Brian Kernighan: UNIX, C, AWK, AMPL, and Go Programming

00:00:00.000 | The following is a conversation with Brian Kernighan,

00:00:03.340 | a professor of computer science at Princeton University.

00:00:07.560 | He was a key figure in the computer science community

00:00:10.140 | in the early Unix days, alongside Unix creators,

00:00:13.760 | Ken Thompson and Dennis Ritchie.

00:00:16.240 | He co-authored the C programming language

00:00:18.540 | with Dennis Ritchie, the creator of C,

00:00:21.440 | and has written a lot of books on programming,

00:00:24.340 | computers and life, including "The Practice of Programming,"

00:00:28.460 | "The Go Programming Language,"

00:00:30.160 | and his latest, "Unix, a History and a Memoir."

00:00:34.100 | He co-created AWK, the text processing language

00:00:36.940 | used by Linux folks like myself.

00:00:39.500 | He co-designed Ample, an algebraic modeling language

00:00:43.800 | that I personally love and have used a lot in my life

00:00:47.580 | for large-scale optimization.

00:00:49.580 | I think I can keep going for a long time

00:00:51.700 | with his creations and accomplishments,

00:00:54.260 | which is funny because given all that,

00:00:56.620 | he's one of the most humble and kind people

00:00:59.180 | I've spoken to on this podcast.

00:01:01.780 | Quick summary of the ads, two new sponsors,

00:01:04.880 | the amazing self-cooling 8Sleep mattress

00:01:09.880 | and Raycon earbuds.

00:01:13.260 | Please consider supporting the podcast

00:01:15.420 | by going to 8sleep.com/lex

00:01:19.060 | and going to buyraycon.com/lex.

00:01:23.020 | Click the links, buy the stuff.

00:01:25.200 | It really is the best way to support this podcast

00:01:27.700 | and the journey I'm on.

00:01:29.660 | If you enjoy this thing, subscribe on YouTube,

00:01:32.020 | review it with Firestarz and Apple Podcasts,

00:01:34.200 | support it on Patreon,

00:01:35.540 | or connect with me on Twitter @LexFriedman.

00:01:38.720 | As usual, I'll do a few minutes of ads now

00:01:41.980 | and never any ads in the middle

00:01:43.300 | that can break the flow of the conversation.

00:01:45.900 | This show is sponsored by 8Sleep

00:01:49.260 | and its incredible Pod Pro mattress

00:01:51.860 | that you can check out at 8sleep.com/lex

00:01:54.700 | to get $200 off.

00:01:57.340 | The mattress controls temperature with an app

00:02:00.700 | and can cool down to as low as 55 degrees.

00:02:03.820 | Research shows the temperature has a big impact

00:02:06.960 | on the quality of our sleep.

00:02:09.020 | Anecdotally, it's been a game changer for me.

00:02:11.260 | I love it.

00:02:12.260 | The Pod Pro is packed with sensors

00:02:14.300 | that track heart rate, heart rate variability

00:02:17.100 | and respiratory rate,

00:02:18.620 | showing it all on their app once you wake up.

00:02:21.400 | Plus, if you have a partner,

00:02:23.500 | you can control the temperature of each side of the bed.

00:02:26.560 | I don't happen to have one,

00:02:28.220 | but the 8Sleep app reminds me

00:02:30.020 | that I should probably get on that.

00:02:32.000 | So ladies, if a temperature controlled mattress

00:02:34.820 | isn't a good reason to apply,

00:02:36.740 | I don't know what is.

00:02:38.880 | The app's health metrics are amazing,

00:02:41.120 | but the cooling alone is honestly worth the money.

00:02:44.420 | As some of you know, I don't always sleep,

00:02:47.260 | but when I do, I choose the 8Sleep Pod Pro mattress.

00:02:51.660 | Check it out at 8sleep.com/lex to get $200 off.

00:02:56.660 | This show is also sponsored by Raycon earbuds.

00:03:01.580 | Get them at buyraycon.com/lex.

00:03:06.020 | They've quickly become my main method

00:03:07.700 | of listening to podcasts, audio books, and music.

00:03:10.440 | When I run, do the pushups and pull-ups

00:03:13.900 | that I've begun to hate at this point,

00:03:15.900 | or just living life.

00:03:17.500 | In fact, I often listen to brown noise with these

00:03:20.260 | when I'm thinking deeply about something.

00:03:22.300 | It helps me focus the mind.

00:03:24.300 | They're super comfortable, pair easily,

00:03:26.460 | great sound, great bass, six hours of playtime.

00:03:30.180 | In fact, for fun, I have one of the earbuds in now,

00:03:33.500 | and I'm listening to Europa by Santana,

00:03:36.340 | probably one of my favorite guitar songs.

00:03:39.120 | It kind of makes me feel like I'm in a music video.

00:03:41.540 | So they told me to say that a bunch of celebrities

00:03:44.820 | use these, like Snoop Dogg, Melissa Etheridge, and Cardi B.

00:03:50.480 | I don't even know who Cardi B is,

00:03:52.720 | but her earbud game is on point.

00:03:55.600 | To mention celebrities I actually care about,

00:03:58.320 | I'm sure if Richard Feynman was still with us,

00:04:01.180 | he'd be listening to the Joe Rogan experience

00:04:03.840 | with Raycon earbuds.

00:04:06.020 | Get them at buyraycon.com/lex.

00:04:09.360 | It's how they know I sent you,

00:04:11.040 | and increases the chance that he'll support this podcast

00:04:13.600 | in the future.

00:04:14.640 | So for all of the sponsors, click all of the links.

00:04:17.640 | It really helps this podcast.

00:04:20.000 | And now, here's my conversation with Brian Kernighan.

00:04:24.040 | Unix started being developed 50 years ago,

00:04:28.600 | maybe more than 50 years ago.

00:04:30.600 | Can you tell the story, like you describe in your new book,

00:04:33.600 | of how Unix was created?

00:04:35.760 | - Ha, if I can remember that far back,

00:04:38.360 | it was some while ago.

00:04:39.580 | So I think the gist of it is that at Bell Labs in 1969,

00:04:45.800 | there were a group of people who had just finished

00:04:48.280 | working on the Multics project,

00:04:49.920 | which was itself a follow-on to CTSS.

00:04:54.300 | So we can go back sort of an infinite regress in time,

00:04:57.040 | but the CTSS was a very, very, very nice time-sharing system.

00:05:01.000 | It was very nice to use.

00:05:02.080 | I actually used it as that summer I spent

00:05:04.560 | in Cambridge in 1966.

00:05:06.920 | - What was the hardware there?

00:05:08.520 | So what's the operating system?

00:05:09.560 | What's the hardware there?

00:05:10.400 | What's the CTSS look like?

00:05:12.160 | - So CTSS looked like,

00:05:14.240 | kind of like a standard time-sharing system.

00:05:17.040 | Certainly at the time, it was the only time-sharing,

00:05:18.920 | if no-

00:05:19.760 | - Let's go back to the basic.

00:05:20.760 | What's a time-sharing system?

00:05:22.360 | - Okay, in the beginning was the word,

00:05:23.760 | and the word was itself.

00:05:24.600 | - And then there was time-sharing systems.

00:05:27.040 | - Yeah, if we go back into, let's call it the 1950s

00:05:29.920 | and early 1960s, most computing was done

00:05:33.280 | on very big computers, physically big,

00:05:35.600 | although not terribly powerful by today's standards,

00:05:38.800 | that were maintained in very large rooms,

00:05:42.120 | and you used things like punch cards

00:05:45.760 | to write your programs on and talk to them.

00:05:47.400 | So you would take a deck of cards,

00:05:49.280 | write your program on it, send it over a counter,

00:05:51.880 | hand it to an operator, and some while later,

00:05:54.480 | back would come something that said,

00:05:55.720 | oh, you made a mistake, and then you'd recycle.

00:05:58.000 | And so it was very, very slow.

00:05:59.440 | So the idea of time-sharing was that you take

00:06:02.240 | basically that same computer, but connect to it

00:06:06.240 | with something that looked like an electric typewriter.

00:06:09.400 | That could be a long distance away, it could be close,

00:06:11.920 | but fundamentally what the operating system did

00:06:14.920 | was to give each person who was connected to it

00:06:18.080 | and wanting to do something a small slice of time

00:06:21.640 | to do a particular job.

00:06:24.840 | So I might be editing a file, so I would be typing,

00:06:28.160 | and every time I hit a keystroke,

00:06:29.480 | the operating system would wake up and said,

00:06:30.960 | oh, he typed character, let me remember that,

00:06:33.400 | and then it'd go back to doing something else.

00:06:34.960 | So it'd be going around and around a group of people

00:06:38.000 | who were trying to get something done,

00:06:39.240 | giving each a small slice of time,

00:06:43.120 | and giving them each the illusion

00:06:45.120 | that they pretty much had the whole machine to themselves,

00:06:47.560 | and hence time-sharing, that is sharing

00:06:49.800 | the computing time resource of the computer

00:06:52.480 | among a number of people who were doing it.

00:06:54.920 | - Without the individual people being aware

00:06:56.840 | that there's others, in a sense, the illusion,

00:06:59.240 | the feelings that the machine is your own.

00:07:02.520 | - Pretty much that was the idea, yes.

00:07:04.240 | You had, if it were well done, and if it were fast enough,

00:07:08.040 | and other people weren't doing too much,

00:07:09.800 | you did have the illusion

00:07:11.000 | that you had the whole machine to yourself,

00:07:13.000 | and it was very much better than the punch card model.

00:07:16.520 | And so CTSS, the compatible time-sharing system,

00:07:19.800 | was, I think, arguably the first of these.

00:07:22.440 | It was done, I guess, technically in '64,

00:07:25.320 | or something like that.

00:07:26.400 | It ran on an IBM 7094, slightly modified

00:07:30.040 | to have twice as much memory as the norm.

00:07:32.800 | It had two banks of 32K words instead of one.

00:07:37.480 | So, (laughs)

00:07:38.840 | - 32K words, yeah.

00:07:40.760 | - Each word was 36 bits, so call it,

00:07:42.880 | about 150 kilobytes times two.

00:07:46.400 | So, by today's standards, that's down in the noise.

00:07:49.600 | But at the time, that was a lot of memory,

00:07:51.480 | and memory was expensive.

00:07:53.240 | So, CTSS was just a wonderful environment to work on.

00:07:56.880 | It was done by people at MIT,

00:07:58.680 | led by Fernando Corbato, Corby,

00:08:02.200 | who died just earlier this year,

00:08:04.280 | and a bunch of other folks.

00:08:06.800 | And so I spent the summer of '66 working on that.

00:08:09.520 | Had a great time, met a lot of really nice people,

00:08:12.640 | and indirectly knew of people at Bell Labs

00:08:17.640 | who were also working on a follow-on to CTSS

00:08:22.520 | that was called Multics.

00:08:24.080 | So, Multics was meant to be the system

00:08:26.000 | that would do everything that CTSS did,

00:08:27.720 | but do it better for a larger population.

00:08:30.760 | All the usual stuff.

00:08:31.720 | - Now, the actual time-sharing, the scheduling,

00:08:35.160 | how much, what's the algorithm

00:08:37.720 | that performs the scheduling?

00:08:39.040 | What's that look like?

00:08:39.880 | How much magic is there?

00:08:40.980 | What are the metrics?

00:08:42.720 | How does it all work in the beginning?

00:08:44.840 | - So, the answer is I don't have a clue.

00:08:46.200 | I think the basic idea was nothing more

00:08:48.280 | than who all wants to get something done.

00:08:50.600 | Suppose that things are very quiet

00:08:52.040 | in the middle of the night,

00:08:53.660 | then I get all the time that I want.

00:08:55.760 | Suppose that you and I are contending at high noon

00:08:58.080 | for something like that,

00:08:59.880 | then probably the simplest algorithm is a round-robin,

00:09:02.420 | one that gives you a bit of time, gives me a bit of time.

00:09:05.120 | And then we could adapt to that.

00:09:07.080 | Like, what are you trying to do?

00:09:08.680 | Are you text editing or are you compiling or something?

00:09:12.000 | And then we might adjust the scheduler

00:09:13.600 | according to things like that.

00:09:15.040 | - So, okay, so Multics was trying to just do some of the,

00:09:18.960 | clean it up a little bit.

00:09:20.280 | - Well, it was meant to be much more than that.

00:09:22.280 | So, Multics was the multiplexed information

00:09:24.320 | and computing service,

00:09:25.460 | and it was meant to be a very large thing

00:09:27.840 | that would provide computing utility,

00:09:29.980 | something that where you could actually think of it

00:09:32.940 | as just a plug-in-the-wall service.

00:09:35.080 | Sort of like cloud computing today, same idea,

00:09:38.360 | but 50 odd years earlier.

00:09:40.680 | And so what Multics offered

00:09:43.800 | was a richer operating system environment,

00:09:46.560 | a piece of hardware that was better designed

00:09:48.880 | for doing the kind of sharing of resources,

00:09:53.160 | and presumably lots of other things.

00:09:55.960 | - Do you think people at that time had the dream

00:09:58.560 | of what cloud computing is starting to become now,

00:10:01.120 | which is computing is everywhere,

00:10:03.160 | that you can just plug in almost,

00:10:05.760 | and you never know how the magic works.

00:10:09.080 | You just kind of plug in, add your little computation

00:10:11.760 | that you need to perform, and it does it.

00:10:13.720 | Was that the dream?

00:10:14.920 | - I don't know where that was the dream.

00:10:16.040 | I wasn't part of it at that point.

00:10:17.440 | Remember, I was an intern for a summer.

00:10:19.280 | But my sense is, given that it was over 50 years ago,

00:10:23.280 | yeah, they had that idea

00:10:24.400 | that it was an information utility,

00:10:26.280 | that it was something where

00:10:27.680 | if you had a computing task to do,

00:10:29.840 | you could just go and do it.

00:10:31.680 | Now, I'm betting that they didn't have the same view

00:10:35.880 | of computing for the masses, let's call it,

00:10:38.840 | the idea that your grandmother would be shopping on Amazon.

00:10:43.240 | I don't think that was part of it.

00:10:45.000 | But if your grandmother were a programmer,

00:10:47.160 | it might be very easy for her to go

00:10:49.080 | and use this kind of utility.

00:10:51.500 | - What was your dream of computers at that time?

00:10:53.680 | What did you see as the future of computers?

00:10:55.680 | 'Cause you have predicted what computers are today.

00:10:59.600 | - Oh, short answer, absolutely not.

00:11:01.760 | I have no clue.

00:11:02.600 | I'm not sure I had a dream.

00:11:03.880 | It was a dream job in the sense that

00:11:05.920 | I really enjoyed what I was doing.

00:11:07.480 | I was surrounded by really, really nice people.

00:11:10.040 | Cambridge is a very fine city to live in in the summer,

00:11:12.880 | less so in the winter when it snows,

00:11:14.300 | but in the summer, it was a delightful time.

00:11:16.880 | And so I really enjoyed all of that stuff

00:11:19.200 | and I learned things.

00:11:20.360 | And I think the good fortune of being there for summer

00:11:25.000 | led me then to get a summer job at Bell Labs

00:11:27.200 | the following summer.

00:11:28.480 | And that was quite useful for the future.

00:11:31.760 | - So Bell Labs is this magical, legendary place.

00:11:35.960 | So first of all, where is Bell Labs?

00:11:39.040 | And can you start talking about that journey

00:11:44.040 | towards Unix at Bell Labs?

00:11:46.600 | - Yeah, so Bell Labs is physically scattered around,

00:11:50.160 | at the time, scattered around New Jersey.

00:11:52.320 | The primary location was in a town called Murray Hill,

00:11:54.920 | or a location called Murray Hill.

00:11:56.320 | It was actually across the boundary

00:11:58.680 | between two small towns in New Jersey

00:12:00.840 | called New Providence and Berkeley Heights.

00:12:03.400 | Think of it as about 15, 20 miles straight west

00:12:05.400 | of New York City, and therefore about an hour north

00:12:08.880 | of here in Princeton.

00:12:10.000 | And at that time, it had, make up a number,

00:12:15.040 | three or 4,000 people there, many of whom had PhDs

00:12:17.960 | and mostly doing physical sciences,

00:12:20.760 | chemistry, physics, materials kinds of things,

00:12:24.400 | but very strong math and rapidly growing interest

00:12:29.040 | in computing as people realized you could do things

00:12:31.160 | with computers that you might not have been able

00:12:33.920 | to do before.

00:12:35.240 | You could replace labs with computers

00:12:37.440 | that had worked on models of what was going on.

00:12:40.360 | So that was the essence of Bell Labs.

00:12:44.080 | And again, I wasn't a permanent employee there.

00:12:46.520 | That was another internship.

00:12:47.880 | I got lucky in internships.

00:12:49.520 | - I mean, if you could just linger on it a little bit,

00:12:52.480 | what was in the air there?

00:12:55.440 | 'Cause some of the, the number of Nobel Prizes,

00:12:57.720 | the number of Turing Awards,

00:12:58.920 | and just legendary computer scientists

00:13:00.960 | that come from there, inventions including,

00:13:04.040 | developments including Unix, it's just, it's unbelievable.

00:13:07.840 | So was there something special about that place?

00:13:11.560 | - Oh, I think there was very definitely something special.

00:13:14.560 | I mentioned the number of people.

00:13:15.760 | It's a very large number of people, very highly skilled,

00:13:19.040 | and working in an environment where there was always

00:13:21.280 | something interesting to work on,

00:13:23.080 | because the goal of Bell Labs,

00:13:25.080 | which was a small part of AT&T,

00:13:27.240 | which provided basically the country's phone service.

00:13:30.120 | The goal of AT&T was to provide service for everybody.

00:13:33.400 | And the goal of Bell Labs was to try and make that service

00:13:36.920 | keep getting better, so improving service.

00:13:39.160 | And that meant doing research on a lot of different things,

00:13:43.880 | physical devices like the transistor,

00:13:46.400 | or fiber optical cables, or microwave systems,

00:13:50.840 | all of these things the labs worked on.

00:13:53.200 | And it was kind of just the beginning of real boom times

00:13:56.560 | in computing as well, 'cause when I was there,

00:13:59.080 | I went there first in '66.

00:14:01.120 | So computing was at that point fairly young,

00:14:04.520 | and so people were discovering that you could do

00:14:06.480 | lots of things with computers.

00:14:08.720 | - So how was Unix born?

00:14:10.800 | - So Multics, in spite of having an enormous number

00:14:14.560 | of really good ideas, lots of good people working on it,

00:14:16.840 | fundamentally didn't live up, at least in the short run,

00:14:20.040 | and I think ultimately really ever,

00:14:22.160 | to its goal of being this information utility.

00:14:25.560 | It was too expensive, and certainly what was promised

00:14:29.200 | was delivered much too late.

00:14:31.280 | And so in roughly the beginning of 1969,

00:14:34.600 | Bell Labs pulled out of the project.

00:14:37.200 | The project at that point had included MIT,

00:14:41.640 | Bell Labs, and General Electric.

00:14:44.040 | General Electric made computers,

00:14:45.480 | so General Electric was the hardware operation.

00:14:48.320 | So Bell Labs, realizing this wasn't going anywhere

00:14:50.880 | on a timescale they cared about, pulled out of the project.

00:14:54.160 | And this left several people with an acquired taste

00:14:59.160 | for really, really nice computing environments,

00:15:01.640 | but no computing environment.

00:15:03.520 | And so they started thinking about what could you do

00:15:06.800 | if you were gonna design a new operating system

00:15:09.480 | that would provide the same kind of comfortable computing

00:15:12.920 | as CTSS had, but also the facilities of something

00:15:16.040 | like Multics sort of brought forward.

00:15:17.960 | And so they did a lot of paper design stuff,

00:15:21.720 | and at the same time, Ken Thompson found

00:15:23.920 | what is characterized as a little-used PDP-7,

00:15:27.280 | where he started to do experiments with file systems,

00:15:31.080 | just how do you store information on a computer

00:15:33.640 | in an efficient way, and then this famous story

00:15:36.360 | that his wife went away to California for three weeks

00:15:39.160 | taking their one-year-old son, and three weeks,

00:15:43.280 | and he sat down and wrote an operating system,

00:15:45.640 | which ultimately became Unix.

00:15:47.440 | So software productivity was good in those days.

00:15:50.280 | - So PDP, what's a PDP-7?

00:15:52.000 | So it's a piece of hardware?

00:15:53.360 | - Yeah, it's a piece of hardware.

00:15:54.520 | It was one of early machines made

00:15:56.720 | by Digital Equipment Corporation, DEC,

00:15:59.840 | and it was a mini-computer, so-called.

00:16:03.440 | It had, I would have to look up the numbers exactly,

00:16:07.440 | but it had a very small amount of memory,

00:16:09.320 | maybe 16K, 16-bit words or something like that,

00:16:13.280 | relatively slow, probably not super expensive,

00:16:17.120 | maybe, again, making this up, I'd have to look it up,

00:16:19.760 | $100,000 or something like that.

00:16:21.840 | - Which is not super expensive in those days, right?

00:16:24.360 | - It was expensive.

00:16:25.400 | It was enough that you and I probably

00:16:26.840 | wouldn't be able to buy one,

00:16:27.680 | but a modest group of people could get together.

00:16:30.880 | But in any case, it came out, if I recall, in 1964,

00:16:34.880 | so by 1969, it was getting a little obsolete,

00:16:38.640 | and that's why it was little-used.

00:16:41.520 | - If you can sort of comment, what do you think

00:16:43.400 | it's like to write an operating system like that?

00:16:45.680 | So that process that Ken went through in three weeks,

00:16:48.680 | because you were, I mean, you're a part of that process.

00:16:52.800 | You contributed a lot to Enix's early development.

00:16:57.600 | So what do you think it takes to do that first step,

00:17:01.360 | that first kind of, from design to reality on the PDP?

00:17:05.460 | - Well, let me correct one thing.

00:17:07.160 | I had nothing to do with it, so I did not write it.

00:17:10.440 | I have never written operating system code.

00:17:12.560 | And so I don't know.

00:17:16.400 | Now, an operating system is simply code,

00:17:18.960 | and this first one wasn't very big,

00:17:21.400 | but it's something that lets you run processes of some,

00:17:24.600 | lets you execute some kind of code that has been written.

00:17:27.320 | It lets you store information for periods of time

00:17:30.820 | so that it doesn't go away when you turn the power off

00:17:33.160 | or reboot or something like that.

00:17:35.220 | And there's kind of a core set of tools

00:17:38.560 | that are technically not part of an operating system,

00:17:40.960 | but you probably need them.

00:17:42.400 | In this case, Ken wrote an assembler for the PDP-7

00:17:46.080 | that worked.

00:17:46.900 | He did a text editor so that he could actually create text.

00:17:49.880 | He had the file system stuff that he had been working on,

00:17:52.160 | and then the rest of it was just a way to load things,

00:17:55.760 | executable code from the file system into the memory,

00:17:58.840 | give it control, and then recover control

00:18:01.880 | when it was finished or in some other way quit.

00:18:04.840 | - What was the code written in,

00:18:06.640 | primarily the programming language?

00:18:08.200 | Was it in assembly?

00:18:09.200 | - Yeah, PDP-7 assembler that Ken created.

00:18:12.320 | These things were assembly language until probably the,

00:18:17.440 | call it 1973 or '74, something like that.

00:18:20.180 | - Forgive me if it's a dumb question,

00:18:23.000 | but it feels like a daunting task

00:18:25.200 | to write any kind of complex system in assembly.

00:18:28.760 | - Absolutely.

00:18:29.600 | - It feels like impossible to do any kind of

00:18:32.940 | what we think of as software engineering with assembly

00:18:36.160 | 'cause to work on a big picture sort of.

00:18:40.080 | - I think it's hard.

00:18:41.480 | It's been a long time since I wrote assembly language.

00:18:43.760 | It is absolutely true that in assembly language,

00:18:45.600 | if you make a mistake, nobody tells you.

00:18:47.140 | There are no training wheels whatsoever.

00:18:49.320 | And so stuff doesn't work.

00:18:51.120 | Now what?

00:18:51.960 | - There's no debuggers.

00:18:53.400 | - Well, there could be debuggers,

00:18:54.440 | but that's the same problem, right?

00:18:56.780 | How do you actually get something

00:18:58.920 | that will help you debug it?

00:19:00.400 | So part of it is an ability to see the big picture.

00:19:05.640 | Now these systems were not big

00:19:07.240 | in the sense that today's pictures are.

00:19:08.680 | So the big picture was in some sense more manageable.

00:19:11.840 | I mean, then realistically, there's an enormous variation

00:19:15.240 | in the capabilities of programmers.

00:19:17.520 | And Ken Thompson, who did that first one,

00:19:20.200 | is kind of the singularity in my experience of programmers

00:19:24.520 | with no disrespect to you or even to me.

00:19:27.760 | He's in several leagues removed.

00:19:31.000 | - I know there's levels.

00:19:33.760 | It's a fascinating thing that there are unique stars

00:19:37.160 | in particular in the programming space

00:19:39.800 | and at a particular time.

00:19:41.160 | The time matters too, the timing

00:19:42.720 | of when that person comes along.

00:19:44.440 | And a wife does have to leave.

00:19:47.800 | There's this weird timing that happens

00:19:49.760 | that and then all of a sudden

00:19:50.760 | something beautiful is created.

00:19:52.280 | I mean, how does it make you feel

00:19:53.440 | that there's a system that was created in three weeks

00:19:58.320 | or maybe you can even say on a whim,

00:20:01.200 | but not really, but of course quickly,

00:20:03.660 | that is now, you could think of most of the computers

00:20:07.140 | in the world run on a Unix-like system.

00:20:10.640 | - Right.

00:20:11.480 | (laughs)

00:20:12.880 | - How do you, if you kind of zoom

00:20:15.280 | from the alien perspective,

00:20:16.560 | if you're just observing Earth,

00:20:18.420 | that all of a sudden these computers took over the world

00:20:21.000 | and they started from this little initial seed of Unix,

00:20:24.840 | how does that make you feel?

00:20:26.640 | - It's quite surprising.

00:20:27.680 | And you asked earlier about prediction.

00:20:30.280 | The answer is no.

00:20:31.120 | There's no way you could predict that kind of evolution.

00:20:33.980 | And I don't know whether it was inevitable

00:20:37.140 | or just a whole sequence of blind luck.

00:20:39.120 | I suspect more of the latter.

00:20:40.960 | And so I look at it and think, gee, that's kind of neat.

00:20:45.360 | (laughs)

00:20:47.760 | I think the real question is what does Ken think about that?

00:20:51.020 | 'Cause he's the guy arguably from whom it really came.

00:20:55.180 | Tremendous contributions from Dennis Ritchie

00:20:57.220 | and then others around in that Bell Labs environment.

00:21:00.220 | But if you had to pick a single person, that would be Ken.

00:21:04.540 | - So you've written a new book,

00:21:06.260 | Unix, a history and a memoir.

00:21:08.340 | Are there some memorable human stories,

00:21:10.940 | funny or profound from that time that just kind of stand out?

00:21:14.180 | - Oh, there's a lot of them in a sense.

00:21:15.740 | And again, it's a question of can you resurrect them

00:21:18.220 | in real time? - Memory?

00:21:19.060 | (laughs)

00:21:20.420 | - Memory fails.

00:21:21.780 | But I think part of it was that Bell Labs at the time

00:21:25.100 | was a very special kind of place to work

00:21:27.100 | because there were a lot of interesting people

00:21:29.020 | and the environment was very, very open and free.

00:21:31.740 | It was a very cooperative environment,

00:21:33.340 | very friendly environment.

00:21:34.420 | And so if you had an interesting problem,

00:21:36.000 | you go and talk to somebody

00:21:37.220 | and they might help you with the solution.

00:21:39.320 | And it was a kind of a fun environment too

00:21:43.660 | in which people did strange things

00:21:46.660 | and often tweaking the bureaucracy in one way or another.

00:21:51.660 | - So rebellious in certain kinds of ways.

00:21:54.940 | - In some ways, yeah, absolutely.

00:21:56.700 | I think most people didn't take too kindly

00:21:58.860 | to the bureaucracy and I'm sure the bureaucracy

00:22:01.540 | put up with an enormous amount

00:22:03.820 | that they didn't really want to.

00:22:05.980 | - So maybe to linger on it a little bit,

00:22:09.500 | do you have a sense of what the philosophy

00:22:11.820 | that characterizes Unix is, the design?

00:22:15.300 | Not just the initial, but just carry through the years,

00:22:18.860 | just being there, being around it.

00:22:20.640 | What's the fundamental philosophy behind the system?

00:22:23.340 | - I think one aspect of the fundamental philosophy

00:22:25.620 | was to provide an environment that made it easy to write

00:22:29.100 | or easier, productive to write programs.

00:22:31.940 | So it was meant as a programmer environment.

00:22:33.700 | It wasn't meant specifically as something

00:22:36.060 | to do some other kind of job.

00:22:38.380 | For example, it was used extensively for word processing,

00:22:41.380 | but it wasn't designed as a word processing system.

00:22:43.700 | It was used extensively for lab control,

00:22:45.800 | but it wasn't designed for that.

00:22:47.340 | It was used extensively as a front end

00:22:49.460 | for big other systems, big dumb systems,

00:22:52.460 | but it wasn't designed for that.

00:22:53.780 | It was meant to be an environment

00:22:55.620 | where it was really easy to write programs.

00:22:57.980 | So the programmers could be highly productive.

00:23:00.260 | And part of that was to be a community.

00:23:03.100 | And there's some observation from Dennis Ritchie,

00:23:05.780 | I think at the end of the book,

00:23:06.940 | it says that from his standpoint,

00:23:09.700 | the real goal was to create a community

00:23:11.900 | where people could work as programmers on a system.

00:23:16.900 | I think in that sense, certainly for many, many years,

00:23:19.580 | it succeeded quite well at that.

00:23:22.660 | And part of that is the technical aspects

00:23:25.020 | of because it made it really easy to write programs,

00:23:27.580 | people did write interesting programs.

00:23:29.500 | Those programs tended to be used by other programmers.

00:23:31.980 | And so it was kind of a virtuous circle

00:23:34.020 | of more and more stuff coming out

00:23:36.540 | that was really good for programmers.

00:23:39.340 | - And you were part of that community of programmers.

00:23:41.780 | So what was it like writing programs on that early Unix?

00:23:45.780 | - It was a blast, it really was.

00:23:47.380 | You know, I liked to program.

00:23:51.100 | I'm not a terribly good programmer,

00:23:52.820 | but it was a lot of fun to write code.

00:23:55.260 | And in the early days,

00:23:56.620 | there was an enormous amount of what you would,

00:23:58.220 | today I suppose, call low-hanging fruit.

00:24:00.100 | People hadn't done things before,

00:24:02.500 | and this was this new environment,

00:24:04.300 | and the whole combination of nice tools

00:24:07.620 | and very responsive system and tremendous colleagues

00:24:11.580 | made it possible to write code.

00:24:13.660 | You could have an idea in the morning,

00:24:16.420 | you could do an experiment with it,

00:24:19.060 | you could have something limping along that night

00:24:21.300 | or the next day, and people would react to it,

00:24:23.580 | and they would say, "Oh, that's wonderful,

00:24:25.860 | "but you're really screwed up here."

00:24:27.780 | And the feedback loop was then very, very short and tight.

00:24:31.660 | And so a lot of things got developed fairly quickly

00:24:34.940 | that in many cases still exist today.

00:24:39.860 | And I think that was part of what made it fun,

00:24:43.220 | because programming itself is fun.

00:24:44.660 | It's puzzle solving in a variety of ways,

00:24:46.860 | but I think it's even more fun when you do something

00:24:49.940 | that somebody else then uses.

00:24:52.260 | Even if they whine about it not working,

00:24:54.540 | the fact that they used it is part of the reward mechanism.

00:24:58.500 | - And what was the method of interaction,

00:25:00.420 | the communication, that feedback loop?

00:25:03.580 | I mean, this is before the internet.

00:25:05.380 | - Certainly before the internet.

00:25:07.460 | It was mostly physical right there,

00:25:11.580 | somebody would come into your office and say something.

00:25:13.620 | - So these places are all close by,

00:25:15.220 | like offices are nearby,

00:25:16.700 | so you're really lively in interaction.

00:25:18.940 | - Yeah, yeah, no, Bell Labs was fundamentally

00:25:20.940 | one giant building, and most of the people

00:25:23.020 | who were involved in this Unix stuff

00:25:24.380 | were in two or three quarters, and there was a room,

00:25:27.620 | oh, how big was it?

00:25:29.380 | Probably, call it 50 feet by 50 feet,

00:25:33.460 | make up a number of that, which had some access

00:25:37.300 | to computers there as well as in offices,

00:25:39.940 | and people hung out there, and it had a coffee machine.

00:25:42.900 | And so it was mostly very physical.

00:25:46.300 | We did use email, of course,

00:25:48.020 | but it was fundamentally, for a long time,

00:25:52.700 | all on one machine, so there was no need for internet.

00:25:56.500 | - It's fascinating to think about what computing

00:25:58.660 | would be today without Bell Labs.

00:26:00.900 | 'Cause it seems so many,

00:26:02.420 | people being in the vicinity of each other,

00:26:06.380 | that sort of getting that quick feedback,

00:26:08.460 | working together, so many brilliant people.

00:26:11.460 | I don't know where else that could have existed

00:26:13.180 | in the world, I mean, given how that came together.

00:26:16.140 | (laughs)

00:26:17.980 | Yeah, how does that make you feel,

00:26:19.700 | that little element of history?

00:26:23.180 | - Well, I think that's very nice,

00:26:24.580 | but in a sense, it's survivor bias,

00:26:26.700 | and if it hadn't happened at Bell Labs,

00:26:29.220 | there were other places that were doing

00:26:30.980 | really interesting work as well.

00:26:32.860 | Xerox PARC is perhaps the most obvious one.

00:26:34.980 | Xerox PARC contributed an enormous amount of good material,

00:26:38.140 | and many of the things we take for granted today

00:26:40.540 | in the same way came from Xerox PARC experience.

00:26:43.300 | I don't think they capitalized in the long run as much.

00:26:46.420 | Their parent company was perhaps not as lucky

00:26:49.740 | in capitalizing on this, who knows?

00:26:52.780 | But that's certainly another place

00:26:54.940 | where there was a tremendous amount of influence.

00:26:58.020 | There were a lot of good university activities.

00:27:00.180 | MIT was obviously no slouch in this kind of thing,

00:27:03.660 | and others as well.

00:27:07.100 | - So Unix turned out to be open source

00:27:10.620 | because of the various ways that AT&T operated,

00:27:13.540 | and sort of it had to, the focus was on telephones.

00:27:17.660 | - I think that's a mischaracterization in a sense.

00:27:21.540 | It absolutely was not open source.

00:27:23.820 | It was very definitely proprietary, licensed,

00:27:27.820 | but it was licensed freely to universities

00:27:30.820 | in source code form for many years,

00:27:33.500 | and because of that, generations of university students

00:27:37.660 | and their faculty people grew up knowing about Unix,

00:27:41.740 | and there was enough expertise in the community

00:27:44.740 | that it then became possible for people

00:27:46.660 | to kind of go off in their own direction

00:27:48.100 | and build something that looked Unix-like.

00:27:50.540 | The Berkeley version of Unix started

00:27:54.780 | with that licensed code and gradually picked up

00:27:58.220 | enough of its own code contributions,

00:28:01.720 | notably from people like Bill Joy,

00:28:04.060 | that eventually it was able to become

00:28:07.740 | completely free of any AT&T code.

00:28:10.500 | Now, there was an enormous amount of legal jockeying

00:28:13.140 | around this in the late, early to late '80s, early '90s,

00:28:17.460 | something like that, and then not,

00:28:22.460 | I guess the open source movement might have started

00:28:26.020 | when Richard Stallman started to think about this

00:28:28.300 | in the late '80s, and by 1991, when Torvalds decided

00:28:32.660 | he was going to do a Unix-like operating system,

00:28:37.060 | there was enough expertise in the community

00:28:40.460 | that first he had a target, he could see what to do,

00:28:44.300 | because the kind of the Unix system call interface

00:28:47.500 | and the tools and so on were there,

00:28:49.380 | and so he was able to build an operating system

00:28:53.440 | that at this point, when you say Unix,

00:28:56.140 | in many cases what you're really thinking is Linux.

00:28:58.380 | - Linux, yeah.

00:28:59.860 | - It's funny that from my distant perception,

00:29:02.900 | I felt that Unix was open source

00:29:04.740 | without actually knowing it, but what you're really saying,

00:29:08.940 | it was just freely licensed.

00:29:11.540 | - It was freely licensed.

00:29:12.940 | - So it felt open source, because universities

00:29:15.860 | are not trying to make money, so it felt open source

00:29:18.540 | in the sense that you can get access if you wanted.

00:29:20.540 | - Right, and a very, very, very large number

00:29:23.020 | of universities had the license, and they were able

00:29:24.900 | to talk to all the other universities who had the license,

00:29:27.380 | and so technically not open, technically belonging to AT&T,

00:29:32.220 | pragmatically pretty open.

00:29:34.860 | - And so there's a ripple effect that all the faculty

00:29:37.340 | and the students, then they all grew up,

00:29:38.980 | and then they went throughout the world

00:29:41.860 | and permeated in that kind of way.

00:29:45.420 | So what kind of features do you think make

00:29:49.820 | for a good operating system?

00:29:51.280 | If you take the lessons of Unix, you said,

00:29:57.240 | make it easy for programmers, that seems to be

00:30:00.980 | an important one, but also Unix turned out

00:30:05.340 | to be exceptionally robust and efficient.

00:30:08.180 | - Right.

00:30:09.020 | - So is that an accident when you focus on the programmer,

00:30:12.100 | or is that a natural outcome?

00:30:14.740 | - I think part of the reason for efficiency

00:30:17.540 | was that it began on extremely modest hardware,

00:30:21.180 | very, very, very tiny, and so you couldn't get carried away,

00:30:23.980 | you couldn't do a lot of complicated things

00:30:26.680 | because you just didn't have the resources,

00:30:30.000 | either processor speed or memory,

00:30:32.360 | and so that enforced a certain minimality of mechanisms,

00:30:37.360 | and maybe a search for generalizations

00:30:40.040 | so that you would find one mechanism

00:30:41.800 | that served for a lot of different things

00:30:43.480 | rather than having lots of different special cases.

00:30:45.880 | I think the file system in Unix is a good example

00:30:48.760 | of that file system interface in its fundamental form

00:30:51.440 | is extremely straightforward,

00:30:53.560 | and that means that you can write code

00:30:55.580 | very, very effectively for the file system,

00:30:58.880 | and then one of those generalizations

00:31:02.680 | is that, gee, that file system interface

00:31:04.560 | works for all kinds of other things as well,

00:31:06.740 | and so in particular, the idea of reading and writing

00:31:09.560 | to devices is the same as reading and writing

00:31:11.960 | to a disk that has a file system,

00:31:14.600 | and then that gets carried further in other parts.

00:31:17.820 | The world processes become, in effect,

00:31:21.800 | files in a file system,

00:31:23.360 | and the Plan 9 operating system,

00:31:25.800 | which came along, I guess, in the late '80s

00:31:27.420 | or something like that, took a lot of those ideas

00:31:30.520 | from the original Unix and tried to push

00:31:32.980 | the generalization even further

00:31:34.720 | so that in Plan 9, a lot of different resources

00:31:37.160 | are file systems.

00:31:38.160 | They all share that interface,

00:31:40.040 | so that would be one example

00:31:42.200 | where finding the right model of how to do something

00:31:45.480 | means that an awful lot of things become simpler,

00:31:48.040 | and it means, therefore, that more people can do useful,

00:31:50.360 | interesting things with them

00:31:51.720 | without having to think as hard about it.

00:31:54.440 | - So you said you're not a very good programmer.

00:31:56.960 | - True.

00:31:58.480 | - You're the world's most modest human being, okay,

00:32:01.400 | but you'll continue saying that.

00:32:03.280 | I understand how this works,

00:32:04.440 | but you do radiate a sort of love for programming,

00:32:07.760 | so let me ask, do you think programming

00:32:10.840 | is more an art or a science?

00:32:13.240 | Is it creativity or kind of rigor?

00:32:16.640 | - I think it's some of each.

00:32:18.040 | It's some combination.

00:32:20.760 | Some of the art is figuring out what it is

00:32:22.640 | that you really wanna do, what should that program be,

00:32:25.480 | what would make a good program,

00:32:27.520 | and that's some understanding of what the task is,

00:32:30.600 | what the people who might use this program want,

00:32:33.560 | and I think that's art in many respects.

00:32:37.720 | The science part is trying to figure out how to do it well,

00:32:40.440 | and some of that is real computer science-y stuff,

00:32:45.200 | like what algorithm should we use at some point,

00:32:48.040 | mostly in the sense of being careful to use algorithms

00:32:52.280 | that will actually work properly, scale properly,

00:32:56.200 | avoiding quadratic algorithms

00:32:57.960 | when a linear algorithm should be the right thing,

00:33:01.240 | that kind of more formal view of it.

00:33:04.040 | Same thing for data structures,

00:33:06.360 | but also it's, I think, an engineering field as well,

00:33:10.280 | and engineering's not quite the same as science

00:33:12.440 | because in engineering you're working with constraints.

00:33:15.240 | You have to figure out not only,

00:33:18.520 | so what is a good algorithm for this kind of thing,

00:33:20.920 | but what's the most appropriate algorithm

00:33:22.800 | given the amount of time we have to compute,

00:33:26.200 | the amount of time we have to program,

00:33:27.920 | what's likely to happen in the future with maintenance,

00:33:30.720 | who's gonna pick this up in the future,

00:33:33.040 | all of those kind of things that if you're an engineer,

00:33:35.800 | you get to worry about,

00:33:37.160 | whereas if you think of yourself as a scientist,

00:33:39.120 | well, you can maybe push them over the horizon in a way,

00:33:42.040 | and if you're an artist, what's that?

00:33:44.080 | (laughing)

00:33:45.320 | - So just on your own personal level,

00:33:48.560 | what's your process like of writing a program,

00:33:50.640 | say a small and large sort of tinkering with stuff?

00:33:55.640 | Do you just start coding right away

00:33:57.960 | and just kind of evolve iteratively with a loose notion,

00:34:02.960 | or do you plan on a sheet of paper first

00:34:05.760 | and then kind of design what they teach you

00:34:09.280 | in the kind of software engineering courses in undergrad

00:34:12.640 | or something like that?

00:34:13.640 | What's your process like?

00:34:14.880 | - It's certainly much more the informal incremental.

00:34:17.440 | First, I don't write big programs at this point.

00:34:20.160 | It's been a long time since I wrote a program

00:34:21.720 | that was more than, I call it a few hundred or more lines,

00:34:25.560 | something like that.

00:34:27.000 | Many of the programs I write are experiments

00:34:29.040 | for either something I'm curious about

00:34:31.640 | or often for something that I wanna talk about in a class,

00:34:34.560 | and so those necessarily tend to be relatively small.

00:34:38.960 | A lot of the kind of code I write these days

00:34:41.400 | tends to be for sort of exploratory data analysis

00:34:44.280 | where I've got some collection of data

00:34:46.240 | and I wanna try and figure out

00:34:47.080 | what on earth is going on in it,

00:34:49.200 | and for that, those programs tend to be very small.

00:34:52.280 | Sometimes you're not even programming,

00:34:53.920 | you're just using existing tools like counting things,

00:34:57.800 | or sometimes you're writing ox scripts

00:35:00.200 | because two or three lines will tell you something

00:35:02.800 | about a piece of data, and then when it gets bigger,

00:35:05.160 | well, then I will probably write something in Python

00:35:08.760 | because that scales better,

00:35:10.720 | up to, call it a few hundred lines or something like that,

00:35:14.560 | and it's been a long time since I wrote programs

00:35:16.280 | that were much more than that.

00:35:18.560 | - Speaking of data exploration and awk,

00:35:21.000 | first, what is awk?

00:35:23.600 | - So awk is a scripting language

00:35:25.400 | that was done by myself, Al Aho, and Peter Weinberger.

00:35:30.240 | We did that originally in the late '70s.

00:35:32.680 | It was a language that was meant to make it really easy

00:35:34.800 | to do quick and dirty tasks like counting things

00:35:39.360 | or selecting interesting information

00:35:42.800 | from basically all text files,

00:35:45.040 | rearranging it in some way or summarizing it.

00:35:48.120 | - It runs a command on each line of a file.

00:35:51.520 | I mean, it's still exceptionally widely used today.

00:35:55.560 | - Oh, absolutely, yeah.

00:35:56.880 | - It's so simple and elegant,

00:35:58.520 | sort of the way to explore data.

00:36:01.520 | Turns out you can just write a script

00:36:03.160 | that does something seemingly trivial on a single line,

00:36:07.160 | and that giving you that slice of the data

00:36:09.880 | somehow reveals something fundamental about the data,

00:36:13.160 | and that keeps, that seems to work still.

00:36:17.120 | - Yeah, it's very good for that kind of thing,

00:36:19.120 | and that's sort of what it was meant for.

00:36:21.200 | I think what we didn't appreciate

00:36:22.560 | was that the model was actually quite good

00:36:24.800 | for a lot of data processing kinds of tasks,

00:36:27.640 | and that it's kept going as long as it has,

00:36:30.760 | 'cause at this point it's over 40 years old,

00:36:32.840 | and it's still, I think, a useful tool.

00:36:35.960 | And well, this is paternal interest, I guess,

00:36:38.440 | but I think in terms of programming languages,

00:36:41.000 | you get the most bang for the buck by learning AWK,

00:36:44.240 | and it doesn't scale to big programs,

00:36:46.600 | but it does pretty darn well on these little things

00:36:49.960 | where you just wanna see all the somethings in something.

00:36:52.920 | So yeah, I find, I probably write more AWK

00:36:56.000 | than anything else at this point.

00:36:58.680 | - So what kind of stuff do you love about AWK?

00:37:01.040 | Like, is there, if you can comment on sort of things

00:37:05.160 | that give you joy when you can, in a simple program,

00:37:10.240 | reveal something about the data,

00:37:11.600 | is there something that stands out from particular features?

00:37:14.600 | - I think it's mostly the selection of default behaviors,

00:37:19.200 | that you sort of hinted at it a moment ago.

00:37:21.080 | What AWK does is to read through a set of files,

00:37:24.800 | and then within each file,

00:37:26.240 | it reads through each of the lines,

00:37:28.440 | and then on each of the lines,

00:37:29.960 | it has a set of patterns that it looks for,

00:37:33.080 | that's your AWK program,

00:37:34.720 | and if one of the patterns matches,

00:37:36.880 | there is a corresponding action that you might perform,

00:37:39.880 | and so it's kind of a quadruply nested loop

00:37:43.440 | or something like that,

00:37:45.040 | and that's all completely automatic.

00:37:46.600 | You don't have to say anything about it,

00:37:48.120 | you just write the pattern and the action,

00:37:49.920 | and then run the data by it,

00:37:52.000 | and so that paradigm for programming

00:37:54.160 | is a very natural and effective one,

00:37:56.880 | and I think we captured that reasonably well in AWK,

00:38:00.160 | and it does other things for free as well.

00:38:01.920 | It splits the data into fields,

00:38:04.240 | so that on each line,

00:38:05.240 | there's fields separated by white space or something,

00:38:07.560 | and so it does that for free,

00:38:08.800 | you don't have to say anything about it,

00:38:11.200 | and it collects information as it goes along,

00:38:13.760 | like what line are we on,

00:38:15.280 | how many fields are there on this line?

00:38:18.000 | So lots of things that just make it so that a program

00:38:20.800 | which in another language, let's say Python,

00:38:24.040 | would be five, 10, 20 lines,

00:38:26.240 | in AWK it's one or two lines.

00:38:28.040 | - And so because it's one or two lines,

00:38:29.600 | you can do it on the shell,

00:38:31.840 | you don't have to open up another whole thing,

00:38:33.720 | you can just do it right there

00:38:35.200 | in the interaction with the operator directly.

00:38:37.960 | Is there other shell commands that you love over the years,

00:38:43.920 | like you really enjoy using?

00:38:46.320 | - Oh, grep. - Grep?

00:38:47.960 | - Grep's the only one.

00:38:49.760 | Yeah, grep does everything.

00:38:51.440 | - So grep is a kind of a, what is it,

00:38:53.240 | a simpler version of AWK, I would say?

00:38:55.320 | - In some sense, yeah, right, because--

00:38:58.000 | - What is grep?

00:38:58.880 | - So grep is, it basically searches the input

00:39:01.800 | for particular patterns, regular expressions,

00:39:03.960 | technically, of a certain class,

00:39:05.880 | and it has that same paradigm that AWK does,

00:39:08.600 | it's a pattern action thing,

00:39:10.040 | it reads through all the files

00:39:11.280 | and then all the lines in each file,

00:39:13.440 | but it has a single pattern,

00:39:15.200 | which is the regular expression you're looking for,

00:39:17.000 | and a single action printed, if it matches.

00:39:20.240 | So in that sense, it's a much simpler version

00:39:22.560 | and you could write grep and AWK as a one-liner,

00:39:25.560 | and I use grep probably more than anything else

00:39:30.440 | at this point, just because it's so convenient and natural.

00:39:35.040 | - Why do you think, it's such a powerful tool, grep and AWK,

00:39:38.640 | why do you think operating systems like Windows,

00:39:41.280 | for example, don't have it?

00:39:44.800 | Sort of, you can, of course, I use, which is amazing now,

00:39:48.240 | there's Windows for Linux,

00:39:51.560 | which you could basically use all the fun stuff

00:39:54.880 | like AWK and grep inside of Windows,

00:39:57.360 | but Windows naturally,

00:39:58.640 | sort of as part of the graphical interface,

00:40:00.800 | the simplicity of grep,

00:40:02.400 | sort of searching through a bunch of files

00:40:04.760 | and just popping up naturally.

00:40:06.600 | Why don't you think that,

00:40:08.120 | why do you think that's unique

00:40:09.040 | to the Unix and Linux environment?

00:40:11.600 | - I don't know, it's not strictly unique,

00:40:14.000 | but it's certainly focused there,

00:40:16.400 | and I think some of it's the weight of history

00:40:19.040 | that Windows came from MS-DOS,

00:40:22.000 | MS-DOS was a pretty pathetic operating system,

00:40:24.480 | although common on an unboundedly large number of machines,

00:40:29.000 | but somewhere in roughly the '90s,

00:40:32.840 | Windows became a graphical system,

00:40:34.640 | and I think Microsoft spent a lot of their energy

00:40:37.920 | on making that graphical interface what it is,

00:40:41.240 | and that's a different model of computing.

00:40:44.000 | It's a model of computing that,

00:40:45.560 | where you point and click

00:40:47.080 | and sort of experiment with menus,

00:40:49.520 | it's a model of computing works rather well

00:40:52.680 | for people who are not programmers,

00:40:54.480 | who just wanna get something done,

00:40:56.080 | whereas teaching something like the command line

00:40:59.120 | to non-programmers turns out

00:41:00.960 | to sometimes be an uphill struggle,

00:41:02.800 | and so I think Microsoft probably was right

00:41:04.840 | in what they did.

00:41:06.280 | Now you mentioned Whistle or whatever it's called,

00:41:08.440 | the Linux.

00:41:09.280 | - Whistle, I wonder what it's pronounced,

00:41:11.040 | W-S-L is what,

00:41:12.240 | I've never actually pronounced the Whistle, I like it.

00:41:13.920 | - I have no idea.

00:41:14.760 | (laughing)

00:41:15.600 | - Whistle.

00:41:16.440 | - And things like that for a long,

00:41:17.560 | a Cygwin, for example,

00:41:18.760 | which is a wonderful collection

00:41:20.600 | of take all your favorite tools from Unix and Linux

00:41:23.280 | and just make them work perfectly on Windows,

00:41:25.400 | and so that's something that's been going on

00:41:27.280 | for at least 20 years, if not longer,

00:41:29.880 | and I use that on my one remaining Windows machine

00:41:33.000 | routinely because if you're doing something

00:41:37.760 | that is batch computing,

00:41:39.200 | suitable for command line,

00:41:41.720 | that's the right way to do it

00:41:42.680 | because the Windows equivalents are,

00:41:44.480 | if nothing else, not familiar to me.

00:41:46.320 | - But I would definitely recommend to people

00:41:49.880 | to if they don't use Cygwin to try Whistle.

00:41:52.480 | - Yes.

00:41:53.320 | - I've been so excited that I could use BASH,

00:41:57.080 | that you bash, write scripts quickly in Windows,

00:42:00.560 | it's changed my life.

00:42:03.240 | Okay, what's your perfect programming setup?

00:42:06.480 | What computer, what operating system,

00:42:08.320 | what keyboard, what editor?

00:42:10.520 | - Yeah, perfect is too strong a word.

00:42:13.360 | It's way too strong a word.

00:42:15.320 | What I use by default, I have a,

00:42:18.320 | at this point, a 13-inch MacBook Air,

00:42:20.920 | which I use because it's kind of a reasonable balance

00:42:24.240 | of the various things I need.

00:42:25.280 | I can carry it around.

00:42:26.600 | It's got enough computing horsepower,

00:42:28.040 | screen's big enough, keyboard's okay,

00:42:30.200 | and so I basically do most of my computing on that.

00:42:34.640 | I have a big iMac in my office

00:42:37.040 | that I use from time to time as well,

00:42:39.480 | especially when I need a big screen,

00:42:41.040 | but otherwise, it tends not to be used that much.

00:42:44.840 | - Editor.

00:42:48.400 | - I use mostly SAM, which is an editor

00:42:51.200 | that Rob Pike wrote long ago at Bell Labs.

00:42:55.600 | - Did that, sorry to interrupt, does that precede VI?

00:42:58.680 | Does that precede Emacs?

00:43:00.040 | - It post-dates both VI and Emacs.

00:43:03.160 | It is derived from Rob's experience with ED and VI.

00:43:10.040 | - ED.

00:43:10.880 | - That's the original Unix editor.

00:43:14.640 | - Oh, wow.

00:43:15.480 | - Dated probably before you were born.

00:43:18.320 | - So what's, actually, what's the history of editors?

00:43:23.440 | Can you briefly, 'cause it's such a,

00:43:26.680 | I use Emacs, I'm sorry to say.

00:43:28.520 | So sorry to come out with that,

00:43:30.280 | but what's the kind of interplay there?

00:43:33.640 | So SAM, yeah, 'cause VI.

00:43:35.160 | - So in ancient times, like,

00:43:37.400 | call it the first time-sharing systems,

00:43:39.280 | going back to what we were talking about,

00:43:40.840 | there were editors, there was an editor on CTSS

00:43:43.880 | that I don't even remember what it was called.

00:43:46.120 | It might've been edit, where you could type text,

00:43:49.880 | program text, and it would do something,

00:43:52.000 | or document text.

00:43:53.760 | - You could save the text.

00:43:54.600 | - And save it, you could edit it,

00:43:57.240 | usual thing that you would get in an editor.

00:44:00.240 | And Ken Thompson wrote an editor called QED,

00:44:03.060 | which was very, very powerful,

00:44:06.040 | but these were all totally a command-based,

00:44:08.680 | they were not mouse or cursor-based,

00:44:10.760 | because it was before mice, and even before cursors,

00:44:13.800 | 'cause they were running on terminals

00:44:15.040 | that printed on paper, okay?

00:44:17.040 | No CRT-type displays, let alone LEDs.

00:44:20.500 | And so then when Unix came along,

00:44:24.240 | Ken took QED and stripped it way, way, way, way down,

00:44:28.760 | and that became an editor that he called ED.

00:44:31.160 | And it was very simple, but it was a line-oriented editor.

00:44:33.880 | And so you could load a file,

00:44:36.160 | and then you could talk about the lines,

00:44:37.840 | one through the last line,

00:44:39.240 | and you could print ranges of lines,

00:44:41.600 | you could add text, you could delete text,

00:44:44.000 | you could change text,

00:44:44.880 | or you could do a substitute command

00:44:46.480 | that would change things within a line,

00:44:47.980 | or within groups of lines.

00:44:49.280 | - So you could work on parts of a file, essentially.

00:44:51.320 | - Yeah, you could work on any part of it,

00:44:53.000 | the whole thing, or whatever,

00:44:53.920 | but it was entirely command-line-based,

00:44:57.340 | and it was entirely on paper, okay?

00:45:00.840 | Paper, and that meant that you changed--

00:45:02.280 | - Actual paper.

00:45:03.120 | - Yeah, right, real paper.

00:45:04.080 | And so if you changed a line,

00:45:06.120 | you had to print that line,

00:45:07.820 | using up another line of paper

00:45:09.140 | to see what the change caused, okay?

00:45:11.760 | - Yeah, it's amazing.

00:45:12.920 | - So when CRT displays came along,

00:45:17.080 | then you could start to use cursor control,

00:45:19.800 | and you could sort of move where you were on the screen in--

00:45:23.280 | - Without reprinting every time.

00:45:26.160 | - Printing, and there were a number of editors there.

00:45:29.840 | The one that I was most familiar with,

00:45:31.680 | and still use, is VI, which was done by Bill Choi.

00:45:35.200 | And so that dates from probably the late '70s, as a guess,

00:45:40.200 | and it took full advantage of the cursor controls.

00:45:45.240 | I suspected Emacs was roughly at the same time,

00:45:48.340 | but I don't know, I've never internalized Emacs.

00:45:51.040 | So I use, at this point, I stopped using ED,

00:45:54.720 | although I still can.

00:45:56.360 | I use VI sometimes, and I use SAM when I can.

00:46:00.160 | - And SAM is available on most systems?

00:46:02.480 | - It is available.

00:46:04.520 | You have to download it yourself

00:46:05.640 | from typically the plan line operating system distribution.

00:46:08.560 | It's been maintained by people there.

00:46:10.960 | And so--

00:46:12.640 | - I'll get home tonight, I'll try it.

00:46:14.360 | It's cool.

00:46:15.200 | It sounds fascinating.

00:46:17.840 | Although my love is with Lisp and Emacs,

00:46:20.640 | I've went into that hippie world of--

00:46:23.080 | (both laughing)

00:46:25.120 | - I think it's a lot of things.

00:46:26.040 | What religion were you brought up with?

00:46:27.600 | - Yeah, that's true.

00:46:28.720 | That's true.

00:46:29.560 | Most of the actual programming I do

00:46:31.920 | is C, C++, and Python,

00:46:34.080 | but my weird sort of, yeah,

00:46:36.240 | my religious upbringing is in Lisp.

00:46:38.080 | So can you take on the impossible task

00:46:41.760 | and give a brief history of programming languages

00:46:44.760 | from your perspective?

00:46:46.400 | - So I guess you could say programming languages

00:46:48.360 | started probably in what, the late '40s

00:46:51.080 | or something like that.

00:46:51.960 | People used to program computers

00:46:53.760 | by basically putting in zeros and ones

00:46:56.200 | using something like switches on a console.

00:46:59.800 | And then, or maybe holes in paper tapes,

00:47:03.560 | something like that.

00:47:04.920 | So extremely tedious, awful, whatever.

00:47:07.960 | And so I think the first programming languages

00:47:10.240 | were relatively crude assembly languages

00:47:14.560 | where people would basically write a program

00:47:18.360 | that would convert mnemonics like add, A-D-D,

00:47:22.360 | into whatever the bit pattern was

00:47:24.880 | that corresponded to an add instruction.

00:47:26.720 | And they would do the clerical work

00:47:28.200 | of figuring out where things were.

00:47:30.040 | So you could put a name on a location in a program

00:47:32.760 | and the assembler would figure out

00:47:34.920 | where that corresponded to

00:47:36.320 | when the thing was all put together

00:47:37.880 | and dropped into memory.

00:47:39.920 | And early on, and this would be the late '40s

00:47:45.560 | and very early '50s,

00:47:46.680 | there were assemblers written for the various machines

00:47:50.000 | that people used.

00:47:51.000 | You may have seen in the paper just a couple of days ago,

00:47:52.920 | Tony Berker died.

00:47:54.160 | He did this thing in Manchester

00:47:55.880 | called AutoCode, a language which I knew only by name.

00:48:00.880 | But it sounds like it was a flavor of assembly language,

00:48:04.400 | sort of a little higher in some ways.

00:48:06.720 | And it replaced a language that Alan Turing wrote,

00:48:09.080 | which you put in zeros and ones,

00:48:11.120 | but you put it in backwards order

00:48:12.440 | because that was a hardware work.

00:48:14.400 | Very strange. - That's right.

00:48:15.480 | Yeah, yeah, that's right, backwards.

00:48:17.880 | - So assembly languages,

00:48:19.400 | and let's call that the early 1950s.

00:48:22.280 | And so every different flavor of computer

00:48:24.280 | has its own assembly language.

00:48:25.800 | So the EdSac had its, and the Manchester had its,

00:48:28.880 | and the IBM, whatever, 7090 or 704 or whatever had its,

00:48:33.880 | and so on.

00:48:34.720 | So everybody had their own assembly language.

00:48:35.840 | - And assembly languages have a few commands, additions,

00:48:38.920 | subtraction, then branching of some kind,

00:48:41.160 | if then type of situation.

00:48:42.920 | - Right, they have exactly, in their simplest form at least,

00:48:46.720 | one assembly language instruction

00:48:50.040 | per instruction in the machine's repertoire.

00:48:52.840 | And so you have to know the machine intimately

00:48:54.960 | to be able to write programs in it.

00:48:56.720 | And if you write an assembly language program

00:48:58.680 | for one kind of machine, and then you say,

00:49:00.360 | "Gee, it's nice, I'd like it in a different machine,"

00:49:02.880 | start over, okay.

00:49:04.760 | So, very bad.

00:49:05.960 | And so what happened in the late '50s

00:49:08.720 | was people realized you could play this game again,

00:49:11.000 | and you could move up a level in writing

00:49:14.400 | or creating languages that were closer

00:49:16.320 | to the way that real people might think about

00:49:18.400 | how to write code.

00:49:20.760 | And there were, I guess, arguably three or four

00:49:24.080 | at that time period.

00:49:25.560 | There was Fortran, which came from IBM,

00:49:28.040 | which was formula translation,

00:49:29.720 | meant to make it easy to do scientific

00:49:31.640 | and engineering computations.

00:49:32.840 | - I didn't know that, formula translation, wow.

00:49:34.600 | - That's what it stood for.

00:49:35.440 | There was COBOL, which is the Common Business Oriented

00:49:37.840 | Language that Grace Hopper and others worked on,

00:49:40.680 | which was aimed at business kinds of tasks.

00:49:44.240 | There was ALGOL, which was mostly meant

00:49:45.920 | to describe algorithmic computations.

00:49:49.360 | I guess you could argue BASIC was in there somewhere.

00:49:51.320 | I think it's just a little later.

00:49:52.960 | And so all of those moved the level up.

00:49:56.360 | And so they were closer to what you and I might think of

00:49:59.880 | as we were trying to write a program.

00:50:02.560 | And they were focused on different domains,

00:50:05.880 | Fortran for formula translation, engineering computations,

00:50:09.120 | let's say COBOL for business, that kind of thing.

00:50:11.720 | - And still used today, at least Fortran probably.

00:50:14.480 | - Oh yeah, COBOL too.

00:50:16.760 | But the deal was that once you moved up that level,

00:50:19.440 | then you, let's call it Fortran, you had a language

00:50:21.960 | that was not tied to a particular kind of hardware

00:50:25.280 | because a different compiler would compile

00:50:26.840 | for a different kind of hardware.

00:50:27.920 | And that meant two things.

00:50:30.000 | It meant you only had to write the program once,

00:50:32.200 | which was very important.

00:50:33.640 | And it meant that you could, in fact,

00:50:35.920 | if you were a random engineer, physicist, whatever,

00:50:38.440 | you could write that program yourself.

00:50:39.760 | You didn't have to hire a programmer to do it for you.

00:50:42.680 | It might not be as good as you'd get with a real programmer,

00:50:44.520 | but it was pretty good.

00:50:45.840 | And so it democratized and made much more broadly available

00:50:49.720 | the ability to write code.

00:50:51.520 | - So it puts the power of programming

00:50:53.120 | into the hands of people like you.

00:50:54.680 | - Yeah, anybody who wants, who is willing to invest

00:50:57.720 | some time in learning a programming language

00:50:59.440 | and is not then tied to a particular kind of computer.

00:51:03.560 | And then in the '70s, you get system programming languages,

00:51:06.320 | of which C is the survivor.

00:51:07.880 | - What does system programming languages mean?

00:51:11.840 | - Programs that, programming languages

00:51:14.920 | that would take on the kinds of things

00:51:16.560 | that would necessary to write so-called system programs,

00:51:19.360 | things like text editors or assemblers or compilers

00:51:22.720 | or operating systems themselves, those kinds of things.

00:51:26.640 | And Fortran-

00:51:28.360 | - Feature rich, they have to be able to do a lot of stuff,

00:51:30.760 | a lot of memory management, access processes,

00:51:33.680 | all that kind of stuff.

00:51:34.680 | - They have to- - Level of processing.

00:51:35.640 | - It's a different flavor of what they're doing.

00:51:37.200 | They're much more in touch with the actual machine

00:51:41.240 | but in a positive way.

00:51:42.360 | That is, you can talk about memory in a more controlled way.

00:51:45.920 | You can talk about the different data types

00:51:47.840 | that the machine supports and the way they're,

00:51:50.720 | and more ways to structure and organize data.

00:51:54.840 | And so the system programming languages,

00:51:57.360 | there was a lot of effort in that in the,

00:51:59.760 | call it the late '60s, early '70s.

00:52:02.160 | C is, I think, the only real survivor of that.

00:52:05.120 | And then what happens after that,

00:52:08.080 | you get things like object-oriented programming languages

00:52:12.120 | because as you write programs in a language like C,

00:52:14.920 | at some point, scale gets to you

00:52:16.560 | and it's too hard to keep track of the pieces

00:52:18.480 | and there's no guardrails or training wheels

00:52:21.080 | or something like that to prevent you from doing bad things.

00:52:24.400 | So C++ comes out of that tradition.

00:52:28.040 | - And then it took off from there.

00:52:29.600 | I mean, there's also a parallel, slightly parallel track

00:52:32.200 | with a little bit of the functional stuff

00:52:33.760 | with Lisp and so on.

00:52:35.160 | But I guess from that point,

00:52:37.000 | it's just an explosion of languages.

00:52:38.680 | There's the Java story, there's the JavaScript,

00:52:41.880 | there's all the stuff that the cool kids these days

00:52:44.960 | are doing with Rust and all that.

00:52:46.640 | So what's to you, so you wrote a book,

00:52:51.040 | C programming language,

00:52:53.120 | and C is probably one of the most important languages

00:52:56.920 | in the history of programming languages

00:52:58.840 | if you kind of look at impact.

00:53:01.080 | What do you think is the most elegant

00:53:02.960 | or powerful part of C?

00:53:06.240 | Why did it survive?

00:53:07.560 | Why did it have such a long-lasting impact?

00:53:11.320 | - I think it found a sweet spot of expressiveness,

00:53:16.320 | that you could really write things

00:53:17.680 | in a pretty natural way, and efficiency,

00:53:20.360 | which was particularly important

00:53:22.200 | when computers were not nearly as powerful

00:53:24.360 | as they are today.

00:53:25.200 | You're gonna put yourself back 50 years,

00:53:28.000 | almost in terms of what computers could do,

00:53:31.240 | and that's roughly four or five generations,

00:53:35.040 | decades of Moore's law, right?

00:53:37.520 | So expressiveness and efficiency,

00:53:40.840 | and I don't know, perhaps the environment

00:53:43.840 | that it came with as well, which was Unix.

00:53:46.360 | So it meant if you wrote a program,

00:53:47.880 | it could be used on all those computers that ran Unix,

00:53:50.520 | and that was all of those computers

00:53:51.960 | because they were all written in C,

00:53:53.400 | and that way Unix, the operating system itself,

00:53:56.520 | was portable, as were all the tools.

00:53:58.640 | So it all worked together, again,

00:54:00.680 | in one of these things where things fed on each other

00:54:03.640 | in a positive cycle.

00:54:05.960 | - What did it take to write sort of a definitive book,

00:54:10.000 | probably definitive book on all of program,

00:54:11.960 | like it's more definitive to a particular language

00:54:14.480 | than any other book on any other language,

00:54:16.640 | and did two really powerful things,

00:54:19.000 | which is popularized the language,

00:54:22.720 | at least from my perspective, maybe you can correct me,

00:54:24.880 | and second is created a standard

00:54:27.400 | of how this language is supposed to be used and applied.

00:54:33.640 | So what did it take?

00:54:34.960 | Did you have those kinds of ambitions in mind

00:54:37.400 | when working on that?

00:54:38.240 | - Is this some kind of joke?

00:54:39.600 | (laughing)

00:54:41.400 | No, of course not.

00:54:42.840 | - So it's an accident of timing, skill, and just luck.

00:54:47.840 | - A lot of it is, clearly, timing was good.

00:54:51.520 | Now, Dennis and I wrote the book in 1977.

00:54:54.000 | - Dennis Ritchie. - Yeah, right.

00:54:56.440 | And at that point, Unix was starting to spread.

00:54:58.920 | I don't know how many there were,

00:55:00.040 | but it would be dozens to hundreds of Unix systems,

00:55:03.320 | and C was also available on other kinds of computers

00:55:06.680 | that had nothing to do with Unix,

00:55:08.320 | and so the language had some potential.

00:55:11.440 | And there were no other books on C,

00:55:17.720 | and Bell Labs was really the only source for it,

00:55:20.360 | and Dennis, of course, was authoritative

00:55:22.560 | because it was his language,

00:55:23.920 | and he had written the reference manual,

00:55:26.800 | which is a marvelous example

00:55:28.040 | of how to write a reference manual.

00:55:29.480 | Really, really very, very well done.

00:55:31.480 | So I twisted his arm until he agreed to write a book,

00:55:34.240 | and then we wrote a book.

00:55:35.440 | And the virtue, or advantage, at least,

00:55:38.560 | I guess, of going first is that then other people

00:55:40.840 | have to follow you if they're gonna do anything.

00:55:43.280 | And I think it worked well

00:55:46.800 | because Dennis was a superb writer.

00:55:50.400 | I mean, he really, really did,

00:55:51.600 | and the reference manual in that book is his, period.

00:55:55.080 | I had nothing to do with that at all.

00:55:58.760 | So just crystal clear prose, very, very well expressed.

00:56:02.720 | And then he and I,

00:56:03.880 | I wrote most of the expository material,

00:56:07.720 | and then he and I sort of did the usual

00:56:09.760 | ping-ponging back and forth, refining it.

00:56:13.400 | But I spent a lot of time trying to find examples

00:56:15.600 | that would sort of hang together

00:56:16.840 | and that would tell people what they might need to know

00:56:18.960 | at about the right time,

00:56:20.160 | that they should be thinking about needing it.

00:56:22.460 | And I'm not sure it completely succeeded,

00:56:25.560 | but it mostly worked out fairly well.

00:56:28.520 | - What do you think is the power of example?

00:56:30.160 | I mean, you're the creator,

00:56:34.280 | at least one of the first people

00:56:36.000 | to do the Hello World program,

00:56:38.360 | which is like the example.

00:56:40.440 | If aliens discover our civilization

00:56:43.080 | hundreds of years from now,

00:56:44.200 | it'll probably be Hello World programs,

00:56:46.840 | just like a half-broken robot

00:56:48.600 | communicating with them with a Hello World.

00:56:50.800 | So what, and that's a representative example,

00:56:53.440 | so what do you find powerful about examples?

00:56:57.080 | - I think a good example will tell you how to do something

00:57:01.560 | and it will be representative of,

00:57:03.880 | you might not wanna do exactly that,

00:57:05.720 | but you will wanna do something

00:57:06.960 | that's at least in that same general vein.

00:57:09.740 | And so a lot of the examples in the C book

00:57:13.400 | were picked for these very, very simple,

00:57:15.720 | straightforward text processing problems

00:57:17.700 | that were typical of Unix.

00:57:19.760 | I want to read input and write it out again.

00:57:23.600 | There's a copy command.

00:57:24.600 | I wanna read input and do something to it

00:57:27.080 | and write it out again.

00:57:28.000 | There's a grab.

00:57:28.840 | And so that kind of find things

00:57:31.360 | that are representative of what people want to do

00:57:35.000 | and spell those out so that they can then take those

00:57:39.400 | and see the core parts and modify them to their taste.

00:57:44.400 | And I think that a lot of programming books that,

00:57:48.800 | I don't look at programming books

00:57:51.160 | a tremendous amount these days,

00:57:52.320 | but when I do, a lot of them don't do that.

00:57:54.480 | They don't give you examples that are both realistic

00:57:59.040 | and something you might want to do.

00:58:01.880 | Some of them are pure syntax.

00:58:03.800 | Here's how you add three numbers.

00:58:05.320 | Well, come on, I could figure that out.

00:58:07.300 | Tell me how I would get those three numbers

00:58:09.200 | into the computer and how we would do something useful

00:58:11.920 | with them and then how I put them back out again,

00:58:14.320 | neatly formatted.

00:58:15.560 | - And especially if you follow that example,

00:58:17.180 | there is something magical of doing something

00:58:19.440 | that feels useful.

00:58:21.020 | - Yeah, right.

00:58:21.860 | And I think it's the attempt,

00:58:23.560 | and it's absolutely not perfect,

00:58:25.560 | but the attempt in all cases was to get something

00:58:28.800 | that was going to be either directly useful

00:58:31.560 | or would be very representative of useful things

00:58:35.600 | that a programmer might want to do.

00:58:37.960 | But within that vein of fundamentally text processing,

00:58:41.120 | reading text, doing something, writing text.

00:58:43.680 | - So you've also written a book on Go language.

00:58:47.420 | I have to admit, so I worked at Google for a while

00:58:50.960 | and I've never used Go.

00:58:53.640 | - Well, you missed something.

00:58:54.680 | - Well, I know I missed something for sure.

00:58:56.760 | So Go and Rust are two languages that I hear very,

00:59:01.660 | spoken very highly of and I wish I would like to,

00:59:05.840 | well, there's a lot of them.

00:59:06.840 | There's Julia, there's all these incredible modern languages

00:59:10.760 | but if you can comment before,

00:59:12.680 | or maybe comment on what do you find,

00:59:16.280 | where does Go sit in this broad spectrum of languages?

00:59:19.640 | And also how do you yourself feel about this wide range

00:59:23.960 | of powerful, interesting languages

00:59:26.480 | that you may never even get to try to explore

00:59:30.480 | because of time?

00:59:31.520 | - So I think, so Go first comes from that same

00:59:36.520 | Bell Labs tradition in part, not exclusively,

00:59:39.240 | but two of the three creators, Ken Thompson and Rob Pike.

00:59:42.480 | - So literally the people.

00:59:44.000 | - Yeah, the people.

00:59:45.560 | And then with this very, very useful influence

00:59:49.040 | from the European school in particular,

00:59:51.840 | the Klaus Spirth influence through Robert Griesemer

00:59:55.560 | who was, I guess, a second generation down student at ETH.

01:00:00.560 | And so that's an interesting combination of things.

01:00:03.200 | And so some ways Go captures the good parts of C,

01:00:08.200 | it looks sort of like C, it's sometimes characterized

01:00:11.040 | as C for the 21st century.

01:00:12.940 | On the surface, it looks very, very much like C.

01:00:17.540 | But at the same time, it has some interesting

01:00:20.000 | data structuring capabilities.

01:00:21.840 | And then I think the part that I would say

01:00:25.200 | is particularly useful, and again, I'm not a Go expert.

01:00:29.640 | In spite of co-authoring the book,

01:00:31.800 | about 90% of the work was done by Alan Donovan,

01:00:34.840 | my co-author who is a Go expert.

01:00:36.860 | But Go provides a very nice model of concurrency.

01:00:42.440 | It's basically the cooperating,

01:00:44.780 | communicating sequential processes that Tony Hoare

01:00:47.400 | set forth, geez, I don't know, 40 plus years ago.

01:00:51.340 | And Go routines are, to my mind, a very natural way

01:00:56.000 | to talk about parallel computation.

01:00:59.400 | And in the few experiments I've done with them,

01:01:01.800 | they're easy to write and typically it's gonna work

01:01:04.800 | and very efficient as well.

01:01:07.240 | So I think that's one place where Go stands out,

01:01:09.960 | that that model of parallel computation

01:01:13.080 | is very, very easy and nice to work with.

01:01:16.140 | - Just to comment on that, do you think C foresaw,

01:01:19.500 | or the early Unix days foresaw threads

01:01:22.740 | and massively parallel computation?

01:01:26.060 | - I would guess not really.

01:01:27.700 | I mean, maybe it was seen, but not at the level

01:01:30.440 | where it was something you had to do anything about.

01:01:33.380 | For a long time, processors got faster.

01:01:38.180 | And then processors stopped getting faster

01:01:41.340 | because of things like power consumption

01:01:43.880 | and heat generation.

01:01:46.200 | And so what happened instead was that instead

01:01:49.160 | of processors getting faster,

01:01:50.520 | there started to be more of them.

01:01:52.600 | And that's where that parallel thread stuff comes in.

01:01:55.680 | - So if you can comment on all the other languages,

01:02:01.040 | does it break your heart that you'll never get

01:02:03.040 | to explore them?

01:02:04.480 | How do you feel about the full variety?

01:02:07.200 | - It's not break my heart, but I would love

01:02:10.080 | to be able to try more of these languages.

01:02:12.980 | The closest I've come is in a class

01:02:14.880 | that I often teach in the spring here.

01:02:17.020 | It's a programming class.

01:02:18.580 | And I often give, I have one sort of small example

01:02:23.580 | that I will write in as many languages as I possibly can.

01:02:27.320 | I've got it in 20 odd languages at this point.

01:02:30.220 | And that's so I do a minimal experiment with a language,

01:02:35.220 | just to say, okay, I have this trivial task,

01:02:37.340 | which I understand the task, and it should,

01:02:39.340 | it takes 15 lines in awk and not much more

01:02:42.800 | in a variety of other languages.

01:02:44.400 | So how big is it?

01:02:45.360 | How fast does it run?

01:02:46.420 | And what pain did I go through to learn how to do it?

01:02:49.880 | And that's like anic data, right?

01:02:55.440 | It's very, very, very narrowly focused.

01:02:58.180 | - Anic data, I like that term.

01:02:59.420 | So yeah, but still, it's a little sample

01:03:01.920 | 'cause you get the, I think the hardest step

01:03:04.000 | of the programming language is probably the first step.

01:03:06.340 | Right, so there you're taking the first step.

01:03:08.760 | - Yeah, and so my experience with some languages

01:03:13.400 | is very positive, like Lua,

01:03:14.880 | a scripting language I had never used.

01:03:17.540 | And I took my little program.

01:03:19.740 | The program is a trivial formatter.

01:03:21.660 | It just takes in lines of text of varying lengths,

01:03:24.620 | and it puts them out in lines

01:03:26.460 | that have no more than 60 characters on each line.

01:03:28.900 | So think of it as just kind of the flow of process

01:03:31.900 | in a browser or something.

01:03:34.460 | So it's a very short program.

01:03:36.240 | And in Lua, I downloaded Lua,

01:03:39.300 | and in an hour I had it working,

01:03:41.020 | never having written Lua in my life,

01:03:43.140 | just going with online documentation.

01:03:44.900 | I did the same thing in Scala,

01:03:46.140 | which you can think of as a flavor of Java,

01:03:48.900 | equally trivial.

01:03:50.060 | I did it in Haskell.

01:03:52.100 | It took me several weeks.

01:03:53.560 | But it did run like a turtle.

01:03:57.820 | And I did it in Fortran 90,

01:04:03.900 | and it was painful, but it worked.

01:04:06.100 | And I tried it in Rust,

01:04:07.980 | and it took me several days to get it working

01:04:10.200 | because the model of memory management

01:04:12.140 | was just a little unfamiliar to me.

01:04:13.820 | And the problem I had with Rust,

01:04:15.940 | and it's back to what we were just talking about,

01:04:18.280 | I couldn't find good, consistent documentation on Rust.

01:04:21.520 | Now, this was several years ago,

01:04:22.740 | and I'm sure things have stabilized,

01:04:24.180 | but at the time, everything in the Rust world

01:04:26.500 | seemed to be changing rapidly.

01:04:27.900 | And so you would find what looked like a working example,

01:04:30.500 | and it wouldn't work with the version

01:04:32.220 | of the language that I had.

01:04:34.820 | So it took longer than it should have.

01:04:37.540 | Rust is a language I would like to get back to,

01:04:39.620 | but probably won't.

01:04:41.180 | I think one of the issues,

01:04:42.060 | you have to have something you want to do.

01:04:44.060 | If you don't have something that is the right combination,

01:04:47.620 | if I want to do it,

01:04:48.700 | and yet I have enough disposable time, whatever,

01:04:53.520 | to make it worth learning a new language at the same time,

01:04:56.460 | it's never gonna happen.

01:04:58.140 | - So what do you think about another language of JavaScript

01:05:02.100 | that's this, well, let me just sort of comment on this.

01:05:06.860 | When I was brought up,

01:05:08.140 | sort of JavaScript was seen as

01:05:09.780 | the probably like the ugliest language possible,

01:05:15.580 | and yet it's quite arguably, quite possibly,

01:05:18.240 | taking over not just the front end

01:05:20.020 | and the back end of the internet,

01:05:21.620 | but possibly in the future taking over everything,

01:05:24.000 | because they've now learned to make it very efficient.

01:05:27.100 | - Yeah.

01:05:27.940 | - And so what do you think about this?

01:05:29.780 | - Yeah, well, I think you've captured it in a lot of ways.

01:05:32.180 | When it first came out,

01:05:33.020 | JavaScript was deemed to be fairly irregular

01:05:35.500 | and an ugly language,

01:05:36.540 | and certainly in the academy,

01:05:37.820 | if you said you were working on JavaScript,

01:05:39.300 | people would ridicule you.

01:05:40.500 | It was just not fit for academics to work on.

01:05:43.820 | I think a lot of that has evolved.

01:05:45.500 | The language itself has evolved,

01:05:47.580 | and certainly the technology of compiling it

01:05:50.700 | is fantastically better than it was.

01:05:53.700 | And so in that sense,

01:05:54.820 | it's absolutely a viable solution on back ends,

01:05:58.860 | as well as the front ends.

01:06:00.260 | Used well, I think it's a pretty good language.

01:06:03.480 | I've written a modest amount of it,

01:06:06.340 | and I've played with JavaScript translators

01:06:09.140 | and things like that.

01:06:10.300 | I'm not a real expert,

01:06:12.000 | and it's hard to keep up even there

01:06:13.660 | with the new things that come along with it.

01:06:15.860 | So I don't know whether it will ever take over the world.

01:06:21.160 | I think not,

01:06:22.580 | but it's certainly an important language

01:06:26.620 | and worth knowing more about.

01:06:29.060 | - Maybe to get your comment on something,

01:06:32.420 | which JavaScript, and actually most languages,

01:06:35.340 | sort of Python,

01:06:36.780 | such a big part of the experience of programming

01:06:40.140 | with those languages includes libraries.

01:06:42.700 | So using, building on top of the code

01:06:44.540 | that other people have built.

01:06:46.060 | I think that's probably different from the experience

01:06:48.060 | that we just talked about from Unix and C days

01:06:51.900 | when you're building stuff from scratch.

01:06:53.260 | What do you think about this world

01:06:55.220 | of essentially leveraging,

01:06:56.660 | building up libraries on top of each other

01:06:58.460 | and leveraging them?

01:06:59.460 | - Yeah, that's a very perceptive kind of question.

01:07:02.860 | One of the reasons programming was fun in the old days

01:07:06.260 | was that you were really building it all yourself.

01:07:09.020 | The number of libraries you had to deal with

01:07:10.980 | was quite small.

01:07:11.820 | Maybe it was printf or the standard library

01:07:14.060 | or something like that.

01:07:16.020 | And that is not the case today.

01:07:17.900 | And if you want to do something in,

01:07:20.220 | you mentioned Python and JavaScript,

01:07:22.660 | and those are the two fine examples

01:07:24.440 | you have to typically download a boatload of other stuff

01:07:27.980 | and you have no idea what you're getting.

01:07:29.820 | Absolutely nothing.

01:07:31.380 | I've been doing some playing with machine learning

01:07:33.660 | over the last couple of days and gee,

01:07:37.700 | something doesn't work.

01:07:38.620 | Well, you pip install this, okay?

01:07:40.940 | And down comes another gazillion megabytes of something

01:07:44.460 | and you have no idea what it was.

01:07:46.420 | And if you're lucky it works.

01:07:48.060 | And if it doesn't work, you have no recourse.

01:07:51.300 | There's absolutely no way you could figure out

01:07:52.880 | which in these thousand different packages.

01:07:55.220 | And I think it's worse in the NPM environment

01:07:59.660 | for JavaScript.

01:08:00.500 | I think there's less discipline, less control there.

01:08:03.100 | - And there's aspects of not just not understanding

01:08:06.220 | how it works, but there's security issues,

01:08:08.020 | there's robustness issues.

01:08:09.140 | So you don't want to run a nuclear power plant

01:08:11.780 | using JavaScript essentially.

01:08:14.100 | - Probably not.

01:08:14.940 | - So speaking to the variety of languages,

01:08:18.860 | do you think that variety is good?

01:08:20.460 | Or do you hope, think that over time

01:08:23.580 | we should converge towards one, two, three

01:08:25.760 | programming languages?

01:08:27.440 | That you mentioned to the Bell a lot of days

01:08:29.720 | when people could sort of, the community of it.

01:08:32.920 | And the more languages you have,

01:08:34.520 | the more you separate the communities.

01:08:36.800 | There's the Ruby community, there's the Python community,

01:08:40.300 | there's C++ community.

01:08:42.700 | Do you hope that they'll unite one day

01:08:45.460 | to just one or two languages?

01:08:47.720 | - I certainly don't hope it.

01:08:48.840 | I'm not sure that that's right

01:08:50.000 | 'cause I honestly don't think there is one language

01:08:51.960 | that will suffice for all the programming needs

01:08:54.240 | of the world.

01:08:55.360 | Are there too many at this point?

01:08:56.880 | Well, arguably, but I think if you look at the,

01:09:00.000 | that sort of the distribution of how they are used,

01:09:03.160 | there's something called a dozen languages

01:09:06.780 | that probably account for 95% of all programming

01:09:10.520 | at this point.

01:09:11.360 | And that doesn't seem unreasonable.

01:09:13.600 | And then there's another, well, 2000 languages

01:09:17.240 | that are still in use that nobody uses.

01:09:19.960 | And, or at least don't use in any quantity.

01:09:23.320 | But I think new languages are a good idea in many respects

01:09:25.920 | 'cause they're often a chance to explore an idea

01:09:30.240 | of how a language might help.

01:09:32.920 | I think that's one of the positive things

01:09:35.240 | about functional languages.

01:09:36.480 | For example, they're a particularly good place

01:09:38.660 | where people have explored ideas

01:09:42.480 | that at the time didn't seem feasible,

01:09:45.720 | but ultimately have wound up

01:09:47.480 | as part of mainstream languages as well.

01:09:50.160 | I mean, you just go back as early as recursion Lisp

01:09:52.720 | and then follow forward.

01:09:54.960 | Functions as first-class citizens

01:09:57.120 | and pattern-based languages, and gee, I don't know,

01:10:01.480 | closures and just on and on and on.

01:10:04.200 | Lambda's interesting ideas that showed up first

01:10:07.000 | in let's call it broadly

01:10:08.840 | the functional programming community

01:10:10.680 | and then find their way into mainstream languages.

01:10:13.320 | - Yeah, it's a playground for rebels.

01:10:15.600 | - Yeah, exactly.

01:10:16.440 | (Lex laughing)

01:10:18.000 | And so I think the languages in the playground themselves

01:10:21.240 | are probably not going to be the mainstream,

01:10:24.840 | at least for some while,

01:10:25.880 | but the ideas that come from there are invaluable.

01:10:28.540 | - So let's go to something that when I found out recently,

01:10:33.880 | so I've known that you've done a million things,

01:10:36.240 | but one of the things I wasn't aware of

01:10:37.720 | that you had a role in AMPL,

01:10:39.680 | and before you interrupt me by minimizing your role in it.

01:10:44.880 | - AMPL is for minimizing functions.

01:10:46.520 | - Yeah, minimizing functions, right, exactly.

01:10:48.720 | Can I just say that the elegance

01:10:52.360 | and abstraction power of AMPL is incredible?

01:10:57.360 | When I first came to it about 10 years ago or so,

01:11:01.360 | can you describe what is the AMPL language?

01:11:04.280 | - Sure, so AMPL is a language for mathematical programming,

01:11:08.200 | technical term, think of it as linear programming,

01:11:10.760 | that is setting up systems of linear equations

01:11:14.720 | that are of some sort of system of constraints,

01:11:18.800 | so that you have a bunch of things

01:11:20.600 | that have to be less than this, greater than that,

01:11:22.600 | or whatever, and you're trying to find a set of values

01:11:25.640 | for some decision variables that will maximize

01:11:29.600 | or minimize some objective function.

01:11:32.200 | So it's a way of solving a particular kind

01:11:35.880 | of optimization problem,

01:11:38.000 | a very formal sort of optimization problem,

01:11:40.040 | but one that's exceptionally useful.

01:11:42.560 | - And it specifies, so there's objective function

01:11:45.240 | constraints and variables that become separate

01:11:48.160 | from the data it operates on.

01:11:50.080 | - Right.

01:11:50.920 | - So that kind of separation allows you

01:11:54.040 | to put on different hats.

01:11:58.000 | One put the hat of an optimization person

01:12:00.360 | and then put another hat of a data person

01:12:03.280 | and dance back and forth,

01:12:04.360 | and also separate the actual solvers,

01:12:08.800 | the optimization systems that do the solving.

01:12:12.000 | Then you can have other people come to the table

01:12:14.240 | and then build their solvers,

01:12:15.480 | whether it's linear or non-linear,

01:12:17.400 | convex, non-convex, that kind of stuff.

01:12:21.800 | So what is the, to you as,

01:12:26.800 | maybe you can comment how you got into that world

01:12:30.160 | and what is the beautiful or interesting idea

01:12:33.480 | to you from the world of optimization?

01:12:35.400 | - Sure.

01:12:36.240 | So I preface it by saying I'm absolutely not an expert

01:12:39.800 | on this and most of the important work in AMPL

01:12:42.960 | comes from my two partners in crime on that,

01:12:45.360 | Bob Forer, who was a professor of,

01:12:48.440 | and in the industrial engineering and management

01:12:50.560 | science department at Northwestern,

01:12:52.480 | and my colleague at Bell Labs, Dave Gay,

01:12:54.800 | who was a numerical analyst and optimization person.

01:12:59.000 | So the deal is linear programming,

01:13:01.040 | preface this by saying.

01:13:03.800 | - Let's stay with linear programming.

01:13:05.120 | - Yeah, linear programming is the simplest example of this.

01:13:07.600 | So linear programming as it's taught in school

01:13:09.700 | is that you have a big matrix, which is always called A,

01:13:12.240 | and you say AX is less than or equal to B.

01:13:14.880 | So B is a set of constraints, X is the decision variables,

01:13:18.560 | and A is how the decision variables are combined

01:13:22.960 | to set up the various constraints.

01:13:24.480 | So A is a matrix and X and B are vectors.

01:13:28.160 | And then there's an objective function,

01:13:30.000 | which is just a sum of a bunch of Xs

01:13:31.960 | and some coefficients on them.

01:13:33.560 | And yet that's the thing you wanna optimize.

01:13:37.140 | The problem is that in the real world,

01:13:39.980 | that matrix A is a very, very, very intricate,

01:13:43.420 | very large and very sparse matrix

01:13:45.520 | where the various components of the model

01:13:47.820 | are distributed among the coefficients

01:13:50.540 | in a way that is totally unobvious to anybody.

01:13:53.820 | And so what you need is some way to express

01:13:58.860 | the original model, which you and I would write,

01:14:01.060 | you know, we'd write mathematics on the board,

01:14:03.420 | the sum of this is greater than the sum

01:14:04.980 | of that kind of thing.

01:14:06.420 | So you need a language to write those kinds of constraints.

01:14:10.260 | And Bob Forer for a long time had been interested

01:14:13.160 | in modeling languages,

01:14:14.380 | languages that made it possible to do this.

01:14:16.500 | There was a modeling language around called GAMS,

01:14:19.020 | the General Algebraic Modeling System,

01:14:21.260 | but it looked very much like Fortran, it was kind of clunky.

01:14:24.660 | And so Bob spent a sabbatical year at Bell Labs in 1984,

01:14:29.180 | and he and, he was in the office across from me,

01:14:32.780 | and it's always geography.

01:14:35.380 | And he and Dave Gay and I started talking

01:14:38.020 | about this kind of thing.

01:14:39.660 | And he wanted to design a language that would make it

01:14:43.760 | so that you could take these algebraic specifications,

01:14:46.480 | you know, summation, sines, oversets,

01:14:48.780 | and that you would write on the board

01:14:51.060 | and convert them into basically this A matrix,

01:14:55.820 | and then pass that off to a solver,

01:14:58.880 | which is an entirely separate thing.

01:15:00.740 | And so we talked about the design of the language.

01:15:05.140 | I don't remember any of the details of this now,

01:15:07.180 | but it's kind of an obvious thing.

01:15:08.940 | You're just writing mathematical expressions

01:15:11.220 | in a Fortran-like, or sorry,

01:15:13.140 | an algebraic but textual-like language.

01:15:15.820 | And I wrote the first version of this AMPL program,

01:15:21.860 | my first C++ program.

01:15:24.260 | (Lex laughs)

01:15:25.100 | And-

01:15:26.180 | - That's written in C++?

01:15:27.420 | - Yeah. - Oh, wow.

01:15:28.620 | - And so I did that fairly quickly.

01:15:30.980 | We wrote, it was, you know, 3,000 lines or something,

01:15:33.460 | so it wasn't very big,

01:15:34.340 | but it sort of showed the feasibility of it,

01:15:36.520 | that you could actually do something

01:15:37.760 | that was easy for people to specify models

01:15:41.740 | and convert it into something that a solver could work with.

01:15:44.700 | At the same time, as you say,

01:15:45.860 | the model and the data are separate things.

01:15:47.860 | So one model would then work

01:15:50.060 | with all kinds of different data

01:15:51.260 | in the same way that lots of programs do the same thing,

01:15:53.500 | but with different data.

01:15:54.420 | - So one of the really nice things

01:15:55.660 | is the specification of the models,

01:15:58.460 | human, just kind of like, as you say,

01:16:00.700 | it's human-readable.

01:16:01.980 | Like I literally, I remember on stuff I worked,

01:16:04.900 | I would send it to colleagues

01:16:07.620 | that I'm pretty sure never programmed in their life,

01:16:10.780 | just to understand what the optimization problem is.

01:16:15.780 | I think, how hard is it to convert that?

01:16:18.060 | You said there's a first prototype in C++,

01:16:20.300 | to convert that into something

01:16:22.060 | that could actually be used by the solver.

01:16:24.300 | - It's not too bad, because most of the solvers

01:16:26.300 | have some mechanism that lets them import a model in a form.

01:16:30.460 | It might be as simple as the matrix itself

01:16:32.980 | in just some representation,

01:16:35.020 | or if you're doing things that are not linear programming,

01:16:38.420 | then there may be some mechanism

01:16:39.820 | that lets you provide things like functions to be called,

01:16:43.420 | or other constraints on the model.

01:16:47.140 | So all Ample does is to generate that kind of thing,

01:16:51.500 | and then solver deals with all the hard work.

01:16:54.220 | And then when the solver comes back with numbers,

01:16:57.380 | Ample converts those back into your original form.

01:17:00.220 | So you know how much of each thing you should be buying,

01:17:03.180 | or making, or shipping, or whatever.

01:17:05.160 | So we did that in '84,

01:17:08.620 | and I haven't had a lot to do with it since,

01:17:11.900 | except that we wrote a couple of versions of a book on it.

01:17:14.300 | - Which is one of the greatest books ever written.

01:17:16.560 | I love that book.

01:17:17.720 | I don't know why.

01:17:20.020 | - It's an excellent book.

01:17:21.020 | Bob Forer wrote most of it.

01:17:22.540 | And so it's really, really well done.

01:17:24.020 | He must have been a dynamite teacher.

01:17:25.700 | - And typeset in LaTeX.

01:17:27.560 | - No, no, no, are you kidding?

01:17:29.220 | (laughing)

01:17:30.340 | I remember liking the typography, so I don't know.

01:17:32.940 | - We did it with T-Rof.

01:17:34.500 | - I don't even know what that is.

01:17:35.460 | - Yeah, exactly, you're too young.

01:17:37.180 | - Oh boy.

01:17:38.380 | - I think of T-Rof as a predecessor

01:17:42.180 | to the tech family of things.

01:17:44.260 | It's a formatter that was done at Bell Labs

01:17:46.180 | in this same period of the very early '70s

01:17:48.860 | that predates tech and things like that

01:17:52.460 | by five to 10 years.

01:17:54.940 | - But it was nevertheless, I'm going by memories.

01:17:58.220 | I remember it being beautiful.

01:18:00.060 | - Yeah, it was nicely done.

01:18:01.780 | - Outside of Unix, C, Ogg, Golang,

01:18:03.820 | all the things we talked about,

01:18:05.780 | all the amazing work you've done,

01:18:07.940 | you've also done work in graph theory.

01:18:09.880 | Let me ask this crazy out there question.

01:18:16.480 | If you had to make a bet,

01:18:17.660 | and I had to force you to make a bet,

01:18:19.180 | do you think P equals NP?

01:18:20.900 | (laughing)

01:18:23.500 | - The answer is no, although I'm told that somebody asked

01:18:26.300 | Jeff Dean if that was, under what conditions

01:18:28.980 | P would equal NP, and he said either P is zero or N is one.

01:18:32.300 | Or vice versa, I've forgotten.

01:18:35.660 | This is why Jeff Dean is a lot smarter than I am.

01:18:38.060 | - Yeah.

01:18:38.900 | So, but your intuition is--

01:18:41.660 | - I have no intuition, but I've got a lot of colleagues

01:18:45.940 | who've got intuition, and their betting is no.

01:18:48.140 | - That's the popular bet.

01:18:51.180 | Okay, so what is computational complexity theory,

01:18:55.620 | and do you think these kinds of complexity classes,

01:18:58.260 | especially as you've taught in this modern world,

01:19:01.540 | are still a useful way to understand

01:19:04.260 | the hardness of problems?

01:19:06.060 | - I don't do that stuff.

01:19:07.420 | The last time I touched anything to do with that--

01:19:09.420 | - Many, many years ago.

01:19:10.300 | - Was before it was invented.

01:19:12.420 | 'Cause I, it's literally true.

01:19:14.660 | I did my PhD thesis on graph--

01:19:17.700 | - Before big O notation.

01:19:18.900 | - Oh, absolutely, before, I did this in 1968,

01:19:24.060 | and I worked on graph partitioning,

01:19:25.980 | which is this question, you've got a graph

01:19:27.820 | that is a nodes and edges kind of graph,

01:19:30.300 | and the edges have weights,

01:19:31.660 | and you just wanna divide the nodes into two piles

01:19:34.460 | of equal size, so that the number of edges

01:19:36.820 | that goes from one side to the other

01:19:38.060 | is as small as possible.

01:19:39.420 | And we--

01:19:41.580 | - You developed, so that problem is hard?

01:19:45.940 | - Well, as it turns out, I worked with Shen Lin

01:19:48.420 | at Bell Labs on this, and we were never able to come up

01:19:52.140 | with anything that was guaranteed to give the right answer.

01:19:54.220 | We came up with heuristics that worked pretty darn well,

01:19:57.940 | and I peeled off some special cases for my thesis,

01:20:01.060 | but it was just hard.

01:20:02.260 | And that was just about the time that Steve Cook

01:20:04.700 | was showing that there were classes of problems

01:20:06.500 | that appeared to be really hard,

01:20:08.140 | of which graph partitioning was one.

01:20:10.720 | But this, my expertise, such as it was,

01:20:13.780 | totally predates that development.

01:20:16.500 | - Oh, interesting, so the heuristic,

01:20:18.100 | which now, it carries the two of yours names

01:20:21.980 | for the traveling salesman problem,

01:20:23.740 | and for the graph partitioning, that was,

01:20:26.380 | like, how did you, you weren't even thinking

01:20:28.300 | in terms of classes, you were just trying to find--

01:20:29.980 | - There was no such idea.

01:20:31.140 | - A heuristic that kinda does the job pretty well.

01:20:34.460 | - You were trying to find something that did the job,

01:20:36.820 | and there was nothing that you would call,

01:20:38.700 | let's say, a closed form or algorithmic thing

01:20:41.740 | that would give you a guaranteed right answer.

01:20:44.340 | I mean, compare graph partitioning to max flow min cut,

01:20:48.300 | or something like that.

01:20:50.180 | That's the same problem, except there's no constraint

01:20:52.860 | on the number of nodes on one side or the other of the cut.

01:20:56.260 | And that means it's an easy problem,

01:20:58.700 | at least as I understand it.

01:20:59.980 | Whereas the constraint that says

01:21:01.500 | the two have to be constrained in size

01:21:03.500 | makes it a hard problem.

01:21:05.500 | - Yeah, so Robert Frost has that poem

01:21:07.580 | where you have to choose two paths.

01:21:09.260 | So why did you, is there another alternate universe

01:21:13.620 | in which you pursued the Don Knuth path

01:21:16.600 | of algorithm design, sort of--

01:21:19.860 | - Not smart enough.

01:21:21.380 | - Not smart enough.

01:21:22.380 | You're infinitely modest,

01:21:27.300 | but so you pursued your kind of love of programming.

01:21:30.280 | I mean, when you look back to those,

01:21:33.620 | I mean, just looking into that world,

01:21:35.300 | does that just seem like a distant world

01:21:37.820 | of theoretical computer science?

01:21:40.340 | Then is it fundamentally different

01:21:42.060 | from the world of programming?

01:21:44.580 | - I don't know.

01:21:45.460 | I mean, certainly, in all seriousness,

01:21:47.660 | I just didn't have the talent for it.

01:21:49.460 | When I got here as a grad student at Princeton

01:21:51.860 | and I started to think about research

01:21:53.540 | at the end of my, I don't know, first year

01:21:55.620 | or something like that,

01:21:56.480 | I worked briefly with John Hopcroft,

01:21:59.060 | who is absolutely, you know,

01:22:00.940 | you mentioned during award winner, et cetera,

01:22:02.620 | a great guy.

01:22:04.060 | And it became crystal clear

01:22:05.460 | I was not cut out for this stuff, period.

01:22:08.260 | Okay.

01:22:09.260 | And so I moved into things

01:22:11.540 | where I was more cut out for it.

01:22:13.620 | And that tended to be things like writing programs

01:22:16.820 | and then ultimately writing books.

01:22:19.500 | - You've said that in Toronto,

01:22:22.140 | as an undergrad, you did a senior thesis

01:22:24.260 | or a literature survey on artificial intelligence.

01:22:28.740 | This was 1964.

01:22:30.540 | - Correct.

01:22:32.220 | - What was the AI landscape, ideas, dreams at that time?

01:22:37.140 | - I think that was one of the,

01:22:39.100 | well, you've heard of AI winters.

01:22:40.420 | This is whatever the opposite was,

01:22:41.820 | AI summer or something.

01:22:43.700 | It was one of these things

01:22:44.540 | where people thought that,

01:22:46.820 | boy, we could do anything with computers,

01:22:49.300 | that all these hard problems,

01:22:51.500 | computers will solve them.

01:22:52.700 | They will do machine translation.

01:22:54.420 | They will play games like chess.

01:22:57.860 | They will do, you know,

01:23:00.780 | prove theorems in geometry.

01:23:02.140 | There are all kinds of examples like that

01:23:04.180 | where people thought,

01:23:05.700 | boy, we could really do those sorts of things.

01:23:08.880 | And, you know, I read the Kool-Aid in some sense.

01:23:14.900 | There's a wonderful collection of papers

01:23:16.900 | called "Computers and Thought"

01:23:18.020 | that was published in about that era.

01:23:20.260 | And people were very optimistic.

01:23:22.500 | And then of course it turned out that

01:23:24.220 | what people had thought was just a few years down the pike

01:23:27.340 | was more than a few years down the pike.

01:23:31.300 | And some parts of that are more or less now

01:23:34.580 | sort of under control.

01:23:36.380 | We finally do play games like go and chess and so on

01:23:39.300 | better than people do,

01:23:41.140 | but there are others,

01:23:42.500 | and machine translation is a lot better than it used to be,

01:23:45.100 | but that's, you know, 50, close to 60 years of progress

01:23:49.700 | and a lot of evolution in hardware

01:23:51.340 | and a tremendous amount more data

01:23:52.820 | upon which you can build systems

01:23:56.580 | that actually can learn from some of that data.

01:23:58.940 | - And the infrastructure to support developers

01:24:02.600 | working together like an open source movement,

01:24:05.660 | the internet period is also an empowering.

01:24:08.780 | But what lessons do you draw from that,

01:24:11.740 | the opposite of winter, that optimism?

01:24:13.780 | - Well, I guess the lesson is that in the short run,

01:24:19.700 | it's pretty easy to be too pessimistic

01:24:23.540 | or maybe too optimistic,

01:24:24.700 | and in the long run, you probably shouldn't be too pessimistic.

01:24:27.180 | I'm not saying that very well.

01:24:28.620 | It reminds me of this remark from Arthur Clarke,

01:24:32.740 | a science fiction author, who says,

01:24:34.620 | "When some distinguished but elderly person

01:24:36.500 | "says that something is possible, he's probably right.

01:24:41.160 | "And if he says it's impossible, he's almost surely wrong."

01:24:44.340 | But you don't know what the timescale is.

01:24:45.780 | - The timescale is critical, right.

01:24:48.380 | So what are your thoughts on this new summer of AI

01:24:52.580 | now in the work with machine learning and neural networks?

01:24:55.420 | You've kind of mentioned that you started to try to explore

01:24:57.900 | and look into this world that seems fundamentally different

01:25:01.420 | from the world of heuristics and algorithms like search,

01:25:06.220 | that it's now purely sort of trying to take

01:25:08.980 | huge amounts of data and learn from that data, right?

01:25:12.480 | Programs from the data.

01:25:14.040 | - Yeah.

01:25:14.940 | Look, I think it's very interesting.

01:25:17.000 | I am incredibly far from an expert.

01:25:19.800 | Most of what I know I've learned from my students,

01:25:21.540 | and they're probably disappointed

01:25:24.520 | in how little I've learned from them.

01:25:26.360 | But I think it has tremendous potential

01:25:29.200 | for certain kinds of things.

01:25:30.560 | I mean, games is one where it obviously has had an effect

01:25:34.640 | on some of the others as well.

01:25:36.000 | I think there's, and this is speaking from

01:25:39.520 | definitely not expertise,

01:25:40.700 | I think there are serious problems in certain kinds

01:25:43.640 | of machine learning, at least,

01:25:45.480 | because what they're learning from

01:25:47.500 | is the data that we give them.

01:25:49.180 | And if the data we give them has something wrong with it,

01:25:52.060 | then what they learn from it is probably wrong too.

01:25:54.900 | And the obvious thing is some kind of bias in the data.

01:25:59.140 | That the data has stuff in it like, I don't know,

01:26:02.340 | women aren't as good as men at something, okay?

01:26:05.440 | That's just flat wrong.

01:26:07.380 | But if it's in the data because of historical treatment,

01:26:11.480 | then that machine learning stuff will propagate that.

01:26:15.020 | And that is a serious worry.

01:26:17.740 | - The positive part of that is what machine learning does

01:26:22.700 | is reveal the bias in the data

01:26:24.680 | and puts a mirror to our own society.

01:26:27.020 | And in so doing, helps us remove the bias,

01:26:30.940 | you know, helps us work on ourselves.

01:26:33.860 | - Puts a mirror to ourselves.

01:26:35.720 | - Yeah, that's an optimistic point of view.

01:26:37.440 | And if it works that way, that would be absolutely great.

01:26:40.000 | And what I don't know is whether it does work that way

01:26:42.560 | or whether the AI mechanisms

01:26:46.440 | or machine learning mechanisms reinforce

01:26:48.520 | and amplify things that have been wrong in the past.

01:26:52.640 | And I don't know, but I think that's a serious thing

01:26:56.000 | that we have to be concerned about.

01:26:58.740 | - Let me ask you an out there question, okay?

01:27:01.200 | I know nobody knows, but what do you think it takes

01:27:03.920 | to build a system of human level intelligence?

01:27:07.400 | That's been the dream from the '60s.

01:27:09.860 | We talk about games, about language,

01:27:12.120 | about image recognition,

01:27:15.140 | but really the dream is to create human level

01:27:18.220 | or superhuman level intelligence.

01:27:19.600 | What do you think it takes to do that?

01:27:21.220 | And are we close?

01:27:23.060 | - I haven't a clue and I don't know, roughly speaking.

01:27:26.200 | I mean, this was Turing--

01:27:27.040 | - I was trying to trick you into a hypothesis.

01:27:30.020 | - Yeah, I mean, Turing talked about this

01:27:31.500 | in his paper on machine intelligence back in,

01:27:34.660 | geez, I don't know, early '50s or something like that.

01:27:36.780 | And he had the idea of the Turing test.

01:27:38.300 | And I don't know whether the Turing test is--

01:27:40.940 | - Is a good test of intelligence.

01:27:41.780 | - I don't know, it's an interesting test.

01:27:43.500 | At least it's in some vague sense objective,

01:27:45.780 | whether you can read anything into the conclusions

01:27:48.460 | is a different story.

01:27:50.420 | - Do you have worries, concerns, excitement

01:27:55.140 | about the future of artificial intelligence?

01:27:56.960 | So there's a lot of people who are worried.

01:27:58.900 | And you can speak broadly than just artificial intelligence.

01:28:01.740 | It's basically computing taking over the world

01:28:05.340 | in various forms.

01:28:06.740 | Are you excited by this future,

01:28:09.220 | this possibility of computing being everywhere?

01:28:12.320 | Or are you worried?

01:28:13.820 | - It's some combination of those.

01:28:16.420 | I think almost all technologies over the long run

01:28:21.200 | are for good, but there's plenty of examples

01:28:24.600 | where they haven't been good,

01:28:26.260 | either over a long run for some people

01:28:28.660 | or over a short run.

01:28:30.540 | And computing is one of those.

01:28:33.220 | And AI within it is gonna be one of those as well.

01:28:36.820 | But computing broadly, I mean,

01:28:38.180 | for just a today example is privacy.

01:28:41.620 | That the use of things like social media and so on means that

01:28:46.620 | and the commercial surveillance means that

01:28:49.820 | there's an enormous amount more known about us by people,

01:28:54.340 | other businesses, government, whatever,

01:28:56.940 | than perhaps one ought to feel comfortable with.

01:28:59.540 | So that's an example.

01:29:01.080 | - So that's an example of a possible negative effect

01:29:07.620 | of computing being everywhere.

01:29:09.700 | It's an interesting one 'cause it could also be a positive

01:29:14.060 | if leveraged correctly.

01:29:16.140 | - There's a big if there.

01:29:18.140 | - So I have a deep interest in human psychology

01:29:22.980 | and humans seem to be very paranoid about this data thing.

01:29:27.460 | But that varies depending on age group.

01:29:31.380 | It seems like the younger folks,

01:29:32.900 | so it's exciting to me to see what society looks like

01:29:35.940 | 50 years from now,

01:29:37.460 | that the concerns about privacy might be flipped

01:29:39.940 | on their head based purely on human psychology

01:29:42.700 | versus actual concerns or not.

01:29:45.620 | What do you think about Moore's law?

01:29:49.580 | Well, you said a lot of stuff we've talked,

01:29:52.020 | you talked about programming languages in their design,

01:29:55.780 | in their ideas come from the constraints

01:29:58.780 | in the systems they operate in.

01:30:00.460 | Do you think Moore's law,

01:30:01.820 | the exponential improvement of systems

01:30:07.180 | will continue indefinitely?

01:30:08.820 | There's a mix of opinions on that currently.

01:30:12.420 | Or do you think there'll be a plateau?

01:30:17.420 | - Well, the frivolous answer is no exponential

01:30:21.620 | can go on forever.

01:30:22.660 | You run out of something.

01:30:26.100 | Just as we said, timescale matters.

01:30:27.740 | So if it goes on long enough, that might be all we need.

01:30:30.860 | - Yeah, right, won't matter to us.

01:30:33.300 | So I don't know,

01:30:34.140 | we've seen places where Moore's law has changed.

01:30:35.940 | For example, mentioned earlier,

01:30:37.420 | processors don't get faster anymore,

01:30:41.300 | but you use that same growth of,

01:30:45.700 | the ability to put more things in a given area

01:30:48.060 | to go horizontally instead of vertically as it were.

01:30:51.100 | So you can get more and more processors

01:30:52.940 | or memory or whatever on the same chip.

01:30:55.580 | Is that gonna run into a limitation?

01:30:57.380 | Presumably, because at some point

01:31:00.660 | you get down to the individual atoms.

01:31:03.100 | And so you gotta find some way around that.

01:31:05.540 | Will we find some way around that?

01:31:07.780 | I don't know, I just said that if I say it won't,

01:31:10.020 | I'll be wrong.

01:31:10.860 | So perhaps we will.

01:31:12.540 | - So I just talked to Jim Keller and he says,

01:31:15.020 | so he actually describes,

01:31:16.380 | he argues that the Moore's law will continue

01:31:18.500 | for a long, long time because you mentioned the atom.

01:31:21.780 | We actually have, I think a thousand fold increase,

01:31:25.260 | decreased in transistor size still possible

01:31:29.940 | before we get to the quantum level.

01:31:32.060 | So there's still a lot of possibilities.

01:31:34.700 | He thinks it'll continue indefinitely,

01:31:36.420 | which is an interesting, optimistic viewpoint.

01:31:40.660 | But how do you think the programming languages

01:31:43.420 | will change with this increase?

01:31:45.420 | Whether we hit a wall or not,

01:31:47.660 | what do you think?

01:31:49.420 | Do you think there'll be a fundamental change

01:31:51.420 | in the way programming languages are designed?

01:31:54.500 | - I don't know about that.

01:31:55.380 | I think what will happen is continuation

01:31:58.620 | of what we see in some areas at least,

01:32:02.020 | which is that more programming will be done

01:32:05.500 | by programs than by people.

01:32:08.220 | And that more will be done by sort of declarative

01:32:12.700 | rather than procedural mechanisms where I say,

01:32:15.740 | I want this to happen, you figure out how.

01:32:19.820 | And that is in many cases at this point,

01:32:24.260 | domain of specialized languages for narrow domains,

01:32:28.700 | but you can imagine that broadening out.

01:32:31.860 | And so I don't have to say so much in so much detail,

01:32:35.660 | some collection of software, let's call it languages

01:32:39.340 | or programs or something,

01:32:41.340 | we'll figure out how to do what I want to do.

01:32:44.820 | - Interesting, so increased levels of abstraction.

01:32:47.220 | - Yeah.

01:32:48.900 | - And one day getting to the human level

01:32:50.980 | where we can just use natural language.

01:32:52.780 | - Could be possible.

01:32:54.580 | - So you taught, still teach a course,

01:32:56.780 | Computers in Our World here at Princeton

01:32:59.740 | that introduces computing and programming to non-majors.

01:33:03.280 | Just from that experience, what advice do you have

01:33:07.740 | for people who don't know anything about programming

01:33:10.580 | but are kind of curious about this world

01:33:12.940 | where programming seems to become more and more

01:33:14.780 | of a fundamental skill that people need

01:33:17.060 | to be at least aware of?

01:33:18.740 | - Well, I can recommend a good book.

01:33:20.380 | - What's that?

01:33:21.220 | (laughing)

01:33:22.060 | - The book I wrote for the course.

01:33:24.340 | I think this is one of these questions

01:33:26.660 | of should everybody know how to program?

01:33:28.500 | And I think the answer is probably not,

01:33:31.260 | but I think everybody should at least understand

01:33:32.980 | sort of what it is so that if you say to somebody,

01:33:35.660 | I'm a programmer, they have a notion of what that might be,

01:33:38.100 | or if you say this is a program,

01:33:40.140 | or this was decided by a computer running a program,

01:33:43.560 | that they have some vague intuitive understanding

01:33:47.580 | and accurate understanding of what that might imply.

01:33:51.400 | So part of what I'm doing in this course,

01:33:55.180 | which is very definitely for non-technical people,

01:33:57.460 | I mean, a typical person in it is a history

01:33:59.340 | or English major, try and explain how computers work,

01:34:03.700 | how they do their thing, what programming is,

01:34:06.300 | how you write a program,

01:34:07.620 | and how computers talk to each other,

01:34:11.400 | and what do they do when they're talking to each other.

01:34:14.300 | And then I would say nobody, very rarely,

01:34:19.300 | and does anybody in that course go on

01:34:21.900 | to become a real serious programmer,

01:34:24.180 | but at least they've got a somewhat better idea

01:34:27.140 | of what all this stuff is about, not just the programming,

01:34:29.580 | but the technology behind computers and communications.

01:34:32.620 | - Do they try and write a program themselves?

01:34:35.700 | - Oh yeah, yeah, a very small amount.

01:34:38.340 | I introduced them to how machines work

01:34:40.620 | at a level below high-level languages,

01:34:43.020 | so we have a kind of a toy machine

01:34:45.220 | that has a very small repertoire, a dozen instructions,

01:34:47.800 | and they write trivial assembly language programs for that.

01:34:51.060 | - Really? - Yeah.

01:34:51.900 | - Wow, so can you, just, if you were to give a flavor

01:34:55.060 | to people of the programming world, of the computing world,

01:34:59.500 | what are the examples they should go with?

01:35:01.940 | So a little bit of assembly to get a sense

01:35:04.300 | at the lowest level of what the program is really doing?

01:35:08.820 | - Yeah, I mean, in some sense,

01:35:10.740 | there's no such thing as the lowest level,

01:35:12.500 | because you can keep going down,

01:35:13.620 | but that's the place where I drew the line.

01:35:15.580 | So the idea that computers have a fairly small repertoire

01:35:19.380 | of very simple instructions that they can do,

01:35:21.740 | like add and subtract and branch and so on,

01:35:25.020 | as you mentioned earlier,

01:35:26.220 | and that you can write code at that level,

01:35:31.540 | and it will get things done,

01:35:33.220 | and then you have the levels of abstraction

01:35:35.540 | that we get with higher-level languages,

01:35:37.820 | like Fortran or C or whatever,

01:35:39.860 | and that makes it easier to write the code

01:35:42.340 | and less dependent on particular architectures.

01:35:44.940 | And then we talk about a lot of the different kinds

01:35:48.020 | of programs that they use all the time

01:35:50.620 | that they don't probably realize are programs,

01:35:52.740 | like they're running Mac OS on their computers

01:35:57.620 | or maybe Windows, and they're downloading apps

01:36:00.220 | on their phones, and all of those things are programs

01:36:03.000 | that are just what we just talked about,

01:36:05.940 | except at a grand scale.

01:36:08.140 | - Yeah, it's easy to forget that they're actual programs

01:36:10.500 | that people program.

01:36:11.820 | There's engineers that wrote those things.

01:36:14.060 | - Yeah, right, and so in a way,

01:36:17.660 | I'm expecting them to make an enormous conceptual leap

01:36:20.580 | from their five or 10-line toy assembly language thing

01:36:24.540 | that adds two or three numbers to something

01:36:28.260 | that is a browser on their phone or whatever,

01:36:31.020 | but it's really the same thing.

01:36:33.460 | - So if you look in broad strokes at history,

01:36:37.160 | what do you think the world,

01:36:39.740 | how do you think the world changed because of computers?

01:36:42.860 | It's hard to sometimes see the big picture

01:36:45.220 | when you're in it, but I guess I'm asking

01:36:48.020 | if there's something you've noticed over the years

01:36:51.540 | that, like you were mentioning,

01:36:54.540 | the students are more distracted looking at their,

01:36:56.820 | now there's a device to look at.

01:36:58.500 | - Right, well, I think computing has changed

01:37:00.580 | a tremendous amount, obviously,

01:37:02.180 | but I think one aspect of that is the way

01:37:04.320 | that people interact with each other,

01:37:06.460 | both locally and far away, and when I was the age

01:37:11.300 | of those kids, making a phone call to somewhere

01:37:14.160 | was a big deal because it cost serious money,

01:37:17.220 | and this was in the '60s, right?

01:37:19.280 | And today, people don't make phone calls,

01:37:22.900 | they send texts or something like that,

01:37:25.660 | so there's an up and down in what people do.

01:37:29.500 | People think nothing of having correspondence,

01:37:34.100 | regular meetings, video, whatever, with friends

01:37:37.460 | or family or whatever in any other part of the world,

01:37:40.380 | and they don't think about that at all.

01:37:43.060 | And so that's just the communication aspect of it.

01:37:47.100 | - Do you think that brings us closer together,

01:37:50.940 | or does it make us, does it take us away

01:37:55.940 | from the closeness of human-to-human contact?

01:37:59.140 | - I think it depends a lot on all kinds of things.

01:38:02.820 | So I trade mail with my brother and sister in Canada

01:38:05.860 | much more often than I used to talk to them on the phone.

01:38:08.860 | So probably every two or three days,

01:38:10.740 | I get something or send something to them,

01:38:14.420 | whereas 20 years ago, I probably wouldn't have talked

01:38:18.860 | to them on the phone nearly as much.

01:38:20.620 | So in that sense, that's brought my brother

01:38:22.660 | and sister and I closer together.

01:38:24.020 | That's a good thing.

01:38:25.020 | I watch the kids on campus, and they're mostly walking

01:38:29.820 | around with their heads down, fooling with their phones

01:38:32.260 | to the point where I have to duck them.

01:38:34.900 | I don't know that that has brought them closer together

01:38:39.420 | in some ways.

01:38:40.500 | There's sociological research that says people are,

01:38:43.740 | in fact, not as close together as they used to be.

01:38:46.220 | I don't know whether that's really true,

01:38:47.580 | but I can see potential downsides in kids

01:38:52.540 | where you think, "Come on, wake up

01:38:54.340 | "and smell the coffee," or whatever.

01:38:56.780 | - That's right, but if you look at, again,

01:38:59.140 | nobody can predict the future, but are you excited

01:39:02.100 | and I kind of touched this a little bit with AI,

01:39:04.820 | but are you excited by the future in the next 10, 20 years

01:39:08.780 | that computing will bring?

01:39:11.580 | You were there when there was no computers, really,

01:39:15.900 | and now computers are everywhere, all over the world,

01:39:19.460 | in Africa and Asia, and just every person,

01:39:23.100 | almost every person in the world has a device.

01:39:25.580 | So are you hopeful, optimistic about that future?

01:39:30.740 | - It's mixed, if the truth be told.

01:39:32.420 | I mean, I think there are some things about that

01:39:34.140 | that are good.

01:39:34.980 | I think there's the potential for people

01:39:36.740 | to improve their lives all over the place,

01:39:39.100 | and that's obviously good.

01:39:40.940 | And at the same time, at least in the short run,

01:39:44.260 | you can see lots and lots of bad

01:39:45.900 | as people become more tribalistic

01:39:48.340 | or parochial in their interests,

01:39:50.360 | and it's an enormous amount more us and them,

01:39:53.180 | and people are using computers in all kinds of ways

01:39:55.700 | to mislead or misrepresent or flat out lie

01:39:58.820 | about what's going on, and that is affecting politics

01:40:01.860 | locally and, I think, everywhere in the world.

01:40:04.220 | - Yeah, the long-term effect on political systems

01:40:08.900 | and so on, it's who knows.

01:40:10.900 | - Who knows, indeed.

01:40:11.860 | (Lex laughing)

01:40:14.440 | - People now have a voice, which is a powerful thing.

01:40:18.620 | People who are oppressed have a voice,

01:40:21.020 | but also everybody has a voice,

01:40:24.100 | and the chaos that emerges from that is fascinating to watch.

01:40:26.820 | - Yeah, yeah, it's kind of scary.

01:40:29.180 | - If you can go back and relive a moment in your life,

01:40:33.700 | one that made you truly happy outside of family,

01:40:37.520 | or was profoundly transformative,

01:40:40.060 | is there a moment or moments that jump out at you

01:40:44.340 | from memory?

01:40:45.180 | - I don't think specific moments.

01:40:48.040 | I think there were lots and lots and lots of good times

01:40:50.340 | at Bell Labs where you would build something,

01:40:52.500 | and it worked.

01:40:55.380 | Huh, Jason, it worked.

01:40:56.740 | - So the moment it worked.

01:40:57.940 | - Yeah, and somebody used it, and they said,

01:41:00.220 | "Gee, that's neat."

01:41:01.260 | Those kinds of things happened quite often

01:41:04.620 | in that sort of golden era in the '70s

01:41:07.820 | when Unix was young, and there was all this

01:41:10.460 | low-hanging fruit and interesting things to work on,

01:41:13.420 | and a group of people who kind of,

01:41:16.140 | we were all together in this, and if you did something,

01:41:18.900 | they would try it out for you.

01:41:20.540 | And I think that was, in some sense,

01:41:22.780 | a really, really good time.

01:41:24.500 | - And awk was awk an example of that,

01:41:27.460 | that when you built it, then people used it?

01:41:29.420 | - Yeah, absolutely.

01:41:30.420 | - And now millions of people use.

01:41:32.700 | - And all your stupid mistakes are right there

01:41:34.540 | for them to look at, right?

01:41:36.500 | So it's mixed.

01:41:37.460 | - Yeah, it's terrifying and vulnerable,

01:41:39.140 | but it's beautiful 'cause it does have a positive impact

01:41:42.020 | on so, so many people.

01:41:43.820 | So I think there's no better way to end it.

01:41:47.220 | Brian, thank you so much for talking to me.

01:41:48.660 | It was an honor.

01:41:49.500 | - Okay, my pleasure.

01:41:51.260 | Good fun.

01:41:53.780 | - Thank you for listening to this conversation

01:41:55.380 | with Brian Kernighan, and thank you to our sponsors,

01:41:58.540 | 8sleep Mattress and Raycon Earbuds.

01:42:02.380 | Please consider supporting this podcast

01:42:05.140 | by going to 8sleep.com/lex

01:42:08.340 | and to buy raycon.com/lex.

01:42:12.140 | Click the links, buy the stuff.

01:42:14.540 | These both are amazing products.

01:42:16.800 | It really is the best way to support this podcast

01:42:19.340 | and the journey I'm on.

01:42:21.180 | It's how they know I sent you and increases the chance

01:42:24.740 | that they'll actually support this podcast in the future.

01:42:27.700 | If you enjoy this thing, subscribe on YouTube,

01:42:30.300 | review it with Firestarz and Apple Podcast,

01:42:32.660 | support it on Patreon, or connect with me on Twitter

01:42:35.980 | at Lex Friedman, spelled somehow, miraculously,

01:42:40.100 | without the letter E, just F-R-I-D-M-A-N,

01:42:44.100 | because when we immigrated to this country,

01:42:46.640 | we were not so good at spelling.

01:42:49.060 | And now let me leave you with some words

01:42:51.380 | from Brian Kernighan.

01:42:53.340 | Don't comment bad code, rewrite it.

01:42:56.200 | Thank you for listening and hope to see you next time.

01:43:00.600 | (upbeat music)

01:43:03.180 | (upbeat music)

01:43:05.760 | [BLANK_AUDIO]

Brian Kernighan: UNIX, C, AWK, AMPL, and Go Programming | Lex Fridman Podcast #109

Chapters