Why TikTok’s Sneak Attack on Facebook Matters | Deep Questions With Cal Newport

00:00:00.000 | Now, before we get into those two parts, though, I'd like to start the shows when possible

00:00:04.720 | reacting to what's going on in the news. Uh, in particular, I have an article I want to talk

00:00:11.200 | about that we didn't really get around to talking on the last episode, but we should have because

00:00:15.120 | it's by me. So when I publish new things, especially new things in the New Yorker,

00:00:20.240 | I like to try to discuss them on the show in a little bit more depth, let you know what I'm

00:00:25.440 | thinking, maybe give you some more angles on it. So a couple of weeks ago, I published my latest

00:00:31.520 | for the New Yorker. It's titled Tik Tok and the fall of the social media giants. If you are

00:00:36.240 | watching this episode instead of just listening. So if you're at youtube.com/calnewportmedia,

00:00:40.800 | you'll see the article on your screen as well. Now, why I thought it was important to talk about

00:00:45.680 | this topic is not that I haven't talked about it before on the show, but because of the opposite

00:00:49.600 | of that, I've talked about this topic a lot on the show. That's kind of what makes this cool for us.

00:00:55.360 | This is one of the first instances, I would say, of an idea whose birth was this podcast.

00:01:01.600 | I began bouncing these ideas around here with you, my listeners, and it evolved into a more

00:01:08.000 | polished form, the New Yorker. So here we are, Deep Questions podcast impacting the world of

00:01:15.120 | news. So you've heard these ideas before, but they're a lot more polished now. So let's go

00:01:19.120 | through it quickly. So I begin the article quoting Blake Chanley, Tik Tok's president of

00:01:28.320 | Global Business Solutions, talking about competition from Facebook. This is where he

00:01:34.720 | said, look, we're not worried. Facebook is a social platform built on a social graph. Tik

00:01:39.360 | Tok is an entertainment platform. These are two different things. Zuckerberg needs to stay in his

00:01:43.760 | lane. We have our own lane. That name should sound familiar. Blake Chanley, I talked about

00:01:48.000 | him on the show because I discovered this interview from one of you, my listeners, sent it

00:01:52.480 | to me at my interesting@calnewport.com email address. So then I went deeper into this.

00:01:58.400 | So here's the next point I want to make. So in the article, I make this point that Facebook

00:02:07.040 | is trying to shift to be more like Tik Tok. Instagram, owned by the same company, shifting

00:02:14.320 | to be more like Tik Tok. And the way they are doing this is by moving towards video,

00:02:20.480 | moving towards short video, but most importantly, as I'll soon elaborate,

00:02:24.480 | moving towards recommendations that have nothing to do with the social graph, moving towards feeds

00:02:32.160 | that are generated purely by algorithms, not by who you follow, not by people you know,

00:02:37.280 | sharing it. So here's the quote from the article. This shift is not surprising, given Tik Tok's

00:02:45.520 | phenomenal popularity, and it has been very successful, but it's also short sighted.

00:02:51.440 | So the whole thesis of the article is the following. I say, platforms like Facebook

00:02:57.920 | could be doomed if they fail to maintain the social graphs upon which they built their kingdoms.

00:03:05.760 | So this is the setup to this whole article. Tik Tok's really popular. Companies like,

00:03:10.960 | platforms like Facebook, I should say, are trying to be more Tik Tok like,

00:03:13.840 | to stave off the competition and try to have faster user growth. This could, however, doom them.

00:03:22.400 | All right, so let me elaborate this argument. What I'm getting at in this,

00:03:29.760 | the original take on this point is there's real advantages to having an online company like

00:03:37.520 | Facebook or Instagram, or as you'll see, Twitter, that relies on a social graph.

00:03:42.000 | And to be clear, by social graph, I mean, all of these individual friend requests,

00:03:48.560 | all these individual follow clicks, this topology of connections that human users built up,

00:03:55.920 | click by click, decision by decision over months and months and years and years of use,

00:03:59.920 | these really rich networks, these give a real advantage to the early mover platforms who have

00:04:06.080 | built them up. One is a network effect advantage. So I'm highlighting these here.

00:04:09.840 | So I mentioned, for example, once Facebook had 100 million followers, active users, I should say,

00:04:17.520 | which it got by 2006, it became hard for anyone else to compete on a model of the people you know

00:04:22.960 | are here. And so you can see what the people you know are up to. Once you have 100 million users

00:04:27.200 | on one platform, how do you start from scratch and say, well, we only have 10,000, but we're growing.

00:04:31.520 | So that's a huge network effect advantage. There are also other advantages to these topologies.

00:04:37.680 | I talk a little bit in this article about the introduction of the retweet button to Twitter.

00:04:42.320 | I think people don't understand the degree to which the introduction of the retweet button

00:04:46.160 | not only made Twitter into a lasting company of cultural influence, but also

00:04:52.160 | really transformed social media. So here's what happened when they introduced the retweet button.

00:04:55.760 | The friction required to send a link or tweet to your entire follower base went down to almost

00:05:04.560 | nothing. Before we had the retweet button, people in Twitter would have to copy tweets and they would

00:05:10.320 | put RT. So you remember this, if you're old enough, you would put RT in caps colon, and then you would

00:05:17.760 | quote, just copy and paste quote, the tweet that you're retweeting. And then you would put at the

00:05:23.840 | original person who retweeted it below it. That's a pain. Retweet button, you see something you like,

00:05:28.320 | click a button and it spreads. That reduction in friction made all of the difference. It unlocked

00:05:34.800 | what I call a fierce viral dynamic. So now a tweet that was catching the attention of the zeitgeist

00:05:42.000 | in the right way, could spread through the power law topology of the Twitter social graph at

00:05:48.560 | frightening speed. Tens of millions, hundreds of millions of people could see something within

00:05:53.360 | an hour or two. And this all was based off these individual user decisions to retweet or not

00:05:59.360 | retweet. They see it from a few people, they retweet it. This turned out to have two hugely

00:06:03.520 | powerful effects. One, it attracted more interesting people to platforms like Twitter,

00:06:09.200 | because there's potential here. You had the potential of reaching millions of people if

00:06:14.400 | what you wrote caught on just right. So now more interesting people came to Twitter. That's another

00:06:18.880 | big network effect advantage. If all of the interesting people are on Twitter, when a

00:06:24.000 | competitor came along like Parler or Gab, they didn't do as well because there wasn't as many

00:06:30.080 | interesting people on there saying interesting things. Perhaps more important, these fierce

00:06:36.400 | viral dynamics also gave way to a frighteningly effective distributed curation mechanism.

00:06:42.720 | So what happened is all these individual decisions that are tweet and retweet,

00:06:46.880 | created this human powered curation algorithm that was really, really good at figuring out

00:06:54.480 | what's the most interesting, controversial, outrageous, funny, off the wall, but perfect

00:07:00.160 | meme, whatever it was. It did a really good job of selecting for these things and amplifying it

00:07:04.800 | so millions of people could see. So that was very effective. So suddenly Twitter not only had

00:07:10.000 | interesting people on there, but it had this really good human powered curation, distributed

00:07:14.560 | curation algorithm that meant when you went on Twitter, you were going to see interesting things

00:07:18.160 | or funny things or outrageous things. And again, this is not sophisticated machine learning

00:07:23.200 | algorithms at play here. This is just the epiphenomenon of a lot of individuals clicking

00:07:27.760 | retweet or not. That's very powerful. So then as I get into it, Facebook noticed this retweet thing

00:07:35.200 | is working. So then by 2012, we get a share button added to the mobile app of Facebook.

00:07:42.000 | It's exactly retweet. They too wanted to take advantage of that distributed curation.

00:07:47.120 | All right, so here's my summary. Both Facebook and Twitter are built on the same general model

00:07:52.080 | of leveraging hard to replicate large social graphs to generate a never ending stream of

00:07:56.960 | engaging content. A strategy that proved to be robust in the face of new competition and

00:08:02.320 | incredibly lucrative. So it's the social graphs and those three advantages I just summarized,

00:08:08.320 | the people you know are on there, the interesting people are on there, and you have the distributed

00:08:12.400 | curation effect of the share and retweets. That made the small number of companies who got there

00:08:18.000 | first. Facebook, Instagram, Instagram, I took this out of the article, but I had a piece in there

00:08:24.080 | before about how Instagram basically, they were able to make a run at the castle that Facebook

00:08:30.080 | was building, because Facebook didn't understand image. And Instagram made it much easier to have

00:08:35.200 | these beautiful images filtered, perfectly sized for iPhones. That was so powerful, that media,

00:08:41.200 | that they were able to build a new audience from scratch before Facebook squashed that by buying

00:08:45.120 | them. So you have Facebook, you have Instagram, you have Twitter, these big social graphs that

00:08:49.440 | can produce a non-stop stream of engaging content of a style that no one else can compete with.

00:08:54.720 | So if I want to go start my own social network tomorrow, to compete with those giants doing

00:09:01.840 | their same game, as many pretenders to the throne have encountered in the past five to 10 years,

00:09:08.480 | good luck. Because until I can get a hundred million people, including a lot of people you

00:09:14.000 | know, and a lot of really interesting people with all these intricate friend and follower

00:09:17.200 | connections that allow retweets and shares to become a really effective distributed curation

00:09:21.600 | mechanism, until that's all there, this is useless. It's not that interesting. I'd rather

00:09:26.720 | just go on Twitter. I'd rather go on Facebook. I'd rather go on Instagram. So they had this

00:09:30.720 | readout, this digital readout that was almost impossible for anyone to raid. And so that's

00:09:38.720 | why these pseudo monopolies grew bigger. It's why their influence on our culture became stronger.

00:09:45.120 | It's why we began to get really worried starting around 2016, 2017, about just how much power

00:09:50.400 | these small number of companies had because they were immune to competition. And in my opinion,

00:09:55.280 | we're strangling the potential of the interactive web 2.0 by taking the democratic weirdness of the

00:10:03.600 | internet, the distributed homespun eccentricities of the internet and capturing it all into a small

00:10:10.160 | number of walled gardens. All right. Then you get tick tock and tick tock is the, is the weird night

00:10:19.040 | that came out of the bog that suddenly threatening all the castles. It's the, I don't know game of

00:10:25.200 | thrones well, but the, uh, the woman with the dragons who comes out of nowhere and suddenly

00:10:31.200 | the Lannisters. Do I have the right chest? I don't know. I don't know. Game of thrones. I

00:10:37.120 | don't know, but there's, you know, there's these people in charge and then there's this,

00:10:40.320 | this woman who had dragons and kind of came out of nowhere and tick tock was tick tock is that.

00:10:44.640 | Um, okay. So why is this the second part of my article? So what I, the case I make is the

00:10:51.360 | effectiveness of the tick tock experience is found in what it doesn't require. So tick tocks

00:10:56.640 | brilliance. Is it, it's the same sort of user generated content to stretch and experience,

00:11:01.200 | but it does not leverage a social graph. And this was the point that Blake Shanley was making in

00:11:05.520 | that very important interview. It tick tock does not care if people, you know, are on tick tock,

00:11:10.560 | tick tock does not care if famous people are on tick tock, tick tock does not care who you follow

00:11:15.840 | or what you can share. There are some social features in tick tock. No one uses them. It is

00:11:20.960 | purely algorithmic. It looks at the total pool of available videos. And by they, I mean, this.

00:11:28.000 | Brutally effective machine learning loop and says, what should I show you next? What should I show

00:11:33.440 | you next? And as far as I can tell, I really went deep into this. I tried to understand what was

00:11:37.280 | happening with this algorithm. It's all proprietary, but there is some hints. I think the best hints

00:11:41.520 | probably come from the wall street journal, which did a big study where they created hundreds.

00:11:46.880 | Well, I should speak the New York, New York, New York fact checker corrected on the, on this

00:11:52.400 | more than a hundred fake accounts, um, which they, they tweaked very carefully and observed to try

00:11:58.720 | to understand how the recommendations were happening and what, what seems to be happening

00:12:01.840 | with the, the tech talk algorithm is that it's, it's a, for you statistical optimizational people,

00:12:07.440 | it's essentially a stochastic multi-arm bandit reinforcement optimization machine learning loop.

00:12:14.640 | So go look that up if you want to bore yourself. But what it does is it's showing you a lot of

00:12:18.960 | different things at first, somewhat randomly. And the main thing it looks at is how long did

00:12:24.240 | you watch the video before you swiped up to the next one? That is its best indication of the

00:12:29.360 | rewards you got from that video. It then uses statistically shows you statistically similar

00:12:35.280 | videos to that one. And again, it's trying to optimize, Hey, this one that's in the same

00:12:40.560 | universe you watched even longer. So let's, let's focus in on this as a constellation of type of

00:12:45.120 | videos you like. That's basically it. Plus some careful tuning about novelty to make sure that,

00:12:50.480 | you know, you're, you're being exposed to different things. You have a chance for your

00:12:53.680 | interest to wander. It's not super complicated, but within as short as 40 minutes, this is what

00:12:58.720 | the wall street journey found. It took about 40 minutes until the experience was almost eerie.

00:13:05.280 | How well this machine learning loop had honed in on a small number of types of videos that really

00:13:11.120 | push your proverbial buttons. So it's not evil genius. You know, how from the movie 2001 is

00:13:19.920 | sitting in a hollowed out volcano somewhere where we're light years ahead. It's actually a pretty

00:13:24.320 | simple machine learning loop, but no social graph. Just we need a bunch of videos. Anyone

00:13:29.520 | can create videos. It doesn't matter if they're famous. It doesn't matter if you know them,

00:13:33.360 | just get enough people creating videos that our algorithm can go to work. And so TikTok can make

00:13:39.200 | a move at Facebook and Twitter because their castle walls are built around their social graph.

00:13:44.480 | And everyone who was trying to attack their castle with a similar strategy, couldn't get over those

00:13:48.560 | walls. TikTok was the woman with the dragons and they flew over. They didn't use the social graph.

00:13:54.320 | Okay. So what's going to happen? Well, here's the key. I think the key thing going on in social

00:14:01.920 | media, digital news is that TikTok success. These developments put traditional social media

00:14:08.000 | companies like Facebook in a perilous bind. They're losing users right now to TikTok because

00:14:15.440 | TikTok again is in the same cognitive space as far as a user is concerned. User created content,

00:14:20.960 | I want distraction on my phone and they're doing a better job of it. So people are going over to

00:14:24.720 | TikTok and what happened? Meta found that their new user growth slowed and they lost over $200

00:14:31.840 | billion in market capitalization in a single day when they released that report. So this is public

00:14:37.600 | companies. There's huge investor pressure. We can't lose, we can't have our revenue go down.

00:14:43.760 | We can't lose this many users to TikTok, which got to a billion users in just a couple of years.

00:14:47.520 | It's just really exploding. So they have to do something. So what are they doing? They're trying

00:14:52.320 | to be more like TikTok, which may be in the short term makes sense. They're like, okay,

00:14:56.000 | if this is what people like, we need to do the same thing. But this is where I think

00:14:59.520 | they accidentally destabilize their entire foundation of all of their protection.

00:15:03.840 | So here's what I wrote. If companies like Facebook, so they instead move away from

00:15:08.240 | their social graph foundations to concentrate on optimizing in the moment engagement,

00:15:13.440 | they'll enter a competitive landscape that pits them directly against the many other

00:15:18.320 | existing sources of mobile distraction, not just TikTok, but also more bespoke and specialized

00:15:23.760 | social networks, streaming services, et cetera, et cetera. So as soon as you are offering

00:15:28.800 | entertainment, that's not based on the social graph, you are competing with every other source

00:15:32.720 | of entertainment and distraction on the phone. I don't think that's a battle that Facebook or

00:15:37.360 | Instagram or Twitter can win long term. Once they leave the protection of their social graph,

00:15:42.800 | they will be chipped away from by this other competition until eventually

00:15:47.600 | their role of dominance is going to wane. And that's my conclusion. This all points to a

00:15:55.440 | possible future in which social media giants like Facebook may soon be past their long stretch of

00:16:00.560 | dominance. They'll continue to chase new engagement models, leaving behind the protection of their

00:16:05.120 | social graphs and in doing so, eventually succumb to the new competitive pressures this introduces.

00:16:09.600 | Now, last time I talked about this idea on this show, some of my listeners, some of you came

00:16:18.080 | wrote back to me and said, "Oh, so you're saying like TikTok is better or like it's somehow we're

00:16:22.640 | in a better world if TikTok just dominates everything? Isn't that just as bad?" So here's

00:16:26.880 | the key nuance, which by the way, based on your feedback as listeners, I knew to really hit this

00:16:33.200 | point in my New Yorker piece. TikTok of course is subject to the same pressures. So in this future,

00:16:39.600 | it too will eventually fade. So no, I do not think TikTok is going to be some 20 year long

00:16:44.080 | cultural force. I don't think TikTok is going to be a four year from now, be a cultural force.

00:16:48.240 | It's entirely shallow in the sense that it has no network effect foundation. It's just purified

00:16:55.040 | distraction. It's incredibly shallow foundation. The Zeitgeist can change on TikTok like that in a

00:17:00.800 | way that it couldn't on Facebook and it couldn't on Twitter because all your friends were already

00:17:04.560 | on Facebook and all the interesting people were on Twitter and all of these connections and those

00:17:08.320 | social graphs were there. And even if you soured on Facebook because you didn't like their role

00:17:12.480 | in the presidential election, or even if you soured on Twitter because you don't like Elon Musk,

00:17:16.320 | there really was no other game in town that could offer you that. TikTok has none of those

00:17:20.080 | advantages. It's just me look pretty screen, me like, me swipe. Other things can serve that

00:17:27.840 | purpose. So if the Zeitgeist changes against TikTok, it could just fall out of people's favor

00:17:32.640 | almost immediately. So no, I don't think TikTok is going to last as some replacement force of

00:17:37.840 | dominance. I think it's going to, it's coming in hot Twitter, maybe, but definitely Facebook,

00:17:44.320 | definitely Instagram chases it, destabilizes their competitive advantage, leads to their downfall.

00:17:49.440 | The only reason why I say maybe with Twitter is it has investor pressure, but as we saw with the

00:17:55.760 | Elon Musk bid, they're small enough. They could be taken private. So if Musk had taken them private

00:18:01.840 | and you got rid of that investor pressure, he could have just said, and someone else could still do

00:18:06.000 | this. Let's just lean into what we do well. I don't care if we lose users in the short term,

00:18:10.080 | no one else can do what we do at this distributed curation of interesting things said by interesting

00:18:14.880 | people. So let's just keep doing that. We're not going to disappear. Let's just fire a bunch of

00:18:18.800 | people, focus, become profitable. Maybe that's a way to stick around. Facebook, Instagram,

00:18:24.480 | way too big. When you're a $600 billion company, you can't go private. You can't let users fall.

00:18:30.480 | So I think they're, they're essentially doomed. So that's the future I see. And that's how I

00:18:34.160 | summarize this article. I think the social media giants, their dominance is going to wane. I think

00:18:40.560 | TikTok is going to come and go. Its main point will be at help to stabilize those. And all of

00:18:45.600 | this could be good because it opens up the territory. You get rid of the warlords who

00:18:50.640 | were keeping everyone else down. It opens up the territory for more innovation, more interesting

00:18:54.320 | services, bespoke services, more fragmented services, more distributed or nonprofit services,

00:18:59.520 | crazy services that are homespun and eccentric. And only so many people use them, but they love

00:19:04.480 | them. There's more room for all of this when you get rid of these giant a hundred billion dollar

00:19:09.680 | BMS that had these pseudo monopolies. And so that is the conclusion, how I conclude the article.

00:19:14.000 | In the end, TikTok's biggest legacy might be less about its current moment of world conquering

00:19:19.360 | success, which will pass and more about how by forcing social media giants like Facebook to

00:19:27.360 | chase its model, it will end up liberating the social internet. So that is the polished form

00:19:34.160 | of my thoughts about TikTok and its impact on the internet. I think I've now talked about this

00:19:40.240 | enough so we can retire the topic for more, read the article at the New Yorker. You can just find

00:19:46.960 | my contributor page. You'll find it. It's called TikTok in the fall of the social media giants.

00:19:50.880 | If you don't subscribe to the New Yorker, well, you should, but if you don't, I also wrote about

00:19:56.640 | the article and added some extra points in my newsletter. So go to calnewport.com/blog and you

00:20:02.160 | can see the article I wrote about this. And while there, you should sign up for that newsletter so

00:20:05.600 | you can get sent straight to your inbox, this type of thinking. So there we go.

00:20:14.080 | Good summary. The podcast, man, podcast made without the podcast, this idea might not have

00:20:19.360 | existed. It was riffing on the news stuff that listeners sent me that I would not have seen.

00:20:24.320 | I would not have seen that article about Blake Chan, Lee and CNBC, et cetera. If a listener had

00:20:31.360 | to send it to us and then since I wouldn't have seen it and then we were riffing on it, you and

00:20:34.720 | I on the show and that sort of set the wheels in motion. So that's cool. Yeah. It's podcast.

00:20:39.680 | There was some vocabulary in there that I didn't even know. So I learned something for sure.

00:20:44.400 | Welcome to the New Yorker world. Yeah, you got out with the source. You got out with the source

00:20:51.200 | handy. By the way, speaking of New Yorker and then we'll move on. But I read the, you told me

00:20:55.440 | about the article from last week, Tad Friend's article about the door-to-door salesman.

00:20:59.520 | Oh yeah. I haven't finished that yet.

00:21:00.800 | It's really good. Yeah. Yeah. So Jesse told me, I always, if people, people ask me, how do I read,

00:21:05.920 | what's my New Yorker reading habits with the magazine? Um, because it's hard, it's long and

00:21:10.960 | it comes every week. So here it is. So here's my New Yorker reading habits. So you can, you can

00:21:15.040 | follow suit if you subscribe. So you get the daily email and that points out like what's going on,

00:21:20.960 | what was posted on the web that day. What's interesting. So I would say 50% of the daily

00:21:26.400 | emails will point out an article that I ended up reading on the web because they published like a

00:21:31.040 | new article every day on the web. Uh, for the magazine, my rule is like, you always have to

00:21:35.520 | read one and if there's more, sometimes you'll read more than one, but you always have to read

00:21:40.000 | one. And so that forces you, even if it's a day where the five articles, none of them are right

00:21:44.080 | in your sweet spot. You read about door-to-door salesman or something else. It's really interesting.

00:21:47.600 | And so that's my role. Read the daily email to see what catches your attention. That's where a lot

00:21:52.640 | of my stuff is featured. So definitely read that. And then you got to read one, one per week,

00:21:56.640 | at least. I have a similar rule too.

00:21:59.520 | Yeah. I just look at the table of contents and then circle two. And then

00:22:02.880 | if there's two I like, then I read them. And then otherwise,

00:22:05.280 | and then it breaks the seal and then you end up reading more a lot of times, but like, you always

00:22:08.320 | have to read at least one. Oh, and then the third rule is you have to leave the magazine out kind

00:22:13.280 | of prominently in your house or apartment and like, Oh, when people come over, Oh, sorry, let me just

00:22:17.840 | clean this up over here. I just was, you got to put it next to your copy of the Harper's

00:22:22.400 | and the Paris.

00:22:24.400 | Do you recycle them a lot quickly thereafter?

00:22:27.520 | Yeah. It was just when the pile grows, I suppose.

Why TikTok’s Sneak Attack on Facebook Matters | Deep Questions With Cal Newport

Chapters