15.960 | whatever kind of things,

00:59:16.800 | but actually genuine love as we have for another person,

00:59:19.680 | love as it would be used in a religious text, right?

00:59:22.660 | I think that capacity to feel love,

00:59:25.440 | more than consciousness, that is the universal thing.

00:59:28.380 | Our feeling of love is actually a sense of that generativity.

00:59:31.260 | When we can look at another person

00:59:32.900 | and see that they can be something more than they are,

00:59:37.380 | and more than just what we,

00:59:38.900 | a pigeonhole we might stick them in.

00:59:42.020 | We see, I mean, I think there's,

00:59:43.180 | in any religious text you'll find

00:59:44.900 | voiced some concept of this,

00:59:47.620 | that you should see the grace of God in the other person.

00:59:50.180 | That they're made in the spirit of what,

00:59:54.300 | the love that God feels for his creation or her creation.

00:59:57.060 | And so I think this thing is actually the root of it.

01:00:00.140 | So I would say before, I don't think molecules of water

01:00:04.760 | feel consciousness, have consciousness,

01:00:06.220 | but there is some proto micro quantum thing of love.

01:00:10.900 | That's the generativity when there's more energy

01:00:14.060 | than what they need to maintain equilibrium.

01:00:16.220 | And that, when you sum it all up,

01:00:18.300 | is something that leads to,

01:00:19.780 | I mean, I had my mind blown one day as an undergrad

01:00:23.460 | at the physics computer lab.

01:00:24.540 | I logged in and when you log into Bash for a long time,

01:00:28.260 | there was a little fortune that would come out.

01:00:29.540 | And it said, "Man was created by water

01:00:32.300 | "to carry itself uphill."

01:00:34.000 | And I was logging in to work on some problem set.

01:00:37.980 | And I logged in and I saw that and I just said,

01:00:40.500 | "Son of a bitch."

01:00:42.020 | I logged out and I went to the coffee shop

01:00:44.380 | and I got a coffee and I sat there on the quad.

01:00:46.180 | I'm like, "You know, it's not wrong and yet WTF, right?"

01:00:51.180 | So when you look at it that way, it's like,

01:00:55.820 | yeah, okay, non-equilibrium physics is a thing.

01:00:59.060 | And so when we think about love,

01:01:00.980 | when we think about these kinds of things,

01:01:03.080 | I would say that in the modern day human condition,

01:01:08.100 | there's a lot of talk about freedom and individual liberty

01:01:12.260 | and rights and all these things.

01:01:14.580 | But that's very Hegelian,

01:01:17.220 | it's very kind of following from the Western philosophy

01:01:19.900 | of the individual as sacrosanct.

01:01:22.700 | But it's not really couched, I think, the right way

01:01:26.300 | because it should be how do we maximize people's ability

01:01:29.420 | to love each other, to love themselves first,

01:01:32.060 | to love each other, their responsibilities

01:01:34.500 | to the previous generation, to the future generations.

01:01:37.780 | Those are the kinds of things

01:01:39.240 | that should be our design criteria, right?

01:01:41.780 | Those should be what we start with to then come up

01:01:45.660 | with the philosophies of self

01:01:47.740 | and of rights and responsibilities.

01:01:50.140 | But that love being at the center of it,

01:01:52.200 | I think when we design systems for cognition,

01:01:55.680 | it should absolutely be built that way.

01:01:58.660 | I think if we simply focus on efficiency and productivity,

01:02:02.200 | these kind of very industrial era,

01:02:04.840 | all the things that Marx had issues with, right?

01:02:08.300 | Those, that's a way to go and really, I think,

01:02:11.500 | go off the deep end in the wrong way.

01:02:13.540 | - So one of the interesting consequences

01:02:16.400 | of thinking of life in this hierarchical way

01:02:21.060 | of an individual human, and then there's groups

01:02:23.380 | and there's societies is I believe that,

01:02:28.260 | you believe that corporations are people.

01:02:30.640 | So this is kind of a politically dense idea,

01:02:36.540 | and all those kinds of things.

01:02:37.380 | If we just throw politics aside,

01:02:39.060 | if we throw all of that aside,

01:02:41.260 | in which sense do you believe that corporations are people?

01:02:44.600 | And how does love connect to that?

01:02:47.620 | - Right, so the belief is that groups of people

01:02:51.980 | have some kind of higher level,

01:02:54.560 | I would say mesoscopic claim to agency.

01:02:56.640 | So where do I, let's start with this.

01:03:00.880 | Most people would say, okay,

01:03:01.980 | individuals have claims to agency and sovereignty.

01:03:05.100 | Nations, we certainly act as if nations,

01:03:07.580 | so at a very large, large scale,

01:03:09.500 | nations have rights to sovereignty and agency.

01:03:13.080 | Like everyone plays the game of modernity

01:03:14.980 | as if that's true, right?

01:03:16.300 | We believe France is a thing.

01:03:17.340 | We believe the United States is a thing.

01:03:18.780 | But to say that groups of people

01:03:21.900 | at a smaller level than that,

01:03:23.760 | like a family unit is the thing.

01:03:26.660 | Well, in our laws, we actually do encode this concept.

01:03:30.460 | I believe that in a relationship, in a marriage, right,

01:03:33.820 | one partner can sue for loss of consortium, right,

01:03:37.740 | if someone breaks up the marriage or whatever.

01:03:39.820 | So these are concepts that even in law,

01:03:41.580 | we do respect that there is something about the union

01:03:44.740 | and about the family.

01:03:45.980 | So for me, I don't think it's so weird

01:03:48.300 | to think that groups of people have a right to,

01:03:51.820 | a claim to rights and sovereignty of some degree.

01:03:54.660 | I mean, we look at our clubs, we look at churches.

01:03:59.020 | These are, we talk about these collectives of people

01:04:02.020 | as if they have a real agency to them.

01:04:04.580 | And they do.

01:04:05.780 | But I think if we take that one step further,

01:04:08.460 | and say, okay, they can accrue resources.

01:04:10.300 | Well, yes, check, you know, and by law, they can.

01:04:12.740 | They can own land, they can engage in contracts,

01:04:17.060 | they can do all these different kinds of things.

01:04:18.860 | So we, in legal terms, support this idea

01:04:22.580 | that groups of people have rights.

01:04:24.660 | Where we go wrong on this stuff

01:04:28.040 | is that the most popular version of this

01:04:31.260 | is the for-profit absentee owner corporation

01:04:35.380 | that then is able to amass larger resources

01:04:38.440 | than anyone else in the landscape, anything else,

01:04:40.860 | any other entity of equivalent size.

01:04:43.100 | And they're able to essentially bully around individuals,

01:04:45.500 | whether it's laborers, whether it's people

01:04:47.060 | whose resources they want to capture.

01:04:48.940 | They're also able to bully around

01:04:50.460 | our system of representation,

01:04:52.140 | which is still tied to individuals, right?

01:04:55.540 | So I don't believe that's correct.

01:04:58.520 | I don't think it's good that they, you know,

01:05:01.160 | they're people, but they're assholes.

01:05:02.320 | I don't think that corporations as people

01:05:03.660 | acting like assholes is a good thing.

01:05:05.500 | But the idea that collectives and collections of people,

01:05:08.420 | that we should treat them philosophically

01:05:10.140 | as having some--

01:05:12.260 | - Agency.

01:05:13.100 | - Some agency and some mass at a mesoscopic level.

01:05:16.900 | I think that's an important thing

01:05:17.980 | because one thing I do think we underappreciate sometimes

01:05:22.340 | is the fact that relationships have relationships.

01:05:26.180 | So it's not just individuals having relationships

01:05:27.900 | with each other.

01:05:29.060 | But if you have eight people seated around a table, right?

01:05:32.060 | Each person has a relationship with each of the others,

01:05:34.180 | and that's obvious.

01:05:35.600 | But then if it's four couples,

01:05:37.840 | each couple also has a relationship

01:05:39.680 | with each of the other couples, right?

01:05:41.680 | The dyads do.

01:05:42.720 | And if it's couples, but one is the father and mother older,

01:05:47.040 | and then, you know, one of their children and their spouse,

01:05:50.920 | that family unit of four has a relationship

01:05:53.940 | with the other family unit of four.

01:05:55.720 | So the idea that relationships have relationships

01:05:57.600 | is something that we intuitively know

01:05:59.920 | in navigating the social landscape,

01:06:01.780 | but it's not something I hear expressed like that.

01:06:04.720 | It's certainly not something that is,

01:06:06.680 | I think, taken into account very well

01:06:08.160 | when we design these kinds of things.

01:06:09.480 | So I think the reason why I care a lot about this

01:06:14.040 | is because I think the future of humanity

01:06:16.520 | requires us to form better sense,

01:06:19.400 | make collective sense-making units

01:06:21.520 | at something, you know, around Dunbar number,

01:06:25.200 | you know, half to 5X Dunbar.

01:06:28.120 | And that's very different than right now

01:06:30.600 | where we defer sense-making

01:06:33.560 | to massive aging zombie institutions.

01:06:37.000 | Or we just do it ourselves.

01:06:38.480 | We go it alone,

01:06:39.320 | go to the dark force of the internet by ourselves.

01:06:41.080 | - So that's really interesting.

01:06:42.320 | So you've talked about agency,

01:06:45.300 | I think maybe calling it a convenient fiction

01:06:47.720 | at all these different levels.

01:06:49.580 | So even at the human individual level,

01:06:52.040 | it's kind of a fiction, we all believe,

01:06:53.920 | because we are, like you said, made of cells,

01:06:55.700 | and cells are made of atoms.

01:06:57.680 | So that's a useful fiction.

01:06:58.960 | And then there's nations.

01:07:00.840 | That seems to be a useful fiction.

01:07:02.860 | But it seems like some fictions are better than others.

01:07:06.880 | You know, there's a lot of people

01:07:07.840 | that argue the fiction of nation is a bad idea.

01:07:10.960 | One of them lives two doors down from me,

01:07:13.840 | Michael Malice, he's an anarchist.

01:07:16.120 | You know, I'm sure there's a lot of people

01:07:18.260 | who are into meditation that believe the idea,

01:07:21.840 | this useful fiction of agency of an individual

01:07:24.880 | is troublesome as well.

01:07:26.680 | We need to let go of that in order to truly,

01:07:29.440 | like to transcend, I don't know,

01:07:32.280 | I don't know what words you wanna use,

01:07:33.720 | but suffering or to elevate the experience of life.

01:07:38.420 | So you're kind of arguing that,

01:07:40.760 | okay, so we have some of these useful fictions of agency.

01:07:44.640 | We should add a stronger fiction

01:07:47.880 | that we tell ourselves about the agency of groups

01:07:51.480 | in the hundreds of the half of Dunbar's number,

01:07:56.160 | five X Dunbar's number.

01:07:57.880 | - Yeah, something on that order.

01:07:58.720 | And we call them fictions,

01:07:59.760 | but really they're rules of the game, right?

01:08:01.540 | Rules that we feel are fair or rules that we consent to.

01:08:05.680 | - I always question the rules when I lose, like a monopoly.

01:08:08.560 | That's when I usually question,

01:08:09.800 | when I'm winning, I don't question the rules.

01:08:11.520 | - We should play a game of Monopoly someday.

01:08:12.840 | There's a trippy version of it that we could do.

01:08:15.180 | - What kind?

01:08:16.240 | - Contract Monopoly is introduced by a friend of mine to me,

01:08:19.160 | where you can write contracts on future earnings

01:08:23.080 | or landing on various things.

01:08:24.560 | And you can hand out, like,

01:08:26.720 | you can land first three times,

01:08:27.780 | you land in a park place is free or whatever.

01:08:29.520 | Just, and then you can start trading

01:08:31.040 | those contracts for money.

01:08:32.340 | - And then you create a human civilization

01:08:35.480 | and somehow Bitcoin comes into it.

01:08:38.280 | Okay, but some of these-

01:08:40.520 | - Actually, I bet if me and you and Eric

01:08:42.680 | sat down to play a game of Monopoly

01:08:44.160 | and we were to make NFTs out of the contracts we wrote,

01:08:46.280 | we could make a lot of money.

01:08:48.000 | Now it's a terrible idea.

01:08:49.080 | - Yes. - I would never do it,

01:08:50.160 | but I bet we could actually sell the NFTs around.

01:08:52.920 | - I have other ideas to make money that I could tell you,

01:08:56.840 | and they're all terrible ideas,

01:08:58.400 | including cat videos on the internet.

01:09:02.280 | Okay, but some of these rules of the game,

01:09:04.580 | some of these fictions are,

01:09:06.020 | it seems like they're better than others.

01:09:09.280 | - They have worked this far to cohere human,

01:09:13.320 | to organize human collective action.

01:09:14.840 | - But you're saying something about,

01:09:16.600 | especially this technological age

01:09:19.240 | requires modified fictions, stories of agency.

01:09:23.640 | Why the Dunbar number?

01:09:25.040 | And also, how do you select the group of people?

01:09:28.440 | Dunbar numbers, I think,

01:09:30.040 | I have the sense that it's overused

01:09:33.920 | as a kind of law that somehow

01:09:38.000 | we can have deep human connection at this scale.

01:09:41.280 | Like some of it feels like an interface problem too.

01:09:45.480 | It feels like if I have the right tools,

01:09:48.080 | I can deeply connect with a larger number of people.

01:09:51.840 | It just feels like there's a huge value

01:09:55.480 | to interacting just in person,

01:09:57.980 | getting to share traumatic experiences together,

01:10:01.000 | beautiful experiences together.

01:10:02.720 | There's other experiences

01:10:04.080 | like that in the digital space that you can share.

01:10:07.560 | It just feels like Dunbar's number

01:10:09.360 | can be expanded significantly,

01:10:10.800 | perhaps not to the level of millions and billions,

01:10:15.000 | but it feels like it could be expanded.

01:10:16.280 | So how do we find the right interface, you think,

01:10:20.240 | for having a little bit of a collective here

01:10:24.840 | that has agency?

01:10:26.080 | - You're right that there's many different ways

01:10:28.080 | that we can build trust with each other.

01:10:30.840 | My friend, Joe Edelman talks about a few different ways

01:10:33.980 | that mutual appreciation, trustful conflict,

01:10:38.980 | just experiencing something.

01:10:41.320 | There's a variety of different things that we can do,

01:10:43.640 | but all those things take time and you have to be present.

01:10:48.480 | The less present you are, I mean, there's just, again,

01:10:50.320 | a no free lunch principle here.

01:10:51.560 | The less present you are, the more of them you can do,

01:10:54.200 | but then the less connection you build.

01:10:56.800 | So I think there is sort of a human capacity issue

01:10:59.440 | around some of these things.

01:11:00.260 | Now, that being said, if we can use certain technologies,

01:11:04.800 | so for instance, if I write a little monograph

01:11:07.540 | on my view of the world,

01:11:08.760 | you read it asynchronously at some point,

01:11:10.600 | and you're like, "Wow, Peter, this is great.

01:11:11.880 | Here's mine."

01:11:12.800 | I read it, I'm like, "Wow, Lex, this is awesome."

01:11:15.320 | We can be friends without having to spend 10 years,

01:11:19.440 | figuring all this stuff out together.

01:11:20.560 | We can just read each other's thing and be like,

01:11:22.120 | "Oh yeah, this guy's exactly in my wheelhouse

01:11:24.800 | and vice versa."

01:11:26.080 | And we can then connect just a few times a year

01:11:30.520 | and maintain a high trust relationship.

01:11:33.040 | It can be expanded a little bit, but it also requires,

01:11:35.840 | these things are not all technological in nature,

01:11:37.320 | it requires the individual themselves

01:11:39.640 | to have a certain level of capacity,

01:11:41.680 | to have a certain lack of neuroticism.

01:11:44.760 | If you wanna use the ocean big five sort of model,

01:11:48.080 | people have to be pretty centered.

01:11:49.680 | The less centered you are,

01:11:50.640 | the fewer authentic connections you can really build

01:11:52.880 | for a particular unit of time.

01:11:54.820 | It just takes more time.

01:11:55.880 | Other people have to put up with your crap.

01:11:57.360 | There's just a lot of the stuff that you have to deal with

01:12:00.000 | if you are not so well-balanced.

01:12:02.260 | So yes, we can help people get better

01:12:04.760 | to where they can develop more relationships faster,

01:12:06.880 | and then you can maybe expand Dunbar number by quite a bit,

01:12:09.560 | but you're not gonna do it.

01:12:10.640 | I think it's gonna be hard to get it beyond 10X,

01:12:12.880 | kind of the rough swag of what it is.

01:12:14.800 | - Well, don't you think that AI systems

01:12:19.880 | could be an addition to Dunbar's number?

01:12:22.640 | So like why-

01:12:23.480 | - Do you count as one system or multiple AI systems?

01:12:25.600 | - Multiple AI systems.

01:12:26.560 | So I do believe that AI systems,

01:12:28.800 | for them to integrate into human society as it is now,

01:12:31.400 | have to have a sense of agency.

01:12:32.640 | So there has to be a individual,

01:12:35.280 | because otherwise we wouldn't relate to them.

01:12:37.600 | - We could engage certain kinds of individuals

01:12:40.240 | to make sense of them for us and be almost like,

01:12:42.680 | did you ever watch "Star Trek," like Voyager?

01:12:45.520 | Like there's the Volta who are like the interfaces,

01:12:47.600 | the ambassadors for the Dominion.

01:12:50.440 | We may have ambassadors that speak

01:12:53.080 | on behalf of these systems.

01:12:54.380 | They're like the Mentats of Dune maybe,

01:12:56.120 | or something like this.

01:12:57.240 | I mean, we already have this to some extent.

01:12:59.280 | If you look at the biggest sort of,

01:13:01.120 | I wouldn't say AI system,

01:13:02.100 | but the biggest cybernetic system in the world

01:13:04.040 | is the financial markets.

01:13:05.120 | It runs outside of any individual's control.

01:13:07.960 | And you have an entire stack of people on Wall Street,

01:13:09.880 | Wall Street analysts, to CNBC reporters, whatever.

01:13:13.240 | They're all helping to communicate what does this mean?

01:13:16.920 | You know, Jim Cramer, like running around

01:13:18.760 | and yelling and stuff.

01:13:19.580 | Like all of these people are part of that lowering

01:13:22.580 | of the complexity there to meet sense,

01:13:26.520 | to help do sense-making for people

01:13:28.440 | at whatever capacity they're at.

01:13:29.760 | And I don't see this changing with AI systems.

01:13:31.560 | I think you would have ringside commentators

01:13:33.360 | talking about all this stuff

01:13:34.560 | that this AI system is trying to do over here, over here.

01:13:36.600 | 'Cause it's actually a super intelligence.

01:13:39.120 | So if you wanna talk about humans interfacing,

01:13:40.800 | making first contact with the super intelligence,

01:13:42.460 | we're already there.

01:13:43.600 | We do it pretty poorly.

01:13:44.800 | And if you look at the gradient of power and money,

01:13:47.240 | what happens is the people closest to it

01:13:48.800 | will absolutely exploit their distance

01:13:50.960 | for personal financial gain.

01:13:54.360 | So we should look at that and be like,

01:13:56.080 | oh, well, that's probably what the future

01:13:57.320 | will look like as well.

01:13:58.880 | But nonetheless, I mean,

01:14:00.220 | we're already doing this kind of thing.

01:14:01.360 | So in the future, we can have AI systems,

01:14:03.800 | but you're still gonna have to trust people

01:14:05.700 | to bridge the sense-making gap to them.

01:14:08.400 | - See, I just feel like there could be

01:14:10.760 | of like millions of AI systems that have agencies.

01:14:14.760 | When you say one super intelligence,

01:14:19.480 | super intelligence in that context means

01:14:22.280 | it's able to solve particular problems extremely well.

01:14:26.080 | But there's some aspect of human-like intelligence

01:14:29.240 | that's necessary to be integrated into human society.

01:14:32.300 | So not financial markets,

01:14:33.720 | not sort of weather prediction systems,

01:14:36.740 | or I don't know, logistics optimization.

01:14:39.680 | I'm more referring to things that you interact with

01:14:43.240 | on the intellectual level.

01:14:45.120 | And that I think requires, there has to be a backstory.

01:14:48.920 | There has to be a personality.

01:14:50.080 | I believe it has to fear its own mortality in a genuine way.

01:14:53.320 | Like there has to be all,

01:14:56.540 | many of the elements that we humans experience

01:14:59.680 | that are fundamental to the human condition,

01:15:01.920 | because otherwise we would not have

01:15:03.840 | a deep connection with it.

01:15:05.840 | But I don't think having a deep connection with it

01:15:07.780 | is necessarily going to stop us from building a thing

01:15:10.580 | that has quite an alien intelligence aspect to it.

01:15:13.340 | So the other kind of alien intelligence on this planet

01:15:16.620 | is octopuses or octopodes or whatever you wanna call them.

01:15:20.180 | Octopi, yeah, there's a little controversy

01:15:22.420 | as to what the plural is, I guess.

01:15:23.700 | But an octopus. - I look forward

01:15:25.860 | to your letters.

01:15:26.700 | - Yeah, an octopus, it really acts

01:15:31.140 | as a collective intelligence of eight intelligent arms.

01:15:34.340 | Its arms have a tremendous amount of neural density to them.

01:15:36.980 | And I see if we can build,

01:15:40.340 | I mean, just let's go with what you're saying.

01:15:42.000 | If we build a singular intelligence

01:15:44.380 | that interfaces with humans, that has a sense of agency

01:15:48.080 | so it can run the cybernetic loop

01:15:49.620 | and develop its own theory of mind,

01:15:51.080 | as well as it's a theory of action,

01:15:52.940 | all of these things, I agree with you,

01:15:54.020 | that that's the necessary components

01:15:56.240 | to build a real intelligence, right?

01:15:57.800 | There's gotta be something at stake.

01:15:58.700 | It's gotta make a decision.

01:15:59.980 | It's gotta then run the OODA loop.

01:16:01.260 | Okay, so we build one of those.

01:16:02.980 | Well, if we can build one of those,

01:16:03.820 | we can probably build 5 million of them.

01:16:05.620 | So we build 5 million of them.

01:16:07.380 | And if their cognitive systems are already digitized

01:16:09.940 | and already kind of there,

01:16:12.020 | we stick a antenna on each of them,

01:16:13.700 | bring it all back to a hive mind

01:16:15.400 | that maybe doesn't make all the individual decisions

01:16:17.580 | for them, but treats each one

01:16:19.340 | as almost like a neuronal input

01:16:21.440 | of a much higher bandwidth and fidelity,

01:16:23.820 | going back to a central system

01:16:25.860 | that is then able to perceive much broader dynamics

01:16:30.180 | that we can't see.

01:16:31.060 | In the same way that a phased array radar, right?

01:16:32.580 | You think about how phased array radar works.

01:16:34.420 | It's just sensitivity.

01:16:36.220 | It's just radars, and then it's hypersensitivity

01:16:39.100 | and really great timing between all of them.

01:16:41.140 | And with a flat array,

01:16:42.620 | it's as good as a curved radar dish, right?

01:16:44.740 | So with these things,

01:16:45.580 | it's a phased array of cybernetic systems

01:16:47.800 | that'll give the centralized intelligence

01:16:50.140 | much, much better, much higher fidelity understanding

01:16:55.060 | of what's actually happening in the environment.

01:16:56.580 | - But the more power,

01:16:57.740 | the more understanding the central superintelligence has,

01:17:02.500 | the dumber the individual fingers

01:17:06.600 | of this intelligence are, I think.

01:17:08.100 | - Not necessarily.

01:17:09.020 | - In my sense, this is an argument.

01:17:11.820 | There has to be the experience of the individual agent

01:17:15.660 | has to have the full richness of the human-like experience.

01:17:20.660 | You have to be able to be driving the car in the rain,

01:17:23.820 | listening to Bruce Springsteen,

01:17:25.220 | and all of a sudden break out in tears

01:17:28.260 | because remembering something that happened to you

01:17:30.580 | in high school.

01:17:31.420 | - We can implant those memories

01:17:32.240 | if that's really needed.

01:17:33.080 | But no.

01:17:33.900 | - No, but the central agency,

01:17:35.180 | I guess I'm saying in my view,

01:17:37.780 | for intelligence to be born,

01:17:39.660 | you have to have a decentralization.

01:17:43.620 | Each one has to struggle and reach.

01:17:47.340 | So each one in excess of energy has to reach for order

01:17:51.940 | as opposed to a central place doing so.

01:17:54.460 | - Have you ever read some sci-fi

01:17:55.700 | where there's hive minds?

01:17:58.060 | The Werner Wenz, I think, has one of these,

01:18:01.140 | and then some of the stuff from,

01:18:03.720 | yes, on the "Commonwealth Saga,"

01:18:05.200 | the idea that you're an individual,

01:18:07.000 | but you're connected with a few other individuals

01:18:09.280 | telepathically as well,

01:18:10.340 | and together you form a swarm.

01:18:12.620 | So if you are, I ask you,

01:18:14.640 | what do you think is the experience of if you are,

01:18:17.900 | like, well, a Borg, right?

01:18:18.960 | If you are one, if you're part of this hive mind,

01:18:22.640 | outside of all the aesthetics, forget the aesthetics,

01:18:25.420 | internally, what is your experience like?

01:18:28.400 | 'Cause I have a theory as to what that looks like.

01:18:30.660 | The one question I have for you about that experience

01:18:33.640 | is how much is there a feeling of freedom, of free will?

01:18:38.640 | Because I, obviously, as a human, very unbiased,

01:18:43.340 | but also somebody who values freedom and biased,

01:18:46.160 | it feels like the experience of freedom is essential

01:18:49.680 | for trying stuff out, to being creative

01:18:54.680 | and doing something truly novel, which is at the core of--

01:18:59.140 | - Yeah, well, I don't think you have to lose any freedom

01:19:00.960 | when you're in that mode,

01:19:02.080 | because I think what happens is we think,

01:19:04.600 | we still think, I mean, you're still thinking about this

01:19:06.960 | in a sense of a top-down command and control hierarchy,

01:19:09.840 | which is not what it has to be at all.

01:19:12.320 | I think the experience, so I'll just show my cards here.

01:19:16.080 | I think the experience of being a robot in that robot swarm,

01:19:19.780 | a robot who has agency over their own local environment

01:19:22.880 | that's doing sense-making

01:19:23.920 | and reporting it back to the hive mind,

01:19:26.960 | I think that robot's experience would be,

01:19:29.140 | one, when the hive mind is working well,

01:19:31.980 | it would be an experience of talking to God,

01:19:35.340 | that you essentially are reporting to,

01:19:38.460 | you're sort of saying, "Here's what I see.

01:19:39.660 | "I think this is what's gonna happen over here.

01:19:41.080 | "I'm gonna go do this thing,

01:19:42.140 | "because I think if I'm gonna do this,

01:19:43.760 | "this will make this change happen in the environment."

01:19:46.420 | And then, God, she may tell you, "That's great.

01:19:51.200 | "And in fact, your brothers and sisters will join you

01:19:53.300 | "to help make this go better."

01:19:55.040 | And then she can let your brothers and sisters

01:19:56.900 | know, "Hey, Peter's gonna go do this thing.

01:19:59.420 | "Would you like to help him?

01:20:00.660 | "Because we think that this will make this thing go better."

01:20:02.420 | And they'll say, "Yes, we'll help him."

01:20:03.840 | So the whole thing could be actually a very emergent,

01:20:06.420 | the sense of what does it feel like to be a cell

01:20:10.080 | in a network that is alive, that is generative?

01:20:12.740 | And I think actually the feeling is serendipity,

01:20:16.140 | that there's random order,

01:20:19.380 | not random disorder or chaos, but random order.

01:20:22.380 | Just when you need it to hear Bruce Springsteen,

01:20:24.740 | you turn on the radio and bam,

01:20:26.780 | it's Bruce Springsteen.

01:20:28.540 | That feeling of serendipity,

01:20:30.100 | I feel like this is a bit of a flight of fancy,

01:20:32.460 | but every cell in your body must have,

01:20:35.940 | what does it feel like to be a cell in your body?

01:20:37.900 | When it needs sugar, there's sugar.

01:20:39.940 | When it needs oxygen, there's just oxygen.

01:20:42.220 | Now, when it needs to go and do its work

01:20:43.740 | and pull as one of your muscle fibers,

01:20:46.740 | it does its work and it's great.

01:20:48.820 | It contributes to the cause.

01:20:50.200 | So this is all, again, a flight of fancy,

01:20:52.540 | but I think as we extrapolate up,

01:20:54.460 | what does it feel like to be an independent individual

01:20:57.140 | with some bounded sense of freedom?

01:20:58.780 | All sense of freedom is actually bounded,

01:21:00.220 | but with a bounded sense of freedom

01:21:01.860 | that still lives within a network that has order to it.

01:21:04.980 | And I feel like it has to be a feeling of serendipity.

01:21:07.220 | - So the cell, there's a feeling of serendipity,

01:21:09.940 | even though-

01:21:11.300 | - It has no way of explaining

01:21:12.260 | why it's getting oxygen and sugar when it gets it.

01:21:14.100 | - So you have to, each individual component

01:21:16.180 | has to be too dumb to understand the big picture.

01:21:20.340 | - No, the big picture is bigger

01:21:21.500 | than what it can understand.

01:21:22.980 | - But isn't that an essential characteristic

01:21:24.780 | of the individual is to be too dumb

01:21:28.060 | to understand the bigger picture?

01:21:29.580 | Like, not dumb necessarily,

01:21:31.260 | but limited in its capacity to understand.

01:21:34.140 | 'Cause the moment, okay.

01:21:35.220 | The moment you understand, I feel like that leads to,

01:21:38.900 | if you tell me now that there's some bigger intelligence

01:21:43.540 | controlling everything I do,

01:21:45.780 | intelligence broadly defined, meaning like,

01:21:48.900 | even the Sam Harris thing, there's no free will.

01:21:51.620 | If I'm smart enough to truly understand

01:21:54.460 | that that's the case, that's gonna, I don't know if I-

01:21:59.020 | - We have a philosophical breakdown, right?

01:22:01.020 | Because we're in the West and we're pumped full

01:22:03.100 | of this stuff of like, you are a golden,

01:22:05.740 | fully free individual with all your freedoms

01:22:07.740 | and all your liberties and go grab a gun

01:22:09.380 | and shoot whatever you want to.

01:22:10.540 | No, it's actually, you don't actually have a lot of these.

01:22:14.540 | You're not unconstrained,

01:22:15.880 | but the areas where you can manifest agency,

01:22:20.220 | you're free to do those things.

01:22:21.900 | You can say whatever you want on this podcast.

01:22:23.300 | You can create a podcast, right?

01:22:24.620 | - Yeah.

01:22:25.460 | - You're not, I mean, you have a lot of this kind of freedom,

01:22:27.860 | but even as you're doing this, you are actually,

01:22:30.140 | I guess where the denouement of all of this is that

01:22:33.500 | we are already intelligent agents in such a system, right?

01:22:37.820 | In that one of these like robots of one of 5 million

01:22:41.020 | little swarm robots or one of the Borg,

01:22:43.580 | they're just posting an internal bulletin board.

01:22:45.420 | I mean, maybe the Borg cube is just a giant Facebook machine

01:22:47.620 | floating in space and everyone's just posting on there.

01:22:50.540 | They're just posting really fast and like, oh yeah.

01:22:52.860 | - It's called the metaverse now.

01:22:53.820 | - The net's called the metaverse.

01:22:54.700 | That's right.

01:22:55.520 | Here's the enterprise.

01:22:56.360 | Maybe we should all go shoot it.

01:22:57.200 | Yeah, everyone upvotes and they're gonna go shoot it.

01:22:58.940 | But we already are part of a human online

01:23:02.220 | collaborative environment

01:23:03.720 | and collaborative sense-making system.

01:23:05.720 | It's not very good yet.

01:23:07.380 | It's got the overhangs of zombie sense-making institutions

01:23:10.880 | all over it, but as that washes away

01:23:13.420 | and as we get better at this,

01:23:15.540 | we are going to see humanity improving at speeds

01:23:19.540 | that are unthinkable in the past.

01:23:21.980 | And it's not because anyone's freedoms were limited.

01:23:23.820 | In fact, the open-source,

01:23:24.660 | I mean, we started this with open-source software, right?

01:23:26.820 | The collaboration, what the internet surfaced

01:23:29.340 | was the ability for people all over the world

01:23:31.340 | to collaborate and produce

01:23:32.620 | some of the most foundational software that's in use today.

01:23:35.780 | Right, that entire ecosystem was created

01:23:37.060 | by collaborators all over the place.

01:23:38.920 | So these online kind of swarm kind of things are not novel.

01:23:44.400 | It's just, I'm just suggesting that future AI systems,

01:23:47.480 | if you can build one smart system,

01:23:49.520 | you have no reason not to build multiple.

01:23:51.560 | If you build multiple,

01:23:52.400 | there's no reason not to integrate them all

01:23:53.840 | into a collective sense-making substrate.

01:23:57.800 | And that thing will certainly have emergent intelligence

01:24:00.480 | that non-individuals

01:24:01.800 | and probably not any of the human designers

01:24:03.560 | will be able to really put a bow around and explain.

01:24:06.880 | - But in some sense, would that AI system

01:24:09.960 | still be able to go like rural Texas, buy a ranch,

01:24:14.560 | go off the grid, go full survivalist?

01:24:17.720 | Can you disconnect from the hive mind?

01:24:20.580 | - You may not want to.

01:24:23.660 | - So to be ineffective, to be intelligent.

01:24:28.120 | - You have access to way more intelligence capability

01:24:30.360 | if you're plugged into 5 million

01:24:31.400 | other really, really smart cyborgs.

01:24:33.460 | Why would you leave?

01:24:34.920 | - So like there's a word control that comes to mind.

01:24:38.280 | So it doesn't feel like control,

01:24:40.560 | like overbearing control.

01:24:44.240 | It's just-

01:24:45.360 | - I think systems, well, this is to your point.

01:24:47.120 | I mean, look at how uncomfortable you are

01:24:49.280 | with this concept, right?

01:24:50.600 | I think systems that feel like overbearing control

01:24:53.240 | will not evolutionarily win out.

01:24:55.760 | I think systems that give their individual elements

01:24:58.380 | the feeling of serendipity and the feeling of agency,

01:25:01.440 | that those systems will win.

01:25:05.120 | But that's not to say

01:25:05.960 | that there will not be emergent higher level order.

01:25:08.000 | On top of it.

01:25:08.840 | And that's the thing, that's the philosophical breakdown

01:25:12.000 | that we're staring right at,

01:25:14.160 | which is in the Western mind,

01:25:15.480 | I think there's a very sharp delineation

01:25:18.180 | between explicit control,

01:25:20.300 | Cartesian, like what is the vector?

01:25:23.940 | Where is the position?

01:25:24.920 | Where is it going?

01:25:26.240 | It's completely deterministic.

01:25:27.920 | And kind of this idea that things emerge.

01:25:31.480 | Everything we see is the emergent patterns of other things.

01:25:35.440 | And there is agency when there's extra energy.

01:25:38.140 | - So you have spoken about a kind of meaning crisis

01:25:43.720 | that we're going through.

01:25:44.980 | But it feels like since we invented sex and death,

01:25:50.940 | we broadly speaking,

01:25:54.040 | we've been searching for a kind of meaning.

01:25:56.320 | So it feels like a human civilization

01:25:58.400 | has been going through a meaning crisis

01:25:59.760 | of different flavors throughout its history.

01:26:02.320 | Why is, how is this particular meaning crisis different?

01:26:07.320 | Or is it really a crisis and it wasn't previously?

01:26:10.560 | What's your sense?

01:26:11.600 | - A lot of human history,

01:26:13.200 | there wasn't so much a meaning crisis.

01:26:14.780 | There was just a like food

01:26:15.840 | and not getting eaten by bears crisis, right?

01:26:18.680 | Once you get to a point where you can make food,

01:26:20.400 | there was the like not getting killed

01:26:21.840 | by other humans crisis.

01:26:23.640 | So sitting around wondering what is it all about,

01:26:26.320 | it's actually a relatively recent luxury.

01:26:29.760 | And to some extent,

01:26:31.600 | the meaning crisis coming out of that is precisely because,

01:26:35.120 | well, it's not precisely because I believe

01:26:37.080 | that meaning is the consequence of

01:26:39.280 | when we make consequential decisions,

01:26:43.380 | it's tied to agency, right?

01:26:45.640 | When we make consequential decisions,

01:26:47.960 | that generates meaning.

01:26:49.840 | So if we make a lot of decisions,

01:26:51.200 | but we don't see the consequences of them,

01:26:53.300 | then it feels like what was the point, right?

01:26:55.480 | But if there's all these big things happening,

01:26:57.140 | but we're just along for the ride,

01:26:58.440 | then it also does not feel very meaningful.

01:27:00.680 | Meaning, as far as I can tell,

01:27:02.040 | this is my working definition of circa 2021,

01:27:04.600 | is generally the result of a person

01:27:08.400 | making a consequential decision,

01:27:09.800 | acting on it and then seeing the consequences of it.

01:27:12.160 | So historically, just when humans are in survival mode,

01:27:16.720 | you're making consequential decisions all the time.

01:27:19.500 | So there's not a lack of meaning

01:27:20.900 | because like you either got eaten or you didn't, right?

01:27:23.440 | You got some food and that's great, you feel good.

01:27:25.560 | Like these are all consequential decisions.

01:27:27.440 | Only in the post fossil fuel and industrial revolution,

01:27:32.440 | could we create a massive leisure class.

01:27:36.920 | I could sit around not being threatened by bears,

01:27:39.320 | not starving to death,

01:27:40.520 | making decisions somewhat,

01:27:44.840 | but a lot of times not seeing the consequences

01:27:47.440 | of any decisions they make.

01:27:49.240 | The general sort of sense of anomie,

01:27:51.600 | I think that's the French term for it,

01:27:53.480 | in the wake of the consumer society,

01:27:55.600 | in the wake of mass media telling everyone,

01:27:58.680 | hey, choosing between Hermes and Chanel

01:28:02.160 | is a meaningful decision.

01:28:03.240 | No, it's not.

01:28:04.080 | - I don't know what either of those mean.

01:28:05.760 | - Oh, there's a high end luxury purses and crap like that.

01:28:10.760 | But the point is that we give people the idea

01:28:13.680 | that consumption is meaning,

01:28:15.200 | that making a choice of this team versus that team,

01:28:17.560 | spectating has meaning.

01:28:20.240 | So we produce all of these different things

01:28:22.580 | that are as if meaning, right?

01:28:25.400 | But really making a decision

01:28:27.000 | that has no consequences for us.

01:28:28.960 | And so that creates the meaning crisis.

01:28:31.080 | - Well, you're saying choosing between Chanel

01:28:33.400 | and the other one has no consequence.

01:28:35.440 | I mean, why is one more meaningful than the other?

01:28:38.320 | - It's not that it's more meaningful than the other,

01:28:39.640 | it's that you make a decision between these two brands

01:28:42.600 | and you're told this brand will make me look better

01:28:45.320 | in front of other people.

01:28:46.140 | If I buy this brand of car,

01:28:47.720 | if I wear that brand of apparel, right?

01:28:50.240 | The idea, like a lot of decisions we make

01:28:52.600 | are around consumption,

01:28:54.320 | but consumption by itself doesn't actually yield meaning.

01:28:57.240 | Gaining social status does provide meaning.

01:29:00.080 | So that's why in this era of abundant production,

01:29:04.000 | so many things turn into status games.

01:29:07.520 | The NFT kind of explosion is a similar kind of thing.

01:29:10.020 | Everywhere there are status games

01:29:12.000 | because we just have so much excess production.

01:29:15.480 | - But aren't those status games a source of meaning?

01:29:18.440 | Like why do the games we play have to be grounded

01:29:22.480 | in physical reality like they are

01:29:24.240 | when you're trying to run away from lions?

01:29:26.200 | Why can't we in this virtuality world, on social media,

01:29:30.440 | why can't we play the games on social media,

01:29:32.280 | even the dark ones?

01:29:33.440 | - Right, we can, we can.

01:29:35.280 | - But you're saying that's creating a meaning crisis.

01:29:37.760 | - Well, there's a meaning crisis

01:29:39.200 | in that there's two aspects of it.

01:29:41.120 | Number one, playing those kinds of status games

01:29:44.440 | oftentimes requires destroying the planet

01:29:47.420 | because it ties to consumption,

01:29:52.620 | consuming the latest and greatest version of a thing,

01:29:55.100 | buying the latest limited edition sneaker

01:29:57.620 | and throwing out all the old ones.

01:29:58.900 | Maybe it keeps in the old ones,

01:29:59.740 | but the amount of sneakers we have to cut up

01:30:01.740 | and destroy every year to create artificial scarcity

01:30:05.020 | for the next generation, right?

01:30:06.460 | This is kind of stuff that's not great.

01:30:08.500 | It's not great at all.

01:30:09.600 | So conspicuous consumption fueling status games

01:30:14.120 | is really bad for the planet, not sustainable.

01:30:17.200 | The second thing is you can play these kinds of status games

01:30:20.780 | but then what it does is it renders you captured

01:30:23.500 | to the virtual environment.

01:30:25.420 | The status games that really wealthy people are playing

01:30:27.780 | are all around the hard resources

01:30:30.260 | where they're gonna build the factories,

01:30:31.460 | they're gonna have the fuel and the rare earths

01:30:32.820 | to make the next generation of robots.

01:30:34.260 | They're then going to run circles around you

01:30:36.660 | and your children.

01:30:38.180 | So that's another reason not to play

01:30:39.700 | those virtual status games.

01:30:41.100 | - So you're saying ultimately the big picture game

01:30:44.700 | is won by people who have access or control

01:30:48.340 | over actual hard resources.

01:30:49.860 | So you don't see a society where most of the games

01:30:54.340 | are played in the virtual space.

01:30:56.860 | - They'll be captured in the physical space.

01:30:58.940 | It all builds.

01:30:59.780 | It's just like the stack of human being, right?

01:31:02.360 | If you only play the game at the cultural

01:31:05.980 | and then intellectual level,

01:31:07.540 | then the people with the hard resources

01:31:08.820 | and access to layer zero physical are going to own you.

01:31:12.940 | - But isn't money not connected to

01:31:15.140 | or less and less connected to hard resources

01:31:17.380 | and money still seems to work?

01:31:18.880 | It's a virtual technology.

01:31:20.580 | - There's different kinds of money.

01:31:22.380 | Part of the reason that some of the stuff

01:31:23.740 | is able to go a little unhinged

01:31:26.260 | is because the big sovereignties where one spends money

01:31:31.260 | and uses money and plays money games and inflates money,

01:31:36.260 | their ability to adjudicate the physical resources

01:31:40.340 | and hard resources and land and things like that,

01:31:42.380 | those have not been challenged in a very long time.

01:31:45.620 | - So, we went off the gold standard.

01:31:47.660 | Most money is not connected to physical resources.

01:31:51.540 | It's an idea.

01:31:52.960 | And that idea is very closely connected to status.

01:31:57.880 | - But it's also tied to, it's actually tied to law.

01:32:03.160 | It is tied to some physical hard things, right?

01:32:05.040 | You have to pay your taxes.

01:32:06.080 | - Yes, so it's always at the end going to be connected

01:32:10.080 | to the blockchain of physical reality.

01:32:12.760 | So, in the case of law and taxes,

01:32:15.560 | it's connected to government

01:32:17.340 | and government is what violence is the,

01:32:21.640 | I'm playing on the stacks of devil's advocates here.

01:32:25.320 | I'm popping one devil off the stack at a time.

01:32:30.680 | Isn't ultimately, of course,

01:32:31.760 | it'll be connected to physical reality,

01:32:33.160 | but just because people control the physical reality

01:32:35.760 | doesn't mean the status,

01:32:37.840 | LeBron James in theory could make more money

01:32:39.800 | than the owners of the teams in theory.

01:32:43.440 | And to me, that's a virtual idea.

01:32:45.040 | So somebody else constructed a game

01:32:47.500 | and now you're playing in the space of virtual,

01:32:50.040 | in the virtual space of the game.

01:32:51.900 | And so it just feels like there could be games

01:32:54.700 | where status, we build realities

01:32:57.580 | that give us meaning in the virtual space.

01:33:00.120 | Like I can imagine such things being possible.

01:33:02.960 | - Oh yeah, okay.

01:33:03.800 | So I see what you're,

01:33:04.640 | I think I see what you're saying there.

01:33:05.640 | With the idea there, I mean,

01:33:07.680 | we'll take the LeBron James side

01:33:08.960 | and put in like some YouTube influencer.

01:33:10.820 | - Yes, sure.

01:33:11.660 | - Right, so the YouTube influencer,

01:33:13.580 | it is status games, but at a certain level,

01:33:16.600 | it precipitates into real dollars.

01:33:19.040 | And into like, well, you look at Mr. Beast, right?

01:33:21.280 | He's like sending off half a million dollars

01:33:23.360 | worth of fireworks or something, right?

01:33:24.600 | Not a YouTube video.

01:33:25.820 | - And also like saving, you know,

01:33:27.200 | like saving trees and so on.

01:33:28.640 | - Sure, right.

01:33:29.480 | And trying to plant a million trees

01:33:30.300 | with Mark Rober or whatever it was.

01:33:31.140 | Yeah, like it's not that those kinds of games

01:33:33.100 | can't lead to real consequences.

01:33:35.480 | It's that for the vast majority of people

01:33:38.680 | in consumer culture,

01:33:41.120 | they are incented by the,

01:33:44.660 | I would say mostly I'm thinking about middle-class consumers.

01:33:47.820 | They're incented by advertisements.

01:33:49.700 | They're scented by their memetic environment

01:33:51.780 | to treat the purchasing of certain things,

01:33:55.660 | the need to buy the latest model of whatever,

01:33:57.360 | the need to appear however,

01:33:59.220 | the need to pursue status games as a driver of meaning.

01:34:03.820 | And my point would be that

01:34:04.900 | it's a very hollow driver of meaning.

01:34:07.220 | And that is what creates a meaning crisis.

01:34:09.600 | Because at the end of the day,

01:34:11.380 | it's like eating a lot of empty calories, right?

01:34:13.400 | Yeah, it tasted good going down,

01:34:14.660 | it's a lot of sugar, but man, it did not,

01:34:16.220 | it was not enough protein to help build your muscles.

01:34:18.380 | And you kind of feel that in your gut.

01:34:20.580 | And I think that's, I mean,

01:34:21.580 | to all this stuff aside

01:34:22.660 | and setting aside our discussion on currency,

01:34:24.220 | which I hope we get back to,

01:34:25.980 | that's what I mean about the meaning crisis,

01:34:28.860 | part of it being created by the fact that we don't,

01:34:31.620 | we're not encouraged to have

01:34:34.700 | more and more direct relationships.

01:34:36.660 | We're actually alienated from relating to,

01:34:40.320 | even our family members sometimes, right?

01:34:42.920 | We're encouraged to relate to brands.

01:34:46.520 | We're encouraged to relate to these kinds of things

01:34:48.760 | that then tell us to do things

01:34:51.760 | that are really of low consequence.

01:34:53.800 | And that's where the meaning crisis comes from.

01:34:55.480 | - So the role of technology in this,

01:34:57.240 | so there's somebody you mentioned who's Jacques Eliel,

01:35:00.640 | his view of technology,

01:35:04.240 | he warns about the towering piles of technique,

01:35:06.940 | which I guess is a broad idea of technology.

01:35:10.260 | So I think, correct me if I'm wrong,

01:35:12.400 | for him, technology is bad,

01:35:15.580 | it moving away from human nature

01:35:17.180 | and it's ultimately is destructive.

01:35:19.340 | My question broadly speaking, this meaning crisis,

01:35:21.460 | can technology, what are the pros and cons of technology?

01:35:24.340 | Can it be a good?

01:35:25.540 | - Yeah, I think it can be.

01:35:26.900 | I certainly draw on some of the Lowell's ideas

01:35:29.420 | and I think some of them are pretty good,

01:35:32.560 | but the way he defines technique is,

01:35:35.900 | well, also Simondon as well.

01:35:37.420 | I mean, he speaks to the general mentality of efficiency,

01:35:40.940 | homogenized processes, homogenized production,

01:35:43.340 | homogenized labor to produce homogenized artifacts

01:35:46.820 | that then are not actually,

01:35:49.100 | they don't sit well in the environment.

01:35:52.780 | So it's essentially,

01:35:53.620 | you can think of it as the antonym of craft,

01:35:57.740 | whereas a craftsman will come to a problem,

01:36:01.940 | maybe a piece of wood and they need to make into a chair,

01:36:04.200 | it may be a site to build a house or build a stable

01:36:06.480 | or build whatever,

01:36:08.440 | and they will consider how to bring various things in

01:36:12.200 | to build something well contextualized

01:36:15.000 | that's in right relationship with that environment.

01:36:20.000 | But the way we have driven technology

01:36:22.280 | over the last 100, 150 years is not that at all.

01:36:25.720 | It is how can we make sure the input materials

01:36:30.480 | are homogenized, cut to the same size,

01:36:33.380 | diluted and doped to exactly the right

01:36:35.240 | alloy concentrations.

01:36:36.820 | How do we create machines that then consume

01:36:38.380 | exactly the right kind of energy

01:36:39.460 | to be able to run at this high speed

01:36:40.620 | to stamp out the same parts,

01:36:42.580 | which then go out the door,

01:36:44.080 | everyone gets the same tickle me Elmo.

01:36:45.740 | And the reason why everyone wants it

01:36:46.780 | is because we have broadcasts that tells everyone

01:36:49.280 | this is the cool thing.

01:36:50.540 | So we homogenize demand, right?

01:36:52.600 | And we're like Baudrillard and other critiques

01:36:55.460 | of modernity coming from that direction,

01:36:57.540 | the situation list as well.

01:36:59.260 | It's that their point is that at this point in time,

01:37:02.100 | consumption is the thing that drives

01:37:04.600 | a lot of the economic stuff, not the need,

01:37:06.700 | but the need to consume and build status games on top.

01:37:09.420 | So we have homogenized when we discovered,

01:37:12.140 | I think this is really like Bernays and stuff, right?

01:37:14.820 | In the early 20th century, we discovered we can create,

01:37:17.940 | we can create demand, we can create desire

01:37:20.900 | in a way that was not possible before

01:37:23.580 | because of broadcast media.

01:37:25.580 | And not only do we create desire,

01:37:27.700 | we don't create a desire for each person

01:37:29.380 | to connect to some bespoke thing,

01:37:31.100 | to build a relationship with their neighbor or their spouse.

01:37:33.660 | We are telling them, you need to consume this brand.

01:37:36.100 | You need to drive this vehicle.

01:37:37.220 | You got to listen to this music.

01:37:38.300 | Have you heard this?

01:37:39.340 | Have you seen this movie, right?

01:37:40.940 | So creating homogenized demand makes it really cheap

01:37:44.820 | to create homogenized product.

01:37:46.540 | And now you have economics of scale.

01:37:48.580 | So we make the same tickle me Elmo,

01:37:50.060 | give it to all the kids and all the kids are like,

01:37:52.820 | hey, I got a tickle me Elmo, right?

01:37:54.440 | So this is ultimately where this ties in then

01:37:58.660 | to runaway hyper capitalism is that we then,

01:38:03.080 | capitalism is always looking for growth.

01:38:04.840 | It's always looking for growth

01:38:05.980 | and growth only happens to the margins.

01:38:07.960 | So you have to squeeze more and more demand out.

01:38:09.960 | You got to make it cheaper and cheaper

01:38:11.080 | to make the same thing,

01:38:12.280 | but tell everyone they're still getting meaning from it.

01:38:15.120 | You're still like, this is still your tickle me Elmo, right?

01:38:18.080 | And we see little bits of this dripping,

01:38:20.960 | critiques of this dripping in popular culture.

01:38:22.840 | You see it sometimes it's when Buzz Lightyear

01:38:25.980 | walks into the thing, he's like,

01:38:27.820 | oh my God, at the toy store, I'm just a toy.

01:38:30.680 | Like there's millions of other,

01:38:31.800 | or there's hundreds of other Buzz Lightyear's

01:38:33.440 | just like me, right?

01:38:34.800 | That is, I think, a fun Pixar critique

01:38:38.120 | on this homogenization dynamic.

01:38:40.120 | - I agree with you on most of the things you're saying.

01:38:42.960 | So I'm playing devil's advocate here,

01:38:44.600 | but this homogenized machine of capitalism

01:38:50.640 | is also the thing that is able to fund,

01:38:54.280 | if channeled correctly, innovation, invention,

01:38:59.120 | and development of totally new things

01:39:00.800 | that in the best possible world

01:39:02.320 | create all kinds of new experiences

01:39:03.880 | that can enrich lives,

01:39:05.640 | the quality of lives for all kinds of people.

01:39:09.880 | So isn't this the machine

01:39:12.400 | that actually enables the experiences

01:39:15.160 | and more and more experiences that would then give meaning?

01:39:18.680 | - It has done that to some extent.

01:39:21.080 | I mean, it's not all good or bad in my perspective.

01:39:24.720 | We can always look backwards

01:39:26.800 | and offer a critique of the path we've taken

01:39:29.160 | to get to this point in time,

01:39:30.680 | but that's a different, that's somewhat different

01:39:33.800 | and informs the discussion,

01:39:35.940 | but it's somewhat different than the question

01:39:37.720 | of where do we go in the future, right?

01:39:40.640 | Is this still the same rocket we need to ride

01:39:42.760 | to get to the next point?

01:39:43.600 | Will it even get us to the next point?

01:39:44.600 | - Well, how does this, so you're predicting the future,

01:39:46.280 | how does it go wrong in your view?

01:39:48.840 | - We have the mechanisms,

01:39:51.120 | we have now explored enough technologies

01:39:54.000 | to where we can actually, I think, sustainably produce

01:39:57.800 | what most people in the world need to live.

01:40:02.440 | We have also created the infrastructures

01:40:07.720 | to allow continued research and development

01:40:10.480 | of additional science and medicine

01:40:13.160 | and various other kinds of things.

01:40:16.160 | The organizing principles that we use

01:40:18.600 | to govern all these things today

01:40:20.200 | have been, a lot of them have been just inherited

01:40:25.600 | from, honestly, medieval times.

01:40:28.520 | Some of them have refactored a little bit

01:40:30.200 | in the industrial era,

01:40:32.000 | but a lot of these modes of organizing people

01:40:35.960 | are deeply problematic.

01:40:38.320 | And furthermore, they're rooted in,

01:40:43.080 | I think, a very industrial mode perspective on human labor.

01:40:47.680 | And this is one of those things,

01:40:49.160 | I'm gonna go back to the open source thing.

01:40:51.080 | There was a point in time when,

01:40:53.240 | well, let me ask you this.

01:40:54.920 | If you look at the core SciPy sort of collection of libraries,

01:40:58.480 | so SciPy, NumPy, Matplotlib, right?

01:41:00.680 | There's IPython Notebook, let's throw Pandas in there,

01:41:02.800 | Scikit-learn, a few of these things.

01:41:04.600 | How much value do you think, economic value,

01:41:09.640 | would you say they drive in the world today?

01:41:12.840 | - That's one of the fascinating things

01:41:14.960 | about talking to you and Travis.

01:41:16.280 | It's like, it's a measure of what's like--

01:41:20.640 | - At least a billion dollars a day, maybe?

01:41:22.480 | - Billion dollars, sure.

01:41:23.600 | I mean, it's like, it's similar question of like,

01:41:26.040 | how much value does Wikipedia create?

01:41:28.520 | - Right.

01:41:29.360 | - It's like, all of it, I don't know.

01:41:32.720 | - Well, I mean, if you look at our systems,

01:41:34.600 | when you do a Google search, right?

01:41:36.040 | Now, some of that stuff runs through TensorFlow,

01:41:37.640 | but when you look at, you know, Siri,

01:41:40.000 | when you do credit card transaction,

01:41:41.600 | fraud, like just everything, right?

01:41:43.280 | Every intelligence agency under the sun,

01:41:45.160 | they're using some aspect of these kinds of tools.

01:41:47.600 | So I would say that these create billions of dollars

01:41:51.120 | of value.

01:41:51.960 | - Oh, you mean like direct use of tools that leverage--

01:41:54.000 | - Yeah, direct, yeah.

01:41:54.840 | - Yeah, even that's billions a day, yeah.

01:41:56.640 | - Yeah, right, easily, I think.

01:41:58.720 | Like the things they could not do

01:41:59.720 | if they didn't have these tools, right?

01:42:01.080 | - Yes.

01:42:01.920 | - So that's billions of dollars a day, great.

01:42:04.840 | I think that's about right.

01:42:05.720 | Now, if we take how many people did it take to make that?

01:42:08.720 | (Travis laughs)

01:42:09.760 | Right, and there was a point in time, not anymore,

01:42:11.680 | but there was a point in time when they could fit in a van.

01:42:13.520 | I could have fit them in my Mercedes Spinter, right?

01:42:15.920 | And so if you look at that, like, holy crap,

01:42:19.200 | literally a van of maybe a dozen people

01:42:22.440 | could create value to the tune of billions of dollars a day.

01:42:26.880 | - What lesson do you draw from that?

01:42:30.040 | - Well, here's the thing.

01:42:31.360 | What can we do to do more of that?

01:42:34.160 | Like that's open source.

01:42:36.280 | The way I've talked about this in other environments is,

01:42:39.680 | when we use generative participatory crowdsourced approaches

01:42:44.360 | we unlock human potential at a level

01:42:48.480 | that is better than what capitalism can do.

01:42:50.720 | I would challenge anyone to go and try to hire

01:42:55.400 | the right 12 people in the world

01:42:58.240 | to build that entire stack

01:43:00.160 | the way those 12 people did that, right?

01:43:02.440 | They'd be very, very hard pressed to do that.

01:43:04.080 | If a hedge fund could just hire a dozen people

01:43:06.680 | and create like something

01:43:07.680 | that is worth billions of dollars a day,

01:43:10.040 | every single one of them would be racing to do it, right?

01:43:12.320 | But finding the right people,

01:43:13.600 | fostering the right collaborations,

01:43:15.080 | getting it adopted by the right other people

01:43:16.760 | to then refine it,

01:43:18.000 | that is a thing that was organic in nature.

01:43:20.640 | That took crowdsourcing,

01:43:22.120 | that took a lot of the open source ethos

01:43:24.080 | and it took the right kinds of people, right?

01:43:26.400 | None of those people who started that said,

01:43:27.800 | I need to have a part of a multi-billion dollar a day

01:43:30.840 | sort of enterprise.

01:43:32.400 | They're like, I'm doing this cool thing

01:43:33.480 | to solve my problem for my friends, right?

01:43:35.440 | So the point of telling the story

01:43:37.800 | is to say that our way of thinking about value,

01:43:40.720 | our way of thinking about allocation of resources,

01:43:42.840 | our ways of thinking about property rights

01:43:44.840 | and all these kinds of things,

01:43:46.120 | they come from finite game,

01:43:48.800 | scarcity mentality, medieval institutions.

01:43:51.240 | As we are now entering,

01:43:54.160 | to some extent we're sort of in a post-scarcity era,

01:43:57.040 | although some people are hoarding a whole lot of stuff.

01:43:59.760 | We are at a point where,

01:44:01.280 | if not now, soon we'll be in a post-scarcity era.

01:44:03.880 | The question of how we allocate resources

01:44:06.440 | has to be revisited at a fundamental level,

01:44:08.680 | because the kind of software these people built,

01:44:10.960 | the modalities that those human ecologies

01:44:13.920 | that built that software,

01:44:15.800 | it treats software as unproperty.

01:44:17.920 | Actually sharing creates value.

01:44:20.440 | Restricting and forking reduces value.

01:44:23.040 | So that's different than any other physical resource

01:44:26.320 | that we've ever dealt with.

01:44:27.160 | It's different than how most corporations

01:44:28.720 | treat software IP, right?

01:44:31.200 | So if treating software in this way

01:44:34.560 | created this much value so efficiently, so cheaply,

01:44:37.520 | 'cause feeding a dozen people for 10 years is really cheap.

01:44:40.480 | Right?

01:44:41.600 | That's the reason I care about this right now,

01:44:44.640 | is because looking forward

01:44:46.000 | when we can automate a lot of labor,

01:44:47.920 | where we can in fact,

01:44:49.520 | the programming for your robot

01:44:51.440 | in your part of neck of the woods

01:44:52.760 | and your part of the Amazon

01:44:54.040 | to build something sustainable for you and your tribe

01:44:56.960 | to deliver the right medicines, to take care of the kids,

01:45:00.160 | that's just software, that's just code.

01:45:02.800 | That could be totally open sourced.

01:45:04.840 | Right?

01:45:05.680 | So we can actually get to a mode

01:45:07.280 | where all of this additional generative things

01:45:10.840 | that humans are doing,

01:45:12.320 | they don't have to be wrapped up in a container

01:45:16.120 | and then we charge

01:45:17.000 | for all the exponential dynamics out of it.

01:45:19.120 | That's what Facebook did.

01:45:20.320 | That's what modern social media did.

01:45:22.120 | Right?

01:45:22.960 | 'Cause the old internet was connecting people just fine.

01:45:24.840 | So Facebook came along and said,

01:45:25.880 | well, anyone can post a picture,

01:45:26.840 | anyone can post some text.

01:45:28.320 | And we're gonna amplify the crap out of it

01:45:29.920 | to everyone else.

01:45:31.040 | And it exploded this generative network

01:45:33.240 | of human interaction.

01:45:34.640 | And then I said, how do I make money off that?

01:45:36.000 | Oh yeah, I'm gonna be a gatekeeper

01:45:38.080 | on everybody's attention.

01:45:39.800 | And that's how we make money.

01:45:40.960 | - So how do we create more than one van?

01:45:45.560 | How do we have millions of vans full of people

01:45:47.760 | that create NumPy, SciPy, that create Python?

01:45:50.920 | So, you know, the story of those people

01:45:54.320 | is often they have some kind of job outside of this.

01:45:57.000 | This is what they're doing for fun.

01:45:58.800 | Don't you need to have a job?

01:46:00.840 | Don't you have to be connected,

01:46:02.160 | plugged in to the capitalist system?

01:46:04.880 | Isn't that what, like, isn't this consumerism

01:46:09.080 | the engine that results in the individuals

01:46:13.800 | that kind of take a break from it

01:46:15.120 | every once in a while to create something magical?

01:46:17.240 | Like at the edges is where the innovation happens.

01:46:18.800 | - Right, the question of surplus, right?

01:46:20.280 | This is the question.

01:46:21.360 | Like if everyone were to go and run their own farm,

01:46:24.400 | no one would have time to go and write NumPy, SciPy, right?

01:46:27.320 | Maybe, but that's what I'm talking about

01:46:29.960 | when I say we're maybe at a post-scarcity point

01:46:32.800 | for a lot of people.

01:46:34.160 | The question that we're never encouraged to ask

01:46:37.280 | in a Super Bowl ad is how much do you need?

01:46:40.680 | How much is enough?

01:46:42.000 | Do you need to have a new car every two years, every five?

01:46:45.080 | If you have a reliable car,

01:46:46.200 | can you drive one for 10 years, is that all right?

01:46:47.960 | You know, I had a car for 10 years and it was fine.

01:46:50.160 | You know, your iPhone,

01:46:51.080 | do you have to upgrade every two years?

01:46:52.800 | I mean, it's sort of, you're using the same apps

01:46:54.320 | you did four years ago, right?

01:46:56.680 | - This should be a Super Bowl ad.

01:46:58.320 | - This should be a Super Bowl ad, that's great.

01:46:59.680 | Maybe somebody- - Do you really need

01:47:00.680 | a new iPhone?

01:47:01.520 | - Maybe one of our listeners will fund something like this

01:47:03.920 | of like, no, but just actually bringing it back,

01:47:06.960 | bringing it back to actually the question of

01:47:09.840 | what do you need?

01:47:11.480 | How do we create the infrastructure

01:47:13.560 | for collectives of people to live on the basis of

01:47:18.200 | providing what we need, meeting people's needs

01:47:21.040 | with a little bit of excess to handle emergencies

01:47:23.240 | and things like that.

01:47:24.440 | Pulling our resources together

01:47:26.320 | to handle the really, really big emergencies,

01:47:28.800 | somebody with a really rare form of cancer

01:47:30.880 | or some massive fire sweeps through, you know,

01:47:32.640 | half the village or whatever.

01:47:34.920 | But can we actually unscale things

01:47:38.200 | and solve for people's needs

01:47:41.560 | and then give them the capacity

01:47:44.120 | to explore how to be the best version of themselves?

01:47:47.320 | And for Travis, that was, you know,

01:47:49.000 | throwing away his shot at tenure in order to write "Numbhigh."

01:47:52.880 | For others, there is a saying in the sci-fi community

01:47:56.840 | that, you know, sci-fi advances

01:47:58.160 | one failed postdoc at a time.

01:47:59.800 | And that's, you know, we can do these things.

01:48:03.800 | We can actually do this kind of collaboration

01:48:05.600 | because code, software, information organization,

01:48:08.360 | that's cheap.

01:48:09.880 | Those bits are very cheap to fling across the oceans.

01:48:13.000 | - So you mentioned Travis.

01:48:14.760 | We've been talking

01:48:15.680 | and we'll continue to talk about open source.

01:48:17.920 | Maybe you can comment.

01:48:20.400 | How did you meet Travis?

01:48:21.600 | Who is Travis Alphont?

01:48:24.080 | What's your relationship been like through the years?

01:48:27.440 | Where did you work together?

01:48:30.160 | How did you meet?

01:48:31.840 | What's the present and the future look like?

01:48:35.120 | - Yeah, so the first time I met Travis

01:48:36.600 | was at a sci-fi conference in Pasadena.

01:48:39.360 | - Do you remember the year?

01:48:40.920 | - 2005.

01:48:42.040 | I was working at, again, at Enthought,

01:48:44.200 | you know, working on scientific computing, consulting.

01:48:47.000 | And a couple of years later,

01:48:51.200 | he joined us at Enthought, I think 2007.

01:48:54.040 | And he came in as the president.

01:48:58.240 | One of the founders of Enthought was the CEO, Eric Jones.

01:49:01.920 | And we were all very excited that Travis was joining us.

01:49:04.080 | And that was, you know, great fun.

01:49:05.080 | And so I worked with Travis

01:49:06.960 | on a number of consulting projects

01:49:08.920 | and we worked on some open source stuff.

01:49:12.120 | I mean, it was just a really, it was a good time there.

01:49:15.080 | And then-

01:49:15.920 | - It was primarily Python related?

01:49:17.840 | - Oh yeah, it was all Python,

01:49:18.760 | NumPy, sci-fi consulting kind of stuff.

01:49:21.000 | Towards the end of that time,

01:49:23.280 | we started getting called into more and more finance shops.

01:49:27.760 | They were adopting Python pretty heavily.

01:49:29.880 | I did some work on like a high-frequency trading shop,

01:49:33.360 | working on some stuff.

01:49:34.200 | And then we worked together on some,

01:49:36.520 | at a couple of investment banks in Manhattan.

01:49:39.880 | And so we started seeing that there was a potential

01:49:42.720 | to take Python in the direction of business computing.

01:49:45.760 | More than just being this niche,

01:49:46.840 | like MATLAB replacement for big vector computing.

01:49:50.560 | What we were seeing was, oh yeah,

01:49:51.840 | you could actually use Python as a Swiss army knife

01:49:53.920 | to do a lot of shadow data transformation kind of stuff.

01:49:56.880 | So that's when we realized the potential is much greater.

01:50:00.560 | And so we started Anaconda.

01:50:03.400 | I mean, it was called Continuum Analytics at the time,

01:50:05.240 | but we started in January of 2012

01:50:07.560 | with a vision of shoring up the parts of Python

01:50:10.800 | that needed to get expanded to handle data at scale,

01:50:13.800 | to do web visualization, application development, et cetera.

01:50:17.240 | And that was that, yeah.

01:50:18.080 | So he was CEO and I was president for the first five years.

01:50:23.080 | And then we raised some money and then the board

01:50:27.840 | sort of put in a new CEO.

01:50:29.000 | They hired a kind of professional CEO.

01:50:31.320 | And then Travis, you laugh out that.

01:50:34.080 | I took over the CTO role.

01:50:35.240 | Travis then left after a year to do his own thing,

01:50:37.920 | to do QuantSight,

01:50:39.600 | which was more oriented around some of the bootstrap years

01:50:43.000 | that we did at Continuum,

01:50:43.920 | where it was open source and consulting.

01:50:46.160 | It wasn't sort of like gung-ho product development.

01:50:48.560 | And it wasn't focused on,

01:50:50.080 | we accidentally stumbled into the package management problem

01:50:53.840 | at Anaconda, but then we had a lot of other visions

01:50:56.920 | of other technology that we built in the open source.

01:50:58.880 | And Travis was really trying to push, again,

01:51:02.360 | the frontiers of numerical computing, vector computing,

01:51:05.280 | handling things like auto differentiation and stuff

01:51:07.680 | intrinsically in the open ecosystem.

01:51:09.920 | So I think that's the,

01:51:13.640 | that's kind of the direction he's working on

01:51:16.600 | in some of his work.

01:51:18.280 | We remain great friends and colleagues and collaborators,

01:51:22.520 | even though he's no longer day-to-day working at Anaconda,

01:51:25.760 | but he gives me a lot of feedback about

01:51:27.840 | this and that and the other.

01:51:29.000 | - What's a big lesson you've learned from Travis

01:51:32.200 | about life or about programming or about leadership?

01:51:35.440 | - Wow, there's a lot.

01:51:36.480 | There's a lot.

01:51:37.320 | Travis is a really, really good guy.

01:51:39.600 | He really, his heart is really in it.

01:51:41.920 | He cares a lot.

01:51:43.080 | - I've gotten that sense having to interact with him.

01:51:46.920 | It's so interesting.

01:51:47.760 | - Yeah. - Such a good human being.

01:51:48.600 | - He's a really good dude.

01:51:49.680 | And he and I, it's so interesting.

01:51:51.360 | We come from very different backgrounds.

01:51:53.200 | We're quite different as people,

01:51:54.900 | but I think we can not talk for a long time

01:52:00.780 | and then be on a conversation and be eye-to-eye

01:52:04.280 | on like 90% of things.

01:52:06.400 | And so he's someone who I believe,

01:52:08.280 | no matter how much fog settles in over the ocean,

01:52:10.600 | his ship, my ship are pointed sort of

01:52:12.480 | in the same direction to the same star.

01:52:14.120 | - Wow, that's a beautiful way to phrase it.

01:52:16.840 | No matter how much fog there is,

01:52:18.600 | we're pointed at the same star.

01:52:20.400 | - Yeah, and I hope he feels the same way.

01:52:21.880 | I mean, I hope he knows that over the years now.

01:52:23.780 | We both care a lot about the community.

01:52:27.000 | For someone who cares so deeply,

01:52:28.120 | I would say this about Travis, it's interesting.

01:52:29.880 | For someone who cares so deeply about the nerd details

01:52:33.360 | of like type system design and vector computing

01:52:36.000 | and efficiency of expressing this and that and the other,

01:52:38.760 | memory layouts and all that stuff,

01:52:40.440 | he cares even more about the people

01:52:43.280 | in the ecosystem, the community.

01:52:45.880 | And I have a similar kind of alignment.

01:52:49.760 | I care a lot about the tech, I really do.

01:52:53.080 | But for me, the beauty of what this human ecology

01:52:58.080 | has produced is I think a touchstone.

01:53:01.740 | It's an early version.

01:53:02.680 | We should look at it and say,

01:53:03.680 | how do we replicate this for humanity at scale?

01:53:05.880 | What this open source collaboration was able to produce?

01:53:08.840 | How can we be generative in human collaboration

01:53:11.640 | moving forward and create that

01:53:12.840 | as a civilizational kind of dynamic?

01:53:15.160 | Like, can we seize this moment to do that?

01:53:17.520 | 'Cause like a lot of the other open source movements,

01:53:19.800 | it's all nerds nerding out on code for nerds.

01:53:22.040 | And this, because it's scientists,

01:53:25.900 | because it's people working on data

01:53:27.240 | that all of it faces real human problems,

01:53:30.140 | I think we have an opportunity

01:53:32.680 | to actually make a bigger impact.

01:53:34.440 | - Is there a way for this kind of open source vision

01:53:37.560 | to make money?

01:53:39.140 | - Absolutely.

01:53:40.180 | - To fund the people involved?

01:53:41.700 | Is that an essential part of it?

01:53:43.140 | - It's hard, but we're trying to do that

01:53:45.740 | in our own way at Anaconda,

01:53:48.620 | because we know that business users,

01:53:49.980 | as they use more of the stuff,

01:53:51.020 | they have needs that like business specific needs

01:53:53.060 | around security, provenance.

01:53:54.960 | They really can't tell their VPs and their investors,

01:54:00.180 | hey, we're having our data scientists

01:54:02.100 | installing random packages from who knows where

01:54:04.540 | and running on customer data.

01:54:05.660 | So they have to have someone to talk to you.

01:54:06.860 | And that's what Anaconda does.

01:54:08.380 | So we are a governed source of packages for them.

01:54:11.460 | And that's great.

01:54:12.300 | That makes some money.

01:54:13.140 | We take some of that and we just take that as a dividend.

01:54:17.140 | We take a percentage of our revenues

01:54:18.240 | and write that as a dividend for the open source community.

01:54:21.040 | But beyond that, I really see the development

01:54:24.440 | of a marketplace for people to create notebooks,

01:54:28.300 | models, data sets, curation

01:54:30.700 | of these different kinds of things,

01:54:33.100 | and to really have a long tail marketplace dynamic with that.

01:54:38.100 | - Can you speak about this problem

01:54:39.620 | that you stumbled into of package management,

01:54:42.580 | Python package management?

01:54:43.900 | What is that?

01:54:45.000 | A lot of people speak very highly of Conda,

01:54:48.980 | which is part of Anaconda, which is a package manager.

01:54:51.540 | There's a ton of packages.

01:54:53.180 | So first, what are package managers?

01:54:55.120 | And second, what was there before?

01:54:57.300 | What is pip?

01:54:58.660 | And why is Conda more awesome?

01:55:01.860 | - The package problem is this,

01:55:03.220 | which is that in order to do

01:55:05.980 | numerical computing efficiently with Python,

01:55:10.980 | there are a lot of low-level libraries

01:55:14.540 | that need to be compiled,

01:55:15.980 | compiled with a C compiler or a C++ compiler

01:55:18.380 | or a Fortran compiler.

01:55:19.900 | They need to not just be compiled,

01:55:21.140 | but they need to be compiled with all of the right settings.

01:55:23.740 | And oftentimes those settings are tuned

01:55:25.220 | for specific chip architectures.

01:55:27.460 | And when you add GPUs to the mix,

01:55:29.340 | when you look at different operating systems,

01:55:32.300 | you may be on the same chip,

01:55:33.820 | but if you're running Mac versus Linux versus Windows

01:55:37.220 | on the same x86 chip, you compile and link differently.

01:55:40.040 | All of this complexity is beyond the capability

01:55:44.820 | of most data scientists to reason about.

01:55:46.780 | And it's also beyond what most of the package developers

01:55:50.220 | want to deal with too.

01:55:51.900 | Because if you're a package developer,

01:55:52.860 | you're like, I code on Linux, this works for me, I'm good.

01:55:55.820 | It is not my problem to figure out how to build this

01:55:58.060 | on an ancient version of Windows.

01:56:00.020 | That's just simply not my problem.

01:56:01.940 | So what we end up with is we have a creator,

01:56:05.140 | or create a very creative crowdsourced environment

01:56:08.580 | where people want to use this stuff, but they can't.

01:56:11.380 | And so we ended up creating a new set of technologies

01:56:15.740 | like a build recipe system, a build system,

01:56:18.420 | and an installer system that is able to,

01:56:21.460 | well, to put it simply,

01:56:24.640 | it's able to build these packages correctly

01:56:27.700 | on each of these different kinds of platforms

01:56:29.300 | and operating systems and make it so

01:56:30.860 | when people want to install something, they can.

01:56:33.620 | It's just one command.

01:56:34.460 | They don't have to set up a big compiler system

01:56:36.940 | and do all these things.

01:56:38.300 | So when it works well, it works great.

01:56:40.420 | Now, the difficulty is we have literally thousands

01:56:43.900 | of people writing code in the ecosystem,

01:56:46.260 | building all sorts of stuff.

01:56:47.300 | And each person writing code,

01:56:48.740 | they may take a dependence on something else.

01:56:50.660 | And so all this web,

01:56:52.260 | incredibly complex web of dependencies.

01:56:54.820 | So installing the correct package

01:56:57.620 | for any given set of packages you want,

01:57:00.340 | getting that right sub graph

01:57:02.180 | is an incredibly hard problem.

01:57:04.620 | And again, most data scientists

01:57:05.780 | don't want to think about this.

01:57:06.620 | They're like, I want to install NumPy and Pandas.

01:57:09.180 | I want this version of some like geospatial library.

01:57:11.700 | I want this other thing.

01:57:13.100 | Like, why is this hard?

01:57:13.980 | These exist, right?

01:57:16.020 | And it is hard because it's, well,

01:57:17.680 | you're installing this on a version of Windows, right?

01:57:20.740 | And half of these libraries are not built for Windows.

01:57:23.400 | Or the latest version isn't available,

01:57:25.140 | but the old version was.

01:57:26.260 | If you go to the old version of this library,

01:57:27.540 | that means you need to go to a different version

01:57:28.620 | of that library.

01:57:30.060 | And so the Python ecosystem,

01:57:32.460 | by virtue of being crowdsourced,

01:57:34.500 | we were able to fill a hundred thousand different niches.

01:57:38.020 | But then we also suffer this problem

01:57:40.380 | that because it's crowdsourced and no one,

01:57:43.140 | it's like a tragedy of the commons, right?

01:57:44.500 | No one really wants to support

01:57:47.080 | their thousands of other dependencies.

01:57:49.200 | So we end up sort of having to do a lot of this.

01:57:52.160 | And of course the Conda Forge community

01:57:53.500 | also steps up as an open source community

01:57:55.060 | that maintains some of these recipes.

01:57:57.540 | That's what Conda does.

01:57:58.660 | Now, Pip is a tool that came along after Conda

01:58:01.900 | to some extent.

01:58:02.740 | It came along as an easier way

01:58:04.460 | for the Python developers writing Python code

01:58:09.460 | that didn't have as much compiled stuff.

01:58:12.920 | They could then install different packages.

01:58:15.360 | And what ended up happening in the Python ecosystem

01:58:17.820 | was that a lot of the core Python and web Python developers,

01:58:20.820 | they never ran into any of this compilation stuff at all.

01:58:24.160 | So even we have, on video,

01:58:27.020 | we have Guido van Rossum saying,

01:58:29.500 | you know what, the scientific community's packaging problems

01:58:31.620 | are just too exotic and different.

01:58:33.140 | I mean, you're talking about Fortran compilers, right?

01:58:35.700 | Like you guys just need to build

01:58:37.060 | your own solution perhaps, right?

01:58:38.960 | So the core Python community went

01:58:41.740 | and built its own sort of packaging technologies,

01:58:45.320 | not really contemplating the complexity

01:58:47.860 | of the stuff over here.

01:58:49.460 | And so now we have the challenge

01:58:50.660 | where you can Pip install some things.

01:58:52.980 | And some libraries,

01:58:53.820 | if you just want to get started with them,

01:58:55.300 | you can Pip install TensorFlow

01:58:56.620 | and that works great.

01:58:57.540 | The instant you want to also install some other packages

01:59:00.180 | that use different versions of NumPy

01:59:02.500 | or some like graphics library

01:59:04.300 | or some OpenCV thing or some other thing,

01:59:06.780 | you now run into dependency hell.

01:59:08.700 | Because you cannot, you know,

01:59:09.860 | OpenCV can have a different version of libjpeg over here

01:59:12.740 | than PyTorch over here.

01:59:14.260 | Like they actually,

01:59:15.100 | and they all have to use the,

01:59:15.920 | if you want to use GPU acceleration,

01:59:17.380 | they have to all use the same underlying drivers

01:59:18.820 | and same GPU CUDA things.

01:59:20.340 | So it gets to be very gnarly.

01:59:22.900 | And it's a level of technology

01:59:24.200 | that both the makers and the users

01:59:26.060 | don't really want to think too much about.

01:59:28.500 | - And that's where you step in

01:59:29.620 | and try to solve the sub graph problem.

01:59:32.020 | How much is that?

01:59:32.860 | I mean, you said that you don't want to think,

01:59:34.900 | they don't want to think about it,

01:59:35.900 | but how much is it a little bit on the developer

01:59:37.940 | and providing them tools

01:59:39.580 | to be a little bit more clear

01:59:42.020 | of that sub graph of dependency that's necessary?

01:59:44.860 | - It is getting to a point

01:59:46.220 | where we do have to think about,

01:59:47.820 | look, can we pull some of the most popular packages together

01:59:51.180 | and get them to work on a coordinated release timeline,

01:59:53.580 | get them to build against the same test matrix,

01:59:55.780 | et cetera, et cetera.

01:59:56.980 | And there is a little bit of dynamic around this,

01:59:58.700 | but again, it is a volunteer community.

02:00:01.780 | People working on these different projects

02:00:04.900 | have their own timelines

02:00:05.940 | and their own things they're trying to meet.

02:00:07.540 | So we end up trying to pull these things together

02:00:11.300 | and then it's this incredibly,

02:00:13.020 | and I would recommend just as a business tip,

02:00:15.380 | don't ever go into business

02:00:16.500 | where when your hard work works, you're invisible.

02:00:19.500 | And when it breaks because of someone else's problem,

02:00:21.620 | you get flack for it.

02:00:23.100 | 'Cause that's in our situation, right?

02:00:25.540 | When something doesn't condo install properly,

02:00:27.180 | usually it's some upstream issue,

02:00:28.900 | but it looks like condo is broken.

02:00:30.180 | It looks like Anaconda screwed something up.

02:00:32.540 | When things do work though, it's like,

02:00:33.980 | oh yeah, cool, it's worked.

02:00:34.980 | Assuming naturally, of course,

02:00:36.100 | that's very easy to make that work, right?

02:00:38.140 | So we end up in this kind of problematic scenario,

02:00:41.900 | but it's okay because I think we're still,

02:00:45.380 | our heart's in the right place.

02:00:46.700 | We're trying to move this forward

02:00:47.780 | as a community sort of affair.

02:00:49.180 | I think most of the people in the community

02:00:50.460 | also appreciate the work we've done over the years

02:00:52.940 | to try to move these things forward

02:00:54.220 | in a collaborative fashion.

02:00:56.340 | So.

02:00:57.260 | - One of the sub graphs of dependencies

02:01:01.140 | that became super complicated

02:01:03.500 | is the move from Python two to Python three.

02:01:05.700 | So there's all these ways to mess with these kinds

02:01:08.100 | of ecosystems of packages and so on.

02:01:11.460 | So I just wanna ask you about that particular one.

02:01:13.700 | What do you think about the move from Python two to three?

02:01:17.940 | Why did it take so long?

02:01:19.380 | What were, from your perspective,

02:01:20.860 | just seeing the packages all struggle

02:01:23.620 | and the community all struggled through this process,

02:01:26.180 | what lessons do you take away from it?

02:01:27.740 | Why did it take so long?

02:01:29.380 | - Looking back, some people perhaps underestimated

02:01:33.300 | how much adoption Python two had.

02:01:36.660 | I think some people also underestimated how much,

02:01:41.700 | or they overestimated how much value

02:01:44.380 | some of the new features in Python three really provided.

02:01:47.340 | Like the things they really loved about Python three

02:01:49.660 | just didn't matter to some of these people in Python two.

02:01:52.540 | 'Cause this change was happening as Python, SciPy

02:01:56.380 | was starting to take off really like past

02:01:58.620 | like a hockey stick of adoption

02:01:59.940 | in the early data science era, in the early 2010s.

02:02:02.820 | A lot of people were learning and onboarding

02:02:04.820 | in whatever just worked.

02:02:06.220 | And the teachers were like, well, yeah,

02:02:08.220 | these libraries I need are not supported in Python three yet,

02:02:10.300 | I'm gonna teach you Python two.

02:02:11.980 | Took a lot of advocacy to get people

02:02:13.900 | to move over to Python three.

02:02:15.660 | So I think it wasn't any particular single thing,

02:02:18.780 | but it was one of those death by a dozen cuts,

02:02:21.740 | which just really made it hard to move off of Python two.

02:02:25.420 | And also Python three itself,

02:02:27.220 | as they were kind of breaking things

02:02:28.740 | and changing these around

02:02:29.580 | and reorganize the standard library,

02:02:30.660 | there's a lot of stuff that was happening there

02:02:32.860 | that kept giving people an excuse to say,

02:02:35.780 | I'll put off till the next version.

02:02:37.700 | Two is working fine enough for me right now.

02:02:39.620 | So I think that's essentially what happened there.

02:02:41.380 | And I will say this though,

02:02:43.660 | the strength of the Python data science movement,

02:02:48.580 | I think is what kept Python alive in that transition.

02:02:52.340 | Because a lot of languages have died

02:02:53.940 | and left their user bases behind.

02:02:56.740 | If there wasn't the use of Python for data,

02:02:58.860 | there's a good chunk of Python users

02:03:01.020 | that during that transition

02:03:02.620 | would have just left for Go and Rust and stayed.

02:03:04.780 | In fact, some people did.

02:03:06.100 | They moved to Go and Rust and they just never looked back.

02:03:08.740 | The fact that we were able to grow by millions of users,

02:03:13.620 | the Python data community,

02:03:15.700 | that is what kept the momentum for Python going.

02:03:18.180 | And now the usage of Python for data is over 50%

02:03:22.100 | of the overall Python user base.

02:03:24.180 | So I'm happy to debate that on stage somewhere

02:03:27.820 | if I can with someone

02:03:28.660 | if they really wanna take issue with that statement.

02:03:30.180 | But from where I sit, I think that's true.

02:03:32.420 | - The statement there,

02:03:33.380 | the idea is that the switch from Python 2 to Python 3

02:03:37.740 | would have probably destroyed Python

02:03:40.260 | if it didn't also coincide with Python for whatever reason,

02:03:45.620 | just overtaking the data science community,

02:03:49.500 | anything that processes data.

02:03:51.620 | So like the timing was perfect,

02:03:53.940 | that this maybe imperfect decision was coupled

02:03:57.580 | with a great timing on the value of data in our world.

02:04:01.900 | - I would say the troubled execution of a good decision.

02:04:04.620 | It was a decision that was necessary.

02:04:07.220 | It's possible if we had more resources,

02:04:08.660 | we could have done in a way that was a little bit smoother,

02:04:10.940 | but ultimately, the arguments for Python 3,

02:04:15.020 | I bought them at the time and I buy them now, right?

02:04:17.340 | Having great text handling

02:04:18.860 | is like a non-negotiable table stakes thing

02:04:21.500 | you need to have in a language.

02:04:23.420 | So that's great.

02:04:26.060 | But the execution, Python is the,

02:04:30.660 | it's volunteer driven.

02:04:32.940 | It's like now the most popular language on the planet,

02:04:34.820 | but it's all literally volunteers.

02:04:37.020 | So the lack of resources meant that they had to really,

02:04:40.300 | they had to do things in a very hamstrung way.

02:04:43.540 | And I think to carry the Python momentum

02:04:46.020 | in the language through that time,

02:04:47.740 | the data movement was a critical part of that.

02:04:49.860 | - So someone was carrot and stick.

02:04:51.380 | I actually have to shamefully admit

02:04:55.660 | that it took me a very long time

02:04:56.980 | to switch from Python 2 and Python 3

02:04:58.700 | because I'm a machine learning person.

02:05:00.260 | It was just for the longest time,

02:05:01.900 | you could just do fine with Python 2.

02:05:03.740 | - Right.

02:05:04.900 | - But I think the moment where I switched

02:05:08.740 | everybody I worked with and switched myself

02:05:12.100 | for small projects and big is when finally,

02:05:15.940 | when NumPy announced that they're going to end support

02:05:19.140 | like in 2020 or something like that.

02:05:22.140 | - Right.

02:05:23.060 | - So like when I realized, oh, this is going to end.

02:05:27.340 | - Right.

02:05:28.180 | - So that's the stick.

02:05:29.020 | - That's the stick. - That's not a carrot.

02:05:30.100 | So for the longest time it was carrots.

02:05:31.660 | It was like all of these packages were saying,

02:05:34.380 | okay, we have Python 3 support now, come join us.

02:05:37.420 | We have Python 2 and Python 3.

02:05:39.060 | But when NumPy, one of the packages,

02:05:40.940 | I sort of love and depend on said like, nope, it's over.

02:05:45.940 | That's when I decided to switch.

02:05:50.340 | I wonder if you think it was possible much earlier

02:05:53.820 | for somebody like NumPy or some major package

02:05:58.820 | to step into the cold and say like, we're an anus.

02:06:01.820 | - Well, it's a chicken and egg problem too, right?

02:06:05.300 | You don't want to cut off a lot of users

02:06:07.540 | unless you see the user momentum going too.

02:06:09.380 | So the decisions for the scientific community

02:06:12.860 | for each of the different projects,

02:06:14.180 | there's not a monolith.

02:06:15.220 | Some projects are like,

02:06:16.060 | we'll only be releasing new features on Python 3.

02:06:18.860 | And that was more of a sticky carrot, right?

02:06:21.380 | A firm carrot, if you will, a firm carrot.

02:06:23.500 | A stick-shaped carrot.

02:06:27.900 | But then for others, yeah, NumPy in particular,

02:06:30.620 | 'cause it's at the base of the dependency stack

02:06:32.620 | for so many things, that was the final stick.

02:06:36.140 | That was a stick-shaped stick.

02:06:37.580 | People were saying, look,

02:06:38.420 | if I have to keep maintaining my releases for Python 2,

02:06:41.660 | that's that much less energy

02:06:43.620 | that I can put into making things better

02:06:45.700 | for the Python 3 folks or in my new version,

02:06:48.180 | which is of course going to be Python 3.

02:06:49.940 | So people were also getting kind of pulled by this tension.

02:06:53.300 | So the overall community sort of had a lot of input

02:06:56.140 | into when the NumPy core folks decided

02:06:58.460 | that they would end of life on Python 2.

02:07:01.380 | - So as these numbers are a little bit loose,

02:07:03.980 | but there are about 10 million Python programmers

02:07:06.820 | in the world, you could argue that number,

02:07:08.380 | but let's say 10 million.

02:07:10.300 | This actually, where I was looking,

02:07:12.260 | said 27 million total programmers,

02:07:15.500 | developers in the world.

02:07:17.220 | You mentioned in a talk that changes need to be made

02:07:20.540 | for there to be 100 million Python programmers.

02:07:23.980 | So first of all, do you see a future

02:07:26.020 | where there's 100 million Python programmers?

02:07:28.300 | And second, what kind of changes need to be made?

02:07:31.340 | - So Anaconda, Minicon,

02:07:32.580 | get downloaded about a million times a week.

02:07:34.900 | So I think the idea that there's only 10 million

02:07:38.100 | Python programmers in the world

02:07:39.340 | is a little bit undercounting.

02:07:41.980 | There are a lot of people who escape traditional counting

02:07:44.860 | that are using Python and data in their jobs.

02:07:48.460 | I do believe that the future world for it to,

02:07:52.100 | well, the world I would like to see

02:07:53.300 | is one where people are data literate.

02:07:56.260 | So they are able to use tools

02:07:58.620 | that let them express their questions and ideas fluidly.

02:08:03.180 | And the data variety and data complexity will not go down.

02:08:06.460 | It will only keep increasing.

02:08:08.340 | So I think some level of code or code-like things

02:08:12.460 | will continue to be relevant.

02:08:15.460 | And so my hope is that we can build systems

02:08:19.740 | that allow people to more seamlessly integrate

02:08:22.900 | Python kinds of expressivity with data systems

02:08:26.060 | and operationalization methods that are much more seamless.

02:08:29.980 | And what I mean by that is,

02:08:32.380 | right now you can't punch Python code into an Excel cell.

02:08:35.660 | I mean, there's some tools you can do to kind of do this.

02:08:37.940 | We didn't build a thing for doing this back in the day,

02:08:39.900 | but I feel like the total addressable market

02:08:43.820 | for Python users, if we do the things right,

02:08:46.860 | is on the order of the Excel users,

02:08:49.100 | which is a few hundred million.

02:08:51.180 | So I think Python has to get better at being embedded,

02:08:56.180 | being a smaller thing that pulls in

02:08:59.660 | just the right parts of the ecosystem

02:09:01.700 | to run numerics and do data exploration,

02:09:05.540 | meeting people where they're already at

02:09:07.900 | with their data and their data tools.

02:09:09.620 | And then I think also it has to be easier

02:09:12.780 | to take some of those things they've written

02:09:14.740 | and flow those back into deployed systems

02:09:17.540 | or little apps or visualizations.

02:09:19.460 | I think if we don't do those things,

02:09:20.740 | then we will always be kept in a silo

02:09:23.140 | as sort of an expert users tool

02:09:25.900 | and not a tool for the masses.

02:09:27.380 | - You know, I work with a bunch of folks

02:09:29.020 | in the Adobe Creative Suite,

02:09:32.500 | and I'm kind of forcing them or inspired them

02:09:35.420 | to learn Python to do a bunch of stuff that helps them.

02:09:38.540 | And it's interesting 'cause they probably

02:09:39.820 | wouldn't call themselves Python programmers,

02:09:41.740 | but they're all using Python.

02:09:43.820 | I would love it if the tools like Photoshop and Premiere

02:09:46.500 | and all those kinds of tools that are targeted

02:09:48.740 | towards creative people, I guess that's where Excel,

02:09:52.100 | Excel is targeted towards a certain kind of audience

02:09:54.220 | that works with data, financial people,

02:09:56.340 | all that kind of stuff.

02:09:58.100 | If there will be easy ways to leverage,

02:10:00.860 | to use Python for quick scripting tasks.

02:10:03.580 | - Yeah.

02:10:04.420 | - And there's an exciting application

02:10:06.740 | of artificial intelligence in this space

02:10:09.740 | that I'm hopeful about looking at OpenAI Codex

02:10:13.260 | with generating programs.

02:10:16.380 | - Yeah.

02:10:17.220 | - So almost helping people bridge the gap

02:10:20.780 | from kind of visual interface to generating programs

02:10:25.780 | to something formal.

02:10:27.340 | And then they can modify it and so on,

02:10:28.980 | but kind of without having to read the manual,

02:10:32.340 | without having to do a Google search and stack overflow,

02:10:34.900 | which is essentially what a neural network does

02:10:36.740 | when it's doing code generation,

02:10:39.020 | is actually generating code and allowing a human

02:10:42.820 | to communicate with multiple programs.

02:10:44.740 | And then maybe even programs to communicate

02:10:46.540 | with each other via Python.

02:10:48.020 | - Right.

02:10:48.860 | - So that to me is a really exciting possibility

02:10:51.220 | 'cause I think there's a friction to kind of,

02:10:55.980 | like how do I learn how to use Python in my life?

02:10:58.940 | There's oftentimes you kind of,

02:11:01.820 | what started a class, you start learning about types.

02:11:05.380 | I don't know, functions.

02:11:07.140 | Like this is, you know, Python is the first language

02:11:09.620 | with which you start to learn to program.

02:11:11.980 | But I feel like that's going to take a long time

02:11:16.700 | for you to understand why it's useful.

02:11:18.580 | You almost want to start with a script.

02:11:20.340 | - Well, you do.

02:11:21.580 | In fact, I think starting with the theory

02:11:23.860 | behind programming languages and types and all that,

02:11:25.580 | I mean, types are there to make

02:11:27.780 | the compiler writer's jobs easier.

02:11:30.020 | Types are not, I mean, heck,

02:11:32.140 | do you have an ontology of types

02:11:33.180 | or just the objects on this table?

02:11:34.380 | No. (laughs)

02:11:35.580 | So types are there because compiler writers are human

02:11:39.180 | and they're limited in what they can do.

02:11:40.860 | But I think that the beauty of scripting,

02:11:45.140 | like there's a Python book

02:11:46.860 | that's called "Automate the Boring Stuff,"

02:11:49.260 | which is exactly the right mentality.

02:11:51.180 | I grew up with computers in a time

02:11:53.900 | when I could, when Steve Jobs was still pitching

02:11:58.500 | these things as bicycles for the mind.

02:11:59.900 | They were supposed to not be just media consumption devices,

02:12:03.300 | but they were actually, you could write some code.

02:12:05.940 | You could write basic,

02:12:06.780 | you could write some stuff to do some things.

02:12:09.020 | And that feeling of a computer as a thing

02:12:12.180 | that we can use to extend ourselves

02:12:14.260 | has all but evaporated for a lot of people.

02:12:17.780 | So you see a little bit in parts

02:12:19.500 | in the generation of youth around Minecraft or Roblox, right?

02:12:23.580 | And I think Python, CircuitPython,

02:12:25.060 | these things could be a renaissance of that,

02:12:28.780 | of people actually shaping and using their computers

02:12:32.620 | as computers, as an extension of their minds

02:12:35.460 | and their curiosity, their creativity.

02:12:37.860 | So you talk about scripting the Adobe suite with Python

02:12:41.500 | in the 3D graphics world.

02:12:42.860 | Python is a scripting language

02:12:46.300 | that some of these 3D graphics suites use.

02:12:48.700 | And I think that's great.

02:12:49.540 | We should better support those kinds of things.

02:12:51.260 | But ultimately the idea that I should be able

02:12:54.220 | to have power over my computing environment.

02:12:56.260 | If I want these things to happen repeatedly all the time,

02:12:59.660 | I should be able to say that somehow to the computer, right?

02:13:02.660 | Now, whether the operating systems get there faster

02:13:06.540 | by having some Siri backed with open AI, with whatever.

02:13:09.580 | So you can just say, "Siri, make this do this,

02:13:10.860 | do this on every other Friday," right?

02:13:12.660 | We probably will get there somewhere.

02:13:14.220 | And Apple's always had these ideas.

02:13:15.900 | There's the Apple script in the menu that no one ever uses,

02:13:19.180 | but you can do these kinds of things.

02:13:21.660 | But when you start doing that kind of scripting,

02:13:23.700 | the challenge isn't learning the type system

02:13:25.900 | or even the syntax of the language.

02:13:27.180 | The challenge is all of the dictionaries

02:13:29.260 | and all the objects of all their properties

02:13:31.140 | and attributes and parameters.

02:13:32.860 | Like who's got time to learn all that stuff, right?

02:13:35.420 | So that's when then programming by prototype

02:13:38.860 | or by example becomes the right way

02:13:40.980 | to get the user to express their desire.

02:13:43.780 | So there's a lot of these different ways

02:13:45.220 | that we can approach programming.

02:13:46.060 | But I do think when, as you were talking

02:13:48.260 | about the Adobe scripting thing,

02:13:49.780 | I was thinking about, you know,

02:13:51.340 | when we do use something like NumPy,

02:13:53.780 | when we use things in the Python data

02:13:55.780 | and scientific, let's say, expression system,

02:14:00.140 | there's a reason we use that,

02:14:01.340 | which is that it gives us mathematical precision.

02:14:04.380 | It gives us actually quite a lot of precision

02:14:06.780 | over precisely what we mean about this data set,

02:14:09.940 | that data set.

02:14:10.780 | And it's the fact that we can have that precision

02:14:13.700 | that lets Python be powerful over as a duct tape for data.

02:14:18.700 | You know, you give me a TSV or a CSV,

02:14:21.220 | and if you give me some massively expensive vendor tool

02:14:25.140 | for data transformation,

02:14:26.460 | I don't know I'm gonna be able to solve your problem.

02:14:28.300 | But if you give me a Python prompt,

02:14:30.620 | you can throw whatever data you want at me.

02:14:32.500 | I will be able to mash it into shape.

02:14:34.420 | So that ability to take it as sort of this like,

02:14:38.380 | you know, machete out into the data jungle

02:14:40.340 | is really powerful.

02:14:41.620 | And I think that's why at some level,

02:14:44.340 | we're not gonna get away from some of these expressions

02:14:47.780 | and APIs and libraries in Python for data transformation.

02:14:51.700 | - You've been at the center of the Python community

02:14:57.300 | for many years.

02:14:58.420 | If you could change one thing about the community

02:15:03.420 | to help it grow, to help it improve,

02:15:05.580 | to help it flourish and prosper, what would it be?

02:15:09.500 | I mean, you know, it doesn't have to be one thing,

02:15:11.740 | but what kind of comes to mind?

02:15:13.900 | What are the challenges?

02:15:15.420 | - Humility is one of the values

02:15:16.540 | that we have at Anaconda, the company,

02:15:17.980 | but it's also one of the values in the community

02:15:21.380 | that it's been breached a little bit in the last few years,

02:15:24.540 | but in general, people are quite decent

02:15:27.420 | and reasonable and nice.

02:15:29.380 | And that humility prevents them from seeing

02:15:32.660 | the greatness that they could have.

02:15:37.940 | I don't know how many people in the core Python community

02:15:41.780 | really understand that they stand perched at the edge

02:15:46.780 | of an opportunity to transform how people use computers.

02:15:50.340 | And actually PyCon, I think it was the last physical PyCon

02:15:53.900 | I went to, Russell Keith-Magee gave a great keynote

02:15:57.740 | about very much along the lines of the challenges I have,

02:16:01.620 | which is Python for a language that doesn't actually,

02:16:05.380 | that can't put an interface up

02:16:07.140 | on like the most popular computing devices.

02:16:09.380 | It's done really well as a language, hasn't it?

02:16:11.820 | You can't write a web front end with Python, really.

02:16:13.740 | I mean, everyone uses JavaScript.

02:16:15.060 | You certainly can't write native apps.

02:16:17.020 | So for a language that you can't actually write apps

02:16:20.580 | in any of the front end runtime environments,

02:16:22.700 | Python's done exceedingly well.

02:16:24.340 | And so that wasn't to pat ourselves on the back.

02:16:28.740 | That was to challenge ourselves as a community to say,

02:16:30.580 | we, through our current volunteer dynamic,

02:16:32.500 | have gotten to this point.

02:16:34.460 | What comes next and how do we seize,

02:16:36.740 | you know, we've caught the tiger by the tail.

02:16:38.620 | How do we make sure we keep up with it as it goes forward?

02:16:40.980 | - So that's one of the questions I have

02:16:42.420 | about sort of open source communities is,

02:16:44.660 | at its best, there's a kind of humility.

02:16:48.500 | Is that humility prevent you to have a vision

02:16:52.500 | for creating something like very new and powerful?

02:16:55.620 | - And you've brought us back to consciousness again.

02:16:57.660 | The collaboration is a swarm emergent dynamic.

02:17:00.700 | Humility lets these people work together

02:17:02.620 | without anyone trouncing anyone else.

02:17:04.820 | How do they, you know, in consciousness,

02:17:07.180 | there's the question of the binding problem.

02:17:08.660 | How does a singular, our attention,

02:17:10.700 | how does that emerge from, you know, billions of neurons?

02:17:13.860 | So how can you have a swarm of people emerge a consensus

02:17:17.660 | that has a singular vision to say, we will do this.

02:17:20.620 | And most importantly, we're not gonna do these things.

02:17:23.860 | Emerging a coherent, pointed, focused leadership dynamic

02:17:28.860 | from a collaboration, being able to do that kind of,

02:17:32.060 | and then dissolve it so people can still do the swarm thing.

02:17:35.060 | That's a problem, that's a question.

02:17:37.220 | - So do you have to have a charismatic leader?

02:17:39.580 | For some reason, Linus Torvald comes to mind,

02:17:42.620 | but you know, there's people who criticize--

02:17:44.460 | - He rules with an iron fist, man.

02:17:46.620 | - But there's still charisma to it.

02:17:48.660 | - There is charisma, right?

02:17:49.500 | - There's a charisma to that iron fist.

02:17:51.780 | There's, every leader is different, I would say,

02:17:55.300 | in their success.

02:17:56.700 | So he doesn't, I don't even know if you can say

02:17:59.460 | he doesn't have humility.

02:18:01.860 | There's such a meritocracy of ideas that like,

02:18:06.620 | this is a good idea and this is a bad idea.

02:18:10.380 | - There's a step function to it.

02:18:11.580 | Once you clear a threshold, he's open.

02:18:13.860 | Once you clear the Bozo threshold,

02:18:15.580 | he's open to your ideas, I think.

02:18:17.500 | - But see, the interesting thing is,

02:18:19.500 | obviously that will not stand in an open source community

02:18:23.380 | if that threshold that is defined

02:18:25.860 | by that one particular person is not actually that good.

02:18:30.260 | So you actually have to be really excellent at what you do.

02:18:33.700 | So he's very good at what he does.

02:18:37.100 | And so there's some aspect of leadership

02:18:39.020 | where you can get thrown out, people can just leave.

02:18:42.620 | You know, that's how it works with open source, the fork.

02:18:45.940 | But at the same time, you want to sometimes be a leader,

02:18:49.900 | like with a strong opinion, 'cause people,

02:18:52.420 | I mean, there's some kind of balance here

02:18:54.420 | for this like hive mind to get like behind.

02:18:57.340 | - Leadership is a big topic.

02:18:58.500 | And I didn't, you know, I'm not one of these guys

02:18:59.860 | that went to MBA school and said,

02:19:01.020 | "I'm gonna be an entrepreneur and I'm gonna be a leader.

02:19:03.500 | "And I'm gonna read all these Harvard Business Review

02:19:05.220 | "articles on leadership and all this other stuff."

02:19:07.740 | Like I was a physicist turned into a software nerd

02:19:10.700 | who then really like nerded out on Python.

02:19:12.780 | And now I am entrepreneurial, right?

02:19:14.860 | I saw a business opportunity around the use of Python

02:19:16.620 | or data, but for me, what has been interesting

02:19:19.980 | over this journey with the last 10 years

02:19:22.140 | is how much I started really enjoying the understanding

02:19:27.140 | and thinking deeper about organizational dynamics

02:19:29.620 | and leadership.

02:19:30.460 | And leadership does come down to a few core things.

02:19:35.300 | Number one, a leader has to create belief

02:19:40.020 | or at least has to dispel disbelief.

02:19:42.300 | Leadership also, you have to have vision,

02:19:46.500 | loyalty and experience.

02:19:49.380 | - So can you say belief in a singular vision?

02:19:52.620 | Like what does belief mean?

02:19:53.700 | - Yeah, belief means a few things.

02:19:55.240 | Belief means here's what we need to do.

02:19:56.980 | And this is a valid thing to do and we can do it.

02:20:00.020 | That you have to be able to drive that belief.

02:20:05.460 | And every step of leadership along the way

02:20:08.780 | has to help you amplify that belief to more people.

02:20:12.660 | I mean, I think at a fundamental level, that's what it is.

02:20:15.900 | You have to have a vision.

02:20:17.100 | You have to be able to show people that,

02:20:20.900 | or you have to convince people to believe in the vision

02:20:23.620 | and to get behind you.

02:20:25.220 | And that's where the loyalty part comes in

02:20:26.500 | and the experience part comes in.

02:20:28.180 | - There's all different flavors of leadership.

02:20:30.180 | So if we talk about Linus,

02:20:31.420 | we could talk about Elon Musk and Steve Jobs.

02:20:36.420 | There's Sunder Pichai.

02:20:38.420 | There's people that kind of put themselves at the center

02:20:40.740 | and are strongly opinionated.

02:20:42.540 | And some people are more like consensus builders.

02:20:46.020 | What works well for open source?

02:20:47.740 | What works well in the space of programmers?

02:20:49.740 | So you've been a programmer, you've led many programmers

02:20:53.300 | and are now sort of at the center of this ecosystem.

02:20:55.620 | What works well in the programming world, would you say?

02:20:58.980 | - It really depends on the people.

02:21:01.140 | What style of leadership is best.

02:21:02.620 | And it depends on the programming community.

02:21:04.180 | I think for the Python community,

02:21:06.340 | servant leadership is one of the values.

02:21:08.300 | Like at the end of the day,

02:21:09.140 | the leader has to also be the high priest of values, right?

02:21:14.140 | So any kind of, any collection of people

02:21:18.020 | has values of their living.

02:21:19.980 | And if you want to maintain certain values

02:21:24.540 | and those values help you as an organization

02:21:27.100 | become more powerful,

02:21:28.340 | then the leader has to live those values unequivocally

02:21:31.340 | and has to hold the values.

02:21:34.380 | So in our case, in this collaborative community

02:21:37.540 | around Python, I think that the humility

02:21:41.940 | is one of those values, servant leadership.

02:21:43.820 | You actually have to kind of do the stuff.

02:21:45.300 | You have to walk the walk, not just talk the talk.

02:21:47.820 | I don't feel like the Python community

02:21:51.140 | really demands that much from a vision standpoint.

02:21:53.820 | - And they should.

02:21:54.780 | - And I think they should.

02:21:56.740 | - This is the interesting thing is

02:21:59.060 | like so many people use Python from where it comes

02:22:03.420 | the vision, you know, like you have a Elon Musk

02:22:06.860 | type character who has makes bold statements

02:22:10.820 | about the vision for particular companies

02:22:13.300 | he's involved with.

02:22:14.660 | And it's like, I think a lot of people that work

02:22:18.700 | at those companies kind of can only last

02:22:22.580 | if they believe that vision.

02:22:24.460 | Because in some of it is super bold.

02:22:26.260 | So my question is, and by the way,

02:22:28.620 | those companies often use Python.

02:22:30.300 | How do you establish a vision?

02:22:33.780 | Like get to a hundred million users, right?

02:22:36.500 | Get to where, you know, Python is at the center

02:22:42.300 | of the machine learning and was it data science,

02:22:46.780 | machine learning, deep learning,

02:22:48.460 | artificial intelligence revolution, right?

02:22:51.720 | Like in many ways, perhaps the Python community

02:22:54.900 | is not thinking of it that way,

02:22:55.860 | but it's leading the way on this.

02:22:58.180 | Like the tooling is like essential.

02:23:01.260 | - Right, well, you know, for a while,

02:23:03.340 | PyCon people in the scientific Python

02:23:05.860 | and the PyData community, they would submit talks.

02:23:09.300 | Those are early 2010s, mid 2010s.

02:23:12.180 | They would submit talks for PyCon

02:23:14.100 | and the talks would all be rejected

02:23:15.740 | because there was the separate sort of PyData conferences.

02:23:18.700 | And they're like, well, these should,

02:23:19.540 | these probably belong more to PyData.

02:23:21.180 | And instead there'd be yet another talk about,

02:23:23.120 | you know, threads and you know, whatever,

02:23:25.200 | some web framework.

02:23:26.400 | And it's like, that was an interesting dynamic

02:23:29.000 | to see that there was, I mean,

02:23:31.320 | at the time it was a little annoying

02:23:32.680 | because we wanted to try to get more users

02:23:34.280 | and get more people talking about these things.

02:23:35.760 | And PyCon is a huge venue, right?

02:23:37.320 | It's thousands of Python programmers,

02:23:40.220 | but then also came to appreciate that, you know,

02:23:42.200 | parallel, having an ecosystem

02:23:44.480 | that allows parallel innovation is not bad, right?

02:23:47.400 | There are people doing embedded Python stuff.

02:23:49.040 | There's people doing web programming,

02:23:50.640 | people doing scripting, there's cyber users of Python.

02:23:53.260 | I think the, ultimately at some point,

02:23:55.100 | if your slide mode mold covers so much stuff,

02:23:58.020 | you have to respect that different things are growing

02:24:00.140 | in different areas and different niches.

02:24:02.300 | Now, at some point that has to come together

02:24:04.140 | and the central body has to provide resources.

02:24:07.540 | The principle here is subsidiarity.

02:24:09.140 | Give resources to the various groups

02:24:11.740 | to then allocate as they see fit in their niches.

02:24:15.020 | That would be a really helpful dynamic.

02:24:16.340 | But again, it's a volunteer community.

02:24:17.500 | It's not like they had that many resources to start with.

02:24:21.200 | - What was or is your favorite programming setup?

02:24:23.960 | What operating system, what keyboard, how many screens?

02:24:27.280 | What are you listening to?

02:24:29.120 | What time of day?

02:24:30.920 | Are you drinking coffee, tea?

02:24:33.000 | - Tea, sometimes coffee, depending on how well I slept.

02:24:36.280 | I used to have-

02:24:37.120 | - How deep do you get?

02:24:38.120 | Are you a night owl?

02:24:39.680 | I remember somebody asked you somewhere

02:24:41.560 | a question about work-life balance

02:24:43.760 | and like not just work-life balance,

02:24:46.000 | but like a family, you know, you lead a company

02:24:49.000 | and your answer was basically like,

02:24:52.060 | I still haven't figured it out.

02:24:54.240 | - Yeah, I think I've gotten a little bit better balance.

02:24:56.320 | I have a really great leadership team now supporting me.

02:24:58.880 | And so that takes a lot of the day-to-day stuff

02:25:01.680 | off my plate and my kids are getting a little older.

02:25:04.360 | So that helps.

02:25:05.200 | So, and of course I have a wonderful wife

02:25:07.500 | who takes care of a lot of the things

02:25:09.200 | that I'm not able to take care of and she's great.

02:25:11.660 | I try to get to sleep earlier now

02:25:13.700 | because I have to get up every morning at six

02:25:15.120 | to take my kid down to the bus stop.

02:25:16.980 | So there's a hard thing.

02:25:18.940 | - Yeah.

02:25:19.780 | - For a while I was doing polyphasic sleep,

02:25:21.080 | which is really interesting.

02:25:22.000 | Like I go to bed at nine, wake up at like 2 a.m.,

02:25:24.600 | work till five, sleep three hours, wake up at eight.

02:25:27.260 | Like that was actually, it was interesting.

02:25:29.320 | It wasn't too bad.

02:25:30.160 | - How did it feel?

02:25:31.120 | - It was good.

02:25:31.960 | I didn't keep it up for years, but once I have travel,

02:25:34.960 | then it just, everything goes out the window, right?

02:25:37.360 | 'Cause then you're like time zones and all these things.

02:25:39.080 | - Socially was it, except like,

02:25:40.880 | were you able to live outside of how you felt?

02:25:43.480 | - Yes.

02:25:44.320 | - Were you able to live normal society?

02:25:45.720 | - Oh yeah, because like on the nights

02:25:47.320 | that wasn't out hanging out with people or whatever,

02:25:48.920 | going to bed at nine, no one cares.

02:25:50.680 | I wake up at two, I'm still responding to their slacks,

02:25:52.680 | emails, whatever.

02:25:54.160 | And shit posting on Twitter or whatever

02:25:57.240 | at two in the morning is great.

02:25:58.880 | - Exactly.

02:25:59.720 | - Right, and then you go to bed for a few hours

02:26:01.720 | and you wake up, it's like you had an extra day

02:26:03.360 | in the middle.

02:26:04.200 | - Yes.

02:26:05.020 | - And I'd read somewhere that humans naturally

02:26:06.320 | have biphasic sleep or something, I don't know.

02:26:08.520 | But-

02:26:09.360 | - I read basically everything somewhere.

02:26:11.160 | So every option of everything is available.

02:26:13.400 | - Every option of everything.

02:26:14.520 | I will say that that worked out for me for a while,

02:26:16.840 | although I don't do it anymore.

02:26:18.480 | In terms of programming setup,

02:26:19.380 | I had a 27 inch high DPI setup that I really liked,

02:26:24.380 | but then I moved to a curved monitor

02:26:26.200 | just because when I moved to the new house,

02:26:28.840 | I want to have a bit more screen for Zoom

02:26:31.280 | plus communications, plus various kinds of things.

02:26:33.840 | - So it's like one large monitor.

02:26:35.800 | - One large curved monitor.

02:26:37.160 | - What operating system?

02:26:39.760 | - Mac.

02:26:40.640 | - Okay.

02:26:41.480 | - Yeah.

02:26:42.300 | - Is that what happens when you become important?

02:26:43.920 | Is you stop using Linux and Windows?

02:26:46.700 | - No, I actually have a Windows box as well

02:26:48.520 | on the next table over, but I have three desks, right?

02:26:53.520 | - Yes.

02:26:54.960 | - So main one is the standing desk so that I can,

02:26:56.880 | you know, whatever, what I'm like,

02:26:58.360 | I have a teleprompter set up and everything else.

02:26:59.920 | And then I've got my iMac and then eGPU and then Windows PC.

02:27:04.920 | The reason I moved to Mac was it's got a Linux prompt

02:27:10.680 | or no, sorry, it's got a Unix prompt.

02:27:12.160 | It's got a Unix prompt so I can do all my stuff.

02:27:14.260 | But then I don't have to worry,

02:27:18.300 | like when I'm presenting for clients or investors, whatever,

02:27:20.920 | like I don't have to worry about any like ACPI related

02:27:25.120 | F-sick things in the middle of a presentation,

02:27:27.160 | like none of that.

02:27:28.000 | It just, it will always wake from sleep

02:27:29.840 | and it won't kernel panic on me.

02:27:32.300 | And this is not a dig against Linux,

02:27:34.200 | except that I just, I feel really bad.

02:27:38.440 | I feel like a traitor to my community saying this, right?

02:27:40.180 | But in 2012, I was just like, okay,

02:27:42.480 | started my own company, what do I get?

02:27:44.040 | And Linux laptops were just not quite there.

02:27:47.520 | And so I've just stuck with Macs.

02:27:48.360 | - Can I just defend something that nobody respectable

02:27:51.360 | seems to do, which is, so I do a boot on Linux Windows,

02:27:55.800 | but in Windows, I have a Windows subsystems for Linux

02:28:00.400 | or whatever, WSL.

02:28:02.940 | And I find myself being able to handle everything I need

02:28:06.500 | and almost everything I need in Linux for basic sort of tasks,

02:28:09.680 | scripting tasks within WSL

02:28:11.280 | and it creates a really nice environment.

02:28:12.960 | So I've been, but like whenever I hang out

02:28:14.960 | with like especially important people,

02:28:17.000 | like they're all on iPhone and a Mac and it's like, yeah.

02:28:22.800 | Like what, there is a messiness to Windows

02:28:26.000 | and a messiness to Linux that makes me feel

02:28:29.960 | like you're still in it.

02:28:31.900 | - Well, the Linux stuff,

02:28:33.600 | Windows subsystem for Linux is very tempting,

02:28:35.720 | but there's still the Windows on the outside

02:28:38.960 | where I don't know where, and I've been, okay,

02:28:41.000 | I've used DOS since version 1.11 or 1.21 or something.

02:28:45.520 | So I've been a long time Microsoft user.

02:28:48.320 | And I will say that like, it's really hard for me to know

02:28:52.640 | where anything is, how to get to the details

02:28:54.400 | behind something when something screws up

02:28:55.800 | as an invariably does.

02:28:57.560 | And just things like changing group permissions

02:28:59.880 | on some shared folders and stuff,

02:29:01.360 | just everything seems a little bit more awkward,

02:29:03.560 | more clicks than it needs to be.

02:29:05.200 | Not to say that there aren't weird things

02:29:07.760 | like hidden attributes and all this other happy stuff

02:29:09.840 | on Mac, but for the most part,

02:29:14.000 | and well, actually, especially now

02:29:15.360 | with the new hardware coming out on Mac,

02:29:16.800 | that'll be very interesting, with the new M1.

02:29:19.040 | There were some dark years, the last few years

02:29:21.600 | when I was like, I think maybe I have to move off of Mac

02:29:23.960 | as a platform.

02:29:24.800 | - Dark years.

02:29:25.640 | - But this, I mean, like my keyboard was just not working.

02:29:29.040 | Like literally my keyboard just wasn't working, right?

02:29:31.160 | I had this touch bar, didn't have a physical escape button

02:29:33.480 | like I needed to, 'cause I used Vim.

02:29:35.120 | And now I think we're back, so.

02:29:37.400 | - So you use Vim and you have a, what kind of keyboard?

02:29:40.320 | - So I use a Realforce 87U.

02:29:43.000 | It's a mechanical, it's a topro key switch.

02:29:45.240 | - Is it weird shape, or is it normal shape?

02:29:47.000 | - It's a normal shape.

02:29:48.480 | - Oh no, 'cause I say that because I use a Kinesis

02:29:51.440 | and I had, you said some dark, you said you had dark moments.

02:29:54.960 | I've recently had a dark moment,

02:29:57.400 | it's like, what am I doing with my life?

02:29:58.760 | So I remember sort of flying in a very kind of tight space

02:30:02.960 | and as I'm working, this is what I do on an airplane.

02:30:06.560 | I pull out a laptop and on top of the laptop,

02:30:09.600 | I'll put a Kinesis keyboard.

02:30:11.080 | - That's hardcore, man.

02:30:12.040 | - I was thinking, is this who I am?

02:30:13.800 | Is this what I'm becoming?

02:30:15.080 | Will I be this person?

02:30:16.160 | 'Cause I'm on Emacs with this Kinesis keyboard

02:30:18.520 | sitting like with everybody around.

02:30:21.920 | - Emacs on Windows.

02:30:23.200 | - WSL, yeah.

02:30:25.600 | - Yeah, Emacs on Linux on Windows.

02:30:27.360 | - Yeah, on Windows.

02:30:28.720 | And like everybody around me is using their iPhone

02:30:32.200 | to look at TikTok.

02:30:33.120 | So I'm like in this land and I thought,

02:30:36.080 | you know what, maybe I need to become an adult

02:30:38.440 | and put the 90s behind me and use like a normal keyboard.

02:30:43.200 | And then I did some soul searching

02:30:45.000 | and I decided like, this is who I am.

02:30:46.760 | This is me like coming out of the closet

02:30:48.520 | to saying I'm Kinesis keyboard all the way.

02:30:50.760 | I'm going to use Emacs.

02:30:52.880 | - You know who else is a Kinesis fan?

02:30:55.160 | Wes McKinney, the creator of Pandas.

02:30:56.720 | - Oh.

02:30:57.560 | - He just, he banged out Pandas

02:30:59.000 | on a Kinesis keyboard, I believe.

02:31:00.000 | I don't know if he's still using one, maybe,

02:31:01.920 | but certainly 10 years ago, like he was.

02:31:03.960 | - If anyone's out there,

02:31:05.320 | maybe we need to have a Kinesis support group.

02:31:07.520 | Please reach out.

02:31:08.360 | - Isn't there already one?

02:31:09.440 | - Is there one?

02:31:10.280 | - I don't know, there's gotta be an IRC channel, man.

02:31:12.200 | (laughing)

02:31:14.360 | - Oh no, and you access it through Emacs.

02:31:16.880 | Okay, do you still program these days?

02:31:19.600 | - I do a little bit.

02:31:20.600 | Honestly, the last thing I did was I had written,

02:31:25.880 | I was working with my son to script some Minecraft stuff.

02:31:28.520 | So I was doing a little bit of that.

02:31:29.560 | That was the last, literally the last code I wrote.

02:31:32.320 | Maybe I did, oh, you know what?

02:31:33.440 | Also I wrote some code to do some cap table evaluation,

02:31:36.600 | waterfall modeling kind of stuff.

02:31:38.240 | - What advice would you give to a young person,

02:31:41.320 | say your son, today in high school,

02:31:44.160 | maybe even college about career, about life?

02:31:47.280 | - This may be where I get into trouble a little bit.

02:31:51.080 | We are coming to the end.

02:31:53.360 | We're rapidly entering a time between worlds.

02:31:56.480 | So we have a world now that's starting to really crumble

02:31:59.880 | under the weight of aging institutions

02:32:01.880 | that no longer even pretend to serve the purposes

02:32:04.440 | they were created for.

02:32:05.720 | We are creating technologies that are hurtling

02:32:08.600 | billions of people headlong into philosophical crises

02:32:11.440 | who they don't even know the philosophical operating systems

02:32:14.240 | in their firmware.

02:32:15.080 | And they're heading into a time when that gets vaporized.

02:32:17.800 | So for people in high school,

02:32:20.040 | and certainly I tell my son this as well,

02:32:21.480 | he's in middle school, people in college,

02:32:24.880 | you are going to have to find your own way.

02:32:29.680 | You're going to have to have a pioneer spirit,

02:32:31.680 | even if you live in the middle

02:32:34.040 | of the most dense urban environment.

02:32:36.320 | All of human reality around you is the result

02:32:42.600 | of the last few generations of humans

02:32:45.600 | agreeing to play certain kinds of games.

02:32:48.480 | A lot of those games no longer operate

02:32:53.280 | according to the rules they used to.

02:32:55.080 | Collapse is non-linear, but it will be managed.

02:32:59.240 | And so if you are in a particular social caste

02:33:03.320 | or economic caste, and it's not,

02:33:07.280 | I think it's not kosher to say that about America,

02:33:10.200 | but America is a very stratified and classist society.

02:33:14.160 | There's some mobility, but it's really quite classist.

02:33:17.080 | And in America, unless you're in the upper middle class,

02:33:20.800 | you are headed into very choppy waters.

02:33:23.520 | So it is really, really good to think and understand

02:33:26.960 | the fundamentals of what you need

02:33:29.240 | to build a meaningful life for you,

02:33:32.080 | your loved ones, with your family.

02:33:33.880 | And almost all of the technology being created

02:33:38.000 | that's consumer-facing is designed to own people,

02:33:41.760 | to take the four stack of people, to delaminate them,

02:33:46.760 | and to own certain portions of that stack.

02:33:50.360 | And so if you want to be an integral human being,

02:33:52.280 | if you want to have your agency

02:33:54.400 | and you want to find your own way in the world,

02:33:57.640 | when you're young would be a great time to spend time

02:34:00.600 | looking at some of the classics around

02:34:03.600 | what it means to live a good life,

02:34:05.880 | what it means to build connection with people.

02:34:08.080 | And so much of the status game, so much of the stuff,

02:34:12.120 | one of the things that I sort of talk about

02:34:14.880 | as we create more and more technology,

02:34:18.640 | there's a gradient of technology,

02:34:19.840 | and a gradient of technology

02:34:20.840 | always leads to a gradient of power.

02:34:22.640 | And this is Jacques Allure's point to some extent as well.

02:34:25.200 | That gradient of power is not going to go away.

02:34:27.280 | The technologies are going so fast

02:34:29.880 | that even people like me who helped create

02:34:32.040 | some of the stuff, I'm being left behind.

02:34:33.600 | That's some cutting edge research.

02:34:34.680 | I don't know what's going on in GANs today.

02:34:36.800 | You know, I'll go read some proceedings.

02:34:38.640 | So as the world gets more and more technological,

02:34:42.600 | it'll create more and more gradients

02:34:44.240 | where people will seize power, economic fortunes,

02:34:48.400 | and the way they make the people who are left behind

02:34:51.320 | okay with their lot in life is they create lottery systems.

02:34:54.480 | They make you take part in the narrative

02:35:00.000 | of your own being trapped in your own economic sort of zone.

02:35:04.200 | So avoiding those kinds of things is really important,

02:35:07.920 | knowing when someone is running game on you basically.

02:35:10.740 | So these are the things I would tell young people.

02:35:12.280 | It's a dark message, but it's realism.

02:35:14.240 | I mean, it's what I see.

02:35:15.880 | - So after you gave some realism, you sit back,

02:35:18.520 | you sit back with your son,

02:35:19.760 | you're looking out at the sunset.

02:35:21.680 | What to him can you give as words of hope?

02:35:26.280 | And to you, from where do you derive hope

02:35:30.420 | for the future of our world?

02:35:32.120 | So you said at the individual level,

02:35:33.620 | you have to have a pioneer mindset

02:35:36.480 | to go back to the classics to understand

02:35:38.760 | what is in human nature you can find meaning.

02:35:41.600 | But at the societal level, what trajectory,

02:35:44.520 | when you look at possible trajectories,

02:35:46.000 | what gives you hope?

02:35:47.480 | - What gives me hope is that we have little tremors now,

02:35:52.480 | shaking people out of the reverie

02:35:54.640 | of the fiction of modernity that they've been living in,

02:35:57.000 | kind of a late 20th century style modernity.

02:35:59.500 | That's good, I think, because,

02:36:03.660 | and also to your point earlier,

02:36:06.400 | people are burning out on some of the social media stuff.

02:36:08.200 | They're sort of seeing the ugly side,

02:36:09.280 | especially the latest news with Facebook

02:36:11.640 | and the whistleblower, right?

02:36:12.840 | It's quite clear these things are not

02:36:15.720 | all they're cracked up to be.

02:36:16.720 | - So do you believe,

02:36:18.400 | I believe better social media can be built

02:36:20.760 | because they are burning out

02:36:21.840 | and it'll incentivize other competitors to be built.

02:36:25.280 | Do you think that's possible?

02:36:26.480 | - Well, the thing about it is that

02:36:29.040 | when you have extractive return on returns,

02:36:32.160 | capital coming in and saying, look, you own a network,

02:36:36.640 | give me some exponential dynamics out of this network.

02:36:39.000 | What are you gonna do?

02:36:39.840 | You're gonna just basically put a toll keeper

02:36:41.460 | at every single node and every single graph edge,

02:36:45.240 | every node, every vertex, every edge.

02:36:47.140 | But if you don't have that need for it,

02:36:49.880 | if no one's sitting there saying,

02:36:51.200 | hey, Wikipedia monetize every character,

02:36:53.240 | every byte, every phrase,

02:36:54.920 | then generative human dynamics

02:36:56.640 | will naturally sort of arise,

02:36:58.280 | assuming we do, we respect a few principles

02:37:01.120 | around online communications.

02:37:03.000 | So the greatest and biggest social network in the world

02:37:05.760 | is still like email, SMS, right?

02:37:08.720 | So we're fine there.

02:37:10.560 | The issue with the social media, as we call it now,

02:37:13.200 | is they're actually just new amplification systems, right?

02:37:16.600 | Now it's benefit of certain people like yourself

02:37:18.640 | who have interesting content to be amplified.

02:37:23.120 | So it's created a greater economy and that's cool.

02:37:25.160 | There's a lot of great content out there,

02:37:26.800 | but giving everyone a shot at the fame lottery,

02:37:29.280 | saying, hey, you could also have your,

02:37:31.480 | if you wiggle your butt the right way on TikTok,

02:37:33.960 | you can have your 15 seconds of micro fame.

02:37:36.100 | That's not healthy for society at large.

02:37:38.180 | So I think if we can create tools

02:37:40.200 | that help people be conscientious about their attention,

02:37:45.160 | spend time looking at the past

02:37:46.680 | and really retrieving memory and calling,

02:37:49.600 | not calling, but processing and thinking about that,

02:37:53.240 | I think that's certainly possible

02:37:55.120 | and hopefully that's what we get.

02:37:57.280 | So the bigger picture, the bigger question

02:38:00.640 | that you're asking about what gives me hope

02:38:02.360 | is that these early shocks of COVID lockdowns

02:38:08.120 | and remote work and all these different kinds of things,

02:38:11.600 | I think it's getting people to a point

02:38:14.000 | where they're sort of no longer in the reverie, right?

02:38:19.000 | As my friend, Jim Rutt says,

02:38:21.040 | there's more people with ears to hear now, right?

02:38:23.680 | With pandemic and education, everyone's like, wait, wait,

02:38:27.160 | what have you guys been doing with my kids?

02:38:28.660 | Like, how are you teaching them?

02:38:30.120 | What is this crap you're giving them as homework, right?

02:38:32.400 | So I think these are the kinds of things

02:38:33.960 | that are getting in the supply chain disruptions,

02:38:36.860 | getting more people to think about

02:38:38.240 | how do we actually just make stuff?

02:38:40.200 | This is all good, but the concern is that

02:38:44.840 | it's still gonna take a while for these things,

02:38:48.380 | for people to learn how to be agentic again

02:38:50.540 | and to be in right relationship with each other

02:38:53.920 | and with the world.

02:38:55.880 | So the message of hope is still people are resilient

02:38:58.420 | and we are building some really amazing technology.

02:39:01.400 | - And I also, like to me, I derive a lot of hope

02:39:03.920 | from individuals in that van.

02:39:08.160 | The power of a single individual to transform the world,

02:39:12.000 | to do positive things for the world is quite incredible.

02:39:14.700 | Now you've been talking about,

02:39:16.000 | it's nice to have as many of those individuals as possible,

02:39:18.840 | but even the power of one, it's kind of magical.

02:39:21.920 | - It is, it is.

02:39:22.760 | We're in a mode now where we can do that.

02:39:24.500 | I think also, part of what I try to do

02:39:26.960 | is in coming to podcasts like yours

02:39:29.020 | and then spamming you with all this philosophical stuff

02:39:31.760 | that I've got going on.

02:39:33.540 | There are a lot of good people out there

02:39:34.800 | trying to put words around the current technological,

02:39:39.800 | social, economic crises that we're facing.

02:39:43.240 | And in the space of a few short years,

02:39:44.600 | I think there has been a lot of great content

02:39:46.360 | produced around this stuff for people who wanna see,

02:39:49.560 | wanna find out more or think more about this.

02:39:52.120 | We're popularizing certain kinds of philosophical ideas

02:39:54.560 | that move people beyond just the,

02:39:56.560 | oh, you're communist, oh, you're capitalist kind of stuff.

02:39:58.720 | Like it's sort of, we're way past that now.

02:40:01.200 | So that also gives me hope that I feel like I myself

02:40:04.620 | am getting a handle on how to think about these things.

02:40:07.320 | It makes me feel like I can hopefully affect,

02:40:11.180 | change for the better.

02:40:12.540 | - We've been sneaking up on this question all over the place.

02:40:15.700 | Let me ask the big, ridiculous question.

02:40:17.580 | What is the meaning of life?

02:40:20.380 | - Wow.

02:40:21.220 | The meaning of life.

02:40:25.260 | Yeah, I don't know.

02:40:29.700 | I mean, I've never really understood that question.

02:40:32.060 | - When you say meaning crisis,

02:40:34.700 | you're saying that there is a search

02:40:39.700 | for a kind of experience that's,

02:40:43.560 | could be described as fulfillment,

02:40:45.460 | as like the aha moment of just like joy.

02:40:50.180 | And maybe when you see something beautiful,

02:40:53.400 | or maybe you have created something beautiful,

02:40:55.220 | that experience that you get,

02:40:58.120 | it feels like it all makes sense.

02:41:01.720 | So some of that is just chemicals coming together

02:41:04.360 | in your mind and all those kinds of things.

02:41:06.440 | But it seems like we're building

02:41:08.800 | a sophisticated collective intelligence

02:41:12.620 | that's providing meaning in all kinds of ways to its members.

02:41:17.120 | And there's a theme to that meaning.

02:41:20.640 | So for a lot of history, I think faith

02:41:24.920 | played an important role.

02:41:26.640 | Faith in God, sort of religion.

02:41:29.640 | I think technology in the modern era

02:41:32.680 | is kind of serving a little bit of a source

02:41:35.200 | of meaning for people, like innovation of different kinds.

02:41:38.100 | I think the old school things of love

02:41:43.520 | and the basics of just being good at stuff.

02:41:46.500 | But you were a physicist, so there's a desire to say,

02:41:51.840 | okay, yeah, but these seem to be like symptoms

02:41:54.400 | of something deeper.

02:41:56.320 | - Right.

02:41:57.160 | - Like why? - With little m meaning.

02:41:58.160 | What's capital M meaning?

02:41:59.080 | - Yeah, what's capital M meaning?

02:42:00.960 | Why are we reaching for order when there is excess of energy?

02:42:05.100 | - I don't know if I can answer the why.

02:42:09.360 | Any why that I come up with, I think, is gonna be,

02:42:11.800 | I'd have to think about that a little more,

02:42:15.480 | maybe get back to you on that.

02:42:17.040 | But I will say this.

02:42:18.220 | We do look at the world through a traditional,

02:42:22.280 | I think most people look at the world through

02:42:24.200 | what I would say is a subject-object

02:42:26.040 | kind of metaphysical lens, that we have our own subjectivity

02:42:29.600 | and then there's all of these object things that are not us.

02:42:34.220 | So I'm me and these things are not me, right?

02:42:37.240 | And I'm interacting with them, I'm doing things to them.

02:42:39.880 | But a different view of the world that looks at it

02:42:42.200 | as much more connected, that realizes,

02:42:45.520 | oh, I'm really quite embedded in a soup of other things.

02:42:50.520 | And I'm simply almost like a standing wave pattern

02:42:54.060 | of different things, right?

02:42:55.440 | So when you look at the world

02:42:58.000 | in that kind of connected sense,

02:42:59.200 | I've recently taken a shine

02:43:02.960 | to this particular thought experiment,

02:43:04.520 | which is what if it was the case

02:43:07.240 | that everything that we touch with our hands,

02:43:12.400 | that we pay attention to,

02:43:13.880 | that we actually give intimacy to,

02:43:16.320 | what if there's actually,

02:43:18.020 | all the mumbo jumbo,

02:43:22.840 | people with the magnetic healing crystals

02:43:25.060 | and all this other kind of stuff and quantum energy stuff,

02:43:28.220 | what if that was a thing?

02:43:30.300 | What if when you,

02:43:31.580 | literally when your hand touches an object,

02:43:34.100 | when you really look at something

02:43:35.380 | and you concentrate and you focus on it

02:43:36.820 | and you really give it attention,

02:43:39.000 | you actually give it,

02:43:40.540 | there is some physical residue of something,

02:43:44.220 | a part of you, a bit of your life force that goes into it.

02:43:48.500 | Okay, now this is of course completely mumbo jumbo stuff.

02:43:51.500 | This is not like, I don't actually think this is real,

02:43:53.720 | but let's do the thought experiment.

02:43:55.680 | What if it was?

02:43:57.740 | What if there actually was some quantum magnetic crystal

02:44:01.760 | and energy field thing that just by touching this can,

02:44:05.360 | this can has changed a little bit somehow.

02:44:09.080 | And it's not much unless you put a lot into it

02:44:11.980 | and you touch it all the time, like your phone, right?

02:44:15.040 | These things gained,

02:44:16.200 | they gain meaning to you a little bit,

02:44:19.680 | but what if there's something that technical objects,

02:44:24.260 | the phone is a technical object.

02:44:25.300 | It does not really receive attention or intimacy

02:44:29.220 | and then allow itself to be transformed by it.

02:44:32.020 | But if it's a piece of wood,

02:44:33.420 | if it's the handle of a knife that your mother used

02:44:36.740 | for 20 years to make dinner for you, right?

02:44:40.380 | What if it's a keyboard that you banged out

02:44:43.900 | your world transforming software library on?

02:44:46.460 | These are technical objects

02:44:47.860 | and these are physical objects,

02:44:48.820 | but somehow there's something to them.

02:44:51.440 | We feel an attraction to these objects

02:44:53.360 | as if we have imbued them with life energy.

02:44:56.000 | - Yeah. - Right?

02:44:57.000 | So if you walk that thought experiment through,

02:44:58.800 | what happens when we touch another person,

02:45:00.480 | when we hug them, when we hold them?

02:45:02.280 | And the reason this ties into my answer for your question

02:45:07.840 | is that there's, if there is such a thing,

02:45:12.840 | if we were to hypothesize, you know,

02:45:15.720 | hypothesize such a thing, it could be that the purpose

02:45:20.720 | of our lives is to imbue as many things

02:45:25.940 | with that love as possible.

02:45:27.580 | - That's a beautiful answer and a beautiful way to end it.

02:45:34.060 | Peter, you're an incredible person.

02:45:36.620 | - Thank you. - Spending so much

02:45:39.220 | in the space of engineering and in the space of philosophy.

02:45:44.780 | I'm really proud to be living in the same city as you.

02:45:48.200 | (Lex laughing)

02:45:49.040 | And I'm really grateful

02:45:51.120 | that you would spend your valuable time with me today.

02:45:53.000 | Thanks so much. - Well, thank you.

02:45:53.840 | I appreciate the opportunity to speak with you.

02:45:56.560 | - Thanks for listening to this conversation

02:45:58.160 | with Peter Wang.

02:45:59.280 | To support this podcast,

02:46:00.640 | please check out our sponsors in the description.

02:46:03.520 | And now let me leave you with some words

02:46:05.680 | from Peter Wang himself.

02:46:07.880 | "We tend to think of people

02:46:09.440 | "as either malicious or incompetent,

02:46:12.440 | "but in a world filled with corruptible

02:46:15.140 | "and unchecked institutions,

02:46:17.060 | "there exists a third thing, malicious incompetence.

02:46:21.020 | "It's a social cancer, and it only appears

02:46:24.120 | "once human organizations scale

02:46:25.940 | "beyond personal accountability."

02:46:27.940 | Thank you for listening and hope to see you next time.

02:46:31.740 | (upbeat music)

02:46:34.320 | (upbeat music)

02:46:36.900 | [BLANK_AUDIO]

Peter Wang: Python and the Source Code of Humans, Computers, and Reality | Lex Fridman Podcast #250

Chapters