Good morning,

I am very excited to welcome back Eugene Wei for a Stratechery Interview. Wei worked for Amazon at its very beginning, went to film school, was an executive at Hulu, was head of product at Flipboard, and Head of Video at Oculus. He writes truly groundbreaking essays at Remains of the Day, his personal blog. I previously interviewed Wei in March 2020.

Earlier this week, Wei and I had a two-and-a-half hour conversation about streaming and social media; sadly, I stopped recording at the 90-minute mark (naturally the last 60 minutes were the best 🙂). Even so, this is the longest Stratechery interview I’ve ever posted, but I think you will find it fascinating throughout.

To listen to this interview as a podcast — which I recommend! — click the link at the top of this email to add Stratechery to your podcast player.

On to the interview:

An Interview with Eugene Wei About Streaming and Social Media

This interview is lightly edited for clarity.

The Netflix Scale Argument

Eugene Wei, it is great to have you back. It has, number one, been too long. And number two, you have not written one of your just incredible pieces on your site for quite a while. I kind of had a sense that you have a bunch of takes to get off your chest and I wanted to make sure and give you a platform to do that. It’s great to see you again.

Eugene Wei: It’s great to talk to you again. It’s been too long, indeed. The pandemic really just evaporated a bunch of time.

Yeah. It’s so weird how that happened. It both made time seem to last forever. It also completely disappear all the same time.

EW: Yep.

I figured would we talk about two big things that are in your wheelhouse. Number one, streaming, and number two, social. We’ll split it in half. Unlike most interviews, I don’t have a super in-depth range of questions prepared for you because I know we’ve talked about this stuff a ton already. We can riff and go back-and-forth, and I want to make sure to give you the full canvas to get your takes out there.

So let’s start with streaming. When you say streaming, the company everyone immediately thinks about is Netflix. To say that it has been an adventurous year for Netflix is definitely the case. But I’m curious, when you think about Netflix and all the stuff that’s happened this year, adding advertising, subscribers actually dropping, all of these challenges that they’ve faced, is that a recent development, or how much of what has happened this year actually goes back to Netflix’s founding and their start and all the things they did from the beginning?

EW: Yeah, I think a good way to think about this whole question is to go back to the early days of Netflix, but after they had launched streaming and they had started becoming ascendent.

Right. Netflix 2.0, basically. The second version of Netflix, it’s like Apple. The ’80s were fun, but we actually care about the iPhone era.

EW: Right. They got the jump on everybody in being pure play streaming. They were growing faster than everybody and they really started from a base of very little. We know now in hindsight, they were able to leverage a lot of licensed content from third parties to give themselves time to bootstrap an originals business of their own, and become really the largest streaming service out there. At that time, I think a lot of people were arguing the same thing, which is that because they had the most subscribers, they had a scale advantage over everybody else, and every dollar they spent on content could be spread across a larger base of subscribers. It was the right thing to do.

I feel personally indicted!

EW: It’s very logical. I began my career at Amazon and I worked in strategic planning in the early days. The thing that we would tell all of our investors is that Amazon is an economies of scale business. Meaning that, at some point, if we just got enough customers, we would be able to cover all of our fixed costs, which were a substantial part of the business — building distribution centers, maintaining a website and the catalog and all of that. We had variable costs, of course, but a huge fixed cost space. We could show people that, “Hey, as long as we get enough market share, we will be a profitable business because our unit economics are positive.”

I think it made a lot of sense at the time to think, “Okay. Netflix has the scale advantage.” If they just outspend their competition on content, they will have the ability to be maybe one of the first streaming services on earth to be everything to everybody. They’d have some amount of content that would appeal to a whole wide variety of subscribers, and that for their competition, it would be very hard to emulate that because their cost base would be much higher.

Now that we’ve reached 2022 where we’ve seen their subscriber growth shoulder off, it’s a good time to reflect back and see if it played out as everybody hadn’t anticipated. I think in some ways, it did, and some ways it didn’t. It may be too early to even judge.

Yeah. Yeah. We’ll get to the too early bit because I think that’s actually one of the most interesting questions, but I’ve been very self-critical on this point because I made this argument and not everything played out as I expected for lots of different reasons, but I’ll let you go first.

EW: Yeah. Sure.

What are the ways that it did play out in the way that it was expected?

EW: Well, I certainly think the parts of the advantage they have in terms of the large subscriber base is that it has been a huge platform for them to be able to really slingshot a lot of their originals to just massive viewing numbers. Not just originals from the US, but shows like Squid Game and other things that they can take from other countries and bring to other markets. You just put it on the homepage and that’s just automatically a massive, massive viewer base on opening weekend. That’s one way that they’ve used their scale really well.

Again I go back to my time at Amazon. Jeff Bezos used to say this thing to all of us whenever we would have all hands and we would fret about the competition, he said, “Hey, just focus on the customer. If you just create a great customer experience, it really doesn’t matter what your competitors do.” I used to subscribe really hard to that theory. It was such a clarifying thing for your team. But in the entertainment space, I’ve come to rethink that. I do think it matters what your competition is doing. One of the reasons is that as you know in the entertainment space, which I would put Netflix and all the competitors in, we’ve reached the state of abundance, which is really unique in the history of the world. There’s just more content now than any one person can watch, even if they were awake 24/7 for years on end with their eyes just plastered open.

Yeah. Well, this is the Internet effect. Right? If you were in the ’60s or ’70s, it was about distribution. If there was a show on TV, it was basically already popular because there just weren’t that many choices. That’s like the newspaper thing. Right? The newspaper was the newspaper. But when you’re online, there’s an infinite number of things to watch and it really flips the dynamics on its head.

EW: Like most people of my generation, you go to school. You’re studying microeconomics. You’re studying supply and demand curves and a lot of classic microeconomics, but where the Internet has made things tricky, especially if you learn traditional business strategy out of business school, is that when you go to a condition of infinite supply, you really get to this weird boundary state, which changes a lot of the conventional strategy and conventional wisdom. In a state of infinite entertainment, the customer’s attention is now the scarce resource and managing that is very difficult.

One of the things about entertainment that you noted is that in the past every show really is unique. Every movie really is unique. You could say this movie is one of one, but in a state of infinite content, I’d argue that a lot of content that is unique has been pushed into the category of a commodity. That reduces the leverage you have with that content to have pricing power. Now, a company like Disney is interesting because they really do try to go for a very specific type of IP that gives them extreme pricing power in the marketplace and they’re probably the best…

Also, pricing power over the talent as well.

EW: Exactly. They’re probably the best at doing this and they’ve made a number of great acquisitions and they’ve executed really well on Marvel, maybe a little less so on Star Wars, but still you would say, “Hey, they have the most unique IP catalog.” They, to me, form the most interesting contrast to a Netflix, which has a lot more content than them, but I think a lot of Netflix content, people tend to just lump into maybe not a category of a perfect substitute, but close enough. A lot of that content, it’s like, “Well, if I didn’t watch this and I watched something else to fill that time, it’s about the same to me. I don’t know if I have to subscribe to get that.” That’s made Disney and Netflix probably the most interest interesting pair to compare and contrast.

I agree. There’s two points that I would add on to what you’ve said. Number one, this bit about business school, when I went to business school, and the first class was like Strategy 101 or whatever it was, there were no Internet companies. I asked the professor “How is this possible?” and the reply was that “Oh, well, the general principles of business apply to all businesses, blah, blah, blah, blah.” I’m like, well, technically that’s true, but when distribution is zero, it’s like if you have a complicated math problem and you put a zero into it, a zero wrecks havoc. Right? It screws it all up and the end equation looks a lot different. That’s number one.

The Experience Curve

But then number two, I think a mistake that’s made in analysis generally — and this can be sports analysis, it can be like politics analysis, it can be anything — is lumping stuff too tightly together, which gets to your bit about content is not content. There are different kinds of content. There are gradations of content. If you use one assumption about content for business and that assumption’s wrong, it’s going to flow through to bad analysis generally. I think your distinction about really just filler or mid-tier content or whatever it might be, versus there’s only one Marvel, I think is a super important point in this regard.

EW: There is a book I read recently called Lords of Strategy, and it’s about the rise of business strategy through the history of four consulting firms. It was like BCG, Bain, McKinsey, and then kind of like Michael Porter and Monitor as the fourth.

Yep.

EW: One of the interesting things in there was the BCG experience curve, a framework that was the start of the business strategy revolution. Back in the ’60s through the ’80s, BCG would go around and win clients based on this idea that, “Hey, if you can gain market share fast, you can move up this experience curve faster than your competitors. You would become more cost efficient.” A lot of people argue whether that really is true or not, but I think I bring it up because it’s an interesting thing to think about. Netflix being the first streaming service to achieve 100 million and then 200 million subscribers scale. Where did that benefit them and where did it maybe not benefit them as much?

Famously with the experience curve, BCG had taken on a number of paper-making and textile companies as clients in the ’70s, and they pushed their experience curve thing. A bunch of companies, they invested in a bunch of paper-making machines. They were more efficient. But when they drove a lot of the competition out of the market, it turned out their profits weren’t as high as they anticipated. One of the main reasons why was the price of paper was really being held up by the least efficient plants that their competitors ran. When those plants went out of business, the price of paper just collapsed generally, so there were just less profit.

Yeah. Differentiation matters, it turns out.

EW: Understandably, Netflix, when they started and they didn’t have many originals, they priced their streaming service really low. I can’t remember now what it was, like $7.99. They got scale, and then their competitors came in to chase them, so Disney. When Disney+ launched, I think it was $6.99, which was really cheap for their entire catalog. Amazon Prime video for most Prime subscribers doesn’t even cost anything extra, and when it launched they hadn’t raised the price of Prime. Everybody, of course, was chasing Netflix, and because they had smaller catalogs, they didn’t feel like they could price higher than Netflix. They were also trying to just get some amount of scale considering how much they were going to invest in the business.

However, as you know, one of the things that economics teaches you is that prices are sticky. It’s crazy because if you were to go back in time to ’80s me, ’80s me knew every show on every channel on each day of the week. I could memorize the entire TV guide, and if you presented me with a catalog of shows that is on Netflix today, and you said, “How much would you pay for this thing?”, I would’ve said, “Oh, my God. I pay so much for cable.” I would easily pay that price probably to get all of that.

But what’s funny is prices are all relative values. It’s hard to put a value on these things, but customers have been trained that the price anchor for streaming services is really low. In a way, that’s why it does matter to pay attention to your competitors. It does matter to pay attention to the environment that you operate in. Because in a world of infinite entertainment options with competition from video games and TikTok and podcasts and all of that, I think we’ve ended up in a place where maybe Netflix is going to continue to have the scale advantage, but maybe the ultimate terminal value of that is less than we would’ve anticipated.

I built a lot of streaming models when I was at Hulu, where we had to build revenue and growth models. I have a lot of friends who work at all the various streaming companies. I know that everybody has a model that drives them to profitability, and it usually triggers off of two variables. In one model, you just grow your subscriber base to a large enough size that even at the current price, you can cover your content spend and become profitable. Maybe back in the day for Netflix, that was, “Hey, we’re going to get to a billion subscribers globally, and we’re going to be just rolling in money”. The other is that you reach some terminal number of subscribers, but then you start raising your price per month. You’re like, “Okay. We’ll raise the price $2 a year this year, and then the next year, we’ll raise it another $2 and eventually we’ll reach profitability.”

However, I think because of the competition, there’s been this kind of red queen effect in streaming. Now, we’ve seen Netflix, they raise their prices and they had a few quarters where they lost subscribers. I had a lot of friends who were like, “Oh, gosh. I have so many streaming services. I don’t really have to keep them all going at once. I can just cancel this one for a while.”

Being super easy to cancel is a big deal, but that’s only half of it. The other bit is a price increase triggers an evaluation by the customer, whereas if you stay at the same price, they may never think about canceling, not because they’re not getting value, but because it’s never entered their evaluation space in their head. I think that’s the real challenge there is that actually none of those customers were really getting the value that they wanted at $13 a month or whatever it might have been, but they never bothered to think about it. But once they went to $15 and they sent out that notification email, then yeah, it’s like, “Well, you know what? It is super easy to cancel. I’m not locked into a contract”.

EW: Yeah. In the book 7 Powers, the very first chapter of 7 Powers is about scale advantages. They use Netflix as the example of a company with a scale advantage and the author talks about other various scale advantages that you might get with scale. It might have the advantage in distribution network density, so local distribution networks. You have more density. You might have more efficiency than your competition. But this doesn’t really hold in streaming because you’re using the Internet. There’s talk of learning economies, “Hey, you just learn faster than your competition because you have more reps.” I also would say this doesn’t seem to have held for Netflix. Their hit rate on shows hasn’t necessarily been so much better than their competition.

Aggregators and Differentiated Content

There was so much talk about how they were either going to use data to drive shows, and that hasn’t really played out. I think what you’re getting at, this goes back to the content point about there being different types of content. Aggregation is where scale really pays off on the Internet, but those manifest themselves in a commoditized market. Google is the ultimate example where every search engine in the world has access to the same corpus of data. Commoditized markets is actually the area where these other scale advantages can pay off, because the way to avoid an Aggregator is to have highly differentiated content. Basically, in a content space, a non-commoditized space, there is a degree of power that is retained by supply. I think one of the takeaways that these Aggregators are actually fairly fragile in that everything has to be accessible. That loop has to be very, very tight, and if there’s sufficient power in supply, it can break down.

I wrote about Netflix versus Spotify earlier this year. One of my takes was that while Netflix is a bigger business than Spotify, and there is no question it should be worth a lot more, there is an argument that from a strategic perspective, Spotify’s actually in a better place than Netflix, because their content is commoditized. It’s very expensive content because the music labels own it and they take a big chunk of Spotify’s revenue, but Spotify has the same catalog as Apple Music, as YouTube Music, as everyone else. That’s actually better if you want to be sort of an online Aggregator. You want commoditized content, which I think is fairly counterintuitive where you would think that, “No, need to have differentiated content.”

EW: That’s absolutely right. One of the things that a lot of people complain about these days, I’m sure you’ve seen the tweets and the jokes about it, is that, well, after you add up all the streaming service prices that you pay today, it costs more than the cable bundle used to or the same amount. Well, one of the reasons is just that the unit economics of producing premium content hasn’t changed. It still costs a million dollars plus per minute to make something like a Marvel film. Brad Pitt still gets paid whatever he gets paid to be in a movie. There hasn’t been some technology revolution where the cost of content production has gone down by huge amount except in UGC…

We’re going to get to UGC in a moment because I think there is a big contrast to draw.

EW: Yeah. For sure.

Sorry. For the listeners, UGC is user-generated content in case people didn’t catch that.

EW: Right. But this is a challenge, and the other thing that’s interesting in this space when it comes to competition for Netflix, is that because it’s not a commodity product with content, you can, as a competitor of Netflix, get away with having a much smaller catalog and still building a pretty healthy scale business. We already talked about Disney, and Disney is obviously the prime example of this. When they launched Disney+, they didn’t even have all of the Marvel movies or Star Wars movies. They had licensed some of them to other services. They couldn’t get them back, but anyone who has kids, it’s just like, “Look, I’m subscribing to Disney+ for life, it doesn’t matter how much it costs because my kids watch it.”

Well, one of the brilliant things about kids and streaming services is that it changes video to be more music-like. One of the things about music that makes it better is you listen to the same song again and again. And that’s also why the music labels have fared so well relatively speaking, because it’s the back catalog that’s more important, and the magic trick is the moment a song is released, it’s instantly back catalog. So their advantage is compounding over time.

EW: Yeah, it’s so true. Right now I’m staying at my sister’s, and I’ll sometimes do movie night with my nieces and nephew, and it’s amazing how many times they’ll watch the same movie. I’m like, “We just watched this yesterday, why do you want to watch it again?” And it’s true, one of Disney’s advantages really was that they capture the customer so young, at a point when your aesthetic tastes are not fully formed, and as you noted, you have just this huge tolerance for repeat viewing.

Desire, not just tolerance.

EW: Yeah, and you capture a customer for life. If you’re doing a LTV model of a streaming customer, you really want to get that person at aged like five or six, they’ll watch that thing for life-

70 years, yep.

EW: It’s like my attachment to the original Star Wars trilogy, it’s probably irrational at this point and I’ll never again see anything that I have that attachment to, but that’s just how the life cycle works.

The last thing that I’ll note here that’s a challenge for Netflix is, you know the whole commoditize your complements idea and tech? When I was at Amazon, I first read about this in Joel Spolsky’s blog, he talked about it, and a lot of people in tech talk about it. And I come from both worlds, I’ve worked in tech but I also worked in the creative business, I went to film school, and the streaming business is an interesting intersection of two different cultures. So, one, the culture of cultural production, that whole industry of Hollywood, they treat their art as a business, but they also derive a lot of pride and they put a lot of value in the creative work they put in the world. The number one complaint you hear from people in that business is that everybody in tech refers to it all as “content”.

Which is kind of the point we made a little bit ago, yeah. That’s exactly what they’re pushing back against, and they have a point.

EW: Yeah, they absolutely are right because if you are a tech platform, and your value comes from owning the platform itself, you really don’t want your suppliers to have any pricing power over you. You want your content to be commoditized.

And that was the Netflix value prop, that they were commoditizing content. And if you want to put it in a nutshell, the whole commoditize your complement argument is that there’s a value chain, it’s like a balloon, if you squeeze out all the value from some parts of the chain, all that value flows to you. But this whole bit about content is that because it’s retaining so much value, by definition, there’s less value that you yourself can capture.

EW: Right, and that’s always been the tension here. That’s the push and pull that you have. This is where I think Amazon Prime and Apple TV are interesting, because if you ask them, they will just come out and say, “Hey, we get value from Amazon Prime Video by having those people buy retail products as well”. So in a way, the video business doesn’t have to stand alone as its own P&L. And Apple TV right now over the past year, has been spending like a drunken sailor and outbidding everybody on huge projects. There’s the Brad Pitt F1 movie…

Which I’m very grumpy about because I want to see that in a theater, so hopefully they release it in a theater first.

EW: Yeah. They outbid everybody for CODA which won best picture last year at Sundance, and I think they set a record at Sundance for how much they bid. And again, they can justify it by saying, “Well, look, if it sells more iPhones and iPads and whatnot, it doesn’t really matter if we turn a profit on this content.” But the whole thing that the culture industry argues about what tech is doing, is effectively being done here where they are trying to drive that content down. I mean, I don’t know what is Apple TV+ per month? It’s like $4.99? I don’t even know what Amazon Prime Video costs now as a standalone, but it’s very low.

Yeah, it was $4.99, but they are raising the price to $6.99. Still, it’s very cheap to get an Academy award-winning movie for less than a movie ticket.

EW: So I think this will be the tension here, and where we ultimately may settle, to come full circle, is that Netflix may still end up with the scale advantage in the end. It may just be less valuable than it was previously, and all of streaming video may just earn less revenue than it once did before. And I think in Netflix’s most recent earnings call, or in their statement to shareholders, they finally had some positive news to share. And one of the things they said was, “Hey, look, maybe we still do have a scale advantage because we estimate that our competitors lost something like $10 billion over the past year, and we turned a $5, $6 billion operating profit. And we do know that Disney and a bunch of the other companies, HBO Max now that it’s part of Discovery, are focused more on profits, and cutting costs, and trying to be more rational.”:

So this was always my thesis. I wrote back in 2019 I think that, “Look, the next five years are going to be terrible for Netflix because you’re going to have this influx of competitors, but at the end, all these companies are going to realize this is a terrible business to be in, because there are two costs: it’s not just the cost of building and maintaining the streaming service, it’s also the opportunity cost of not selling your content to Netflix.” Like Sony’s sitting there just raking in the money because they didn’t build their streaming service, and so they’re standing alone as the only content that’s available to buy and getting crazy prices for it.

Now, obviously, that timing was wrong because the pandemic came along and exploded the streaming market, but this is where I’m not sure because, to your point, that might still happen. You might still get the case that services like Paramount+ I don’t understand why it exists. It blows my mind they’re putting sports on it and devaluing the overall bundle. Or Peacock, I guess is the one that I’m thinking about. You are a cable company, why are you killing the bundle? So it might still turn out that at the end, Netflix is one of the last companies standing. It will rationalize and then they can buy content at arguably lower prices, but I think this bit about what is definitely clear is they don’t have the power to commoditize content. They might be the last buyer standing, but they’re probably always going to pay more than those models what sort of suggested they would.

Building Franchises and Binge Watching

EW: Yeah, right. And I do think one of the advantages that Disney has always had, I think I talked about this last time I was on with you, was just that their monetization rate per unit of content is so high with merchandise and theme parks. Again, there’s an advantage here too, selling things to children who have an insatiable appetite for all of that attendant paraphernalia. But I do think there’s room for Netflix to both be last ones standing, and to start to just understand, ultimately, to try to get better at doing some of the things that Disney does with getting ancillary revenue streams from Stranger Things, or whatnot. Because one of the challenges of the Netflix business model, and all these streaming services, is that your revenue is capped by your monthly subscription fee. So no matter how much a customer watches, how much they love Stranger Things, you don’t earn any more from them.

Where that becomes a challenge, and I think this is a fair criticism, is that a lot of franchises and shows on Netflix have been canceled after a season or two. Typically with shows, the more seasons you run, the more your costs are because talent contracts typically get renegotiated. Like the last season of Game of Thrones cost a hell of a lot more than the first season of Game of Thrones when the actors were unknown. And so Netflix has this weird situation where if the show goes on a number of seasons and is fairly popular but has a declining viewership, the costs go up while their value to their catalog goes down. And so you’re more inclined to pull the trigger on canceling a show, but conversely, that makes it harder for you to generate the kind of durable IP that typically results in merchandise and other ancillary revenues.

This might be one of those quotes that is falsely attributed to Walt Disney, but they said that Disney used to say, “Hey, you have to remake all of our animated movies every 10 years or re-release them.” When I first worked at Amazon and we were just selling VHS and DVD, you couldn’t always buy Snow White or Pinocchio on VHS tape. Disney would just release it for a month and then they would yank it off of the market. So they were very good, the entire business of Hollywood was, at generating artificial scarcity for their product, and sustaining that demand over many years. And now you see it with live action versions of all their films, and then they put them on Broadway. They’re trying to treat each of their IPs as kind of an annuity.

In contrast, I feel like a lot of people right now consider a lot of Netflix content to be more of the disposable variety where the half-life is very short. And it’s one of the weird things about the entire cultural content business, is that in a way they devalue their own products by focusing so much of their marketing on new content and head content.

Oh, this was the big thing about Netflix that I definitely had wrong. Because when you would evaluate a company, I always considered the cost of this large original content catalog as not being just the cost of goods sold, it was also marketing spend. The idea was that for your marginal customer, when they sign up to Netflix, they don’t just get the new content, they get all the old content as well, and so the marginal value of Netflix is increasing, which is exactly what you need for the marginal customer who’s perceived value of the service is sort of decreasing as you go down the curve, and you have a nice sort of crisscross there.

What I think has happened that I definitely got wrong about Netflix, is it turns out that a lot of their content really is just cost of goods sold. Like it’s there and gone, and no one signs up to Netflix to watch a show that’s a few years old. And I think your bit about cancellation is super important here, because if you’re bored on Netflix, and you find some show that had 10 seasons, then you watch it, that takes up a week of time, or two weeks of time, or three weeks of time. And you’ve now developed this expectation that Netflix has this treasure trove of adventures, but right now, why would you want to go watch a season a show that was canceled after two seasons? There’s no real payoff there.

EW: Yes. I think one of the things that I’ve learned over the years in working a bit in the media and streaming side of things, is that one of the ways you need to model each of these pieces of IP is as a social network with a network effect. This is most apparent when it comes to things like Disney shows, which are built on this globally popular IP. But it matters for other shows too. When we used to talk about the water cooler and talking about shows at the water cooler, we really were talking about a social network, right? It’s about being a member of the cultural discourse. The cultural discourse is driven both by what other people are watching, and by what is being pushed by those companies themselves in billboards and advertising everywhere. And the goal, I mean, ideally you have IP that has a very strong network effect. So, the network effect of Star Wars as an IP, or Marvel as an IP, is that I can talk to almost anyone in the world and have a great fun conversation about, I don’t know, Avengers Endgame, or the latest Star Wars movie, or whatnot. That has value.

Even if it sucked, right?

EW: Yeah, exactly. That goes above and beyond just the raw entertainment value of the show, is that it’s a unit of cultural discourse that you carry around with you. And I think there’s room for Netflix to probably get better at this part of things.

This goes back to the whole release strategy, though. You talk about this with their content, but one my criticisms of Netflix is they have an all-or-nothing approach to everything. It’s like we’re definitely having no advertising, or we’re definitely having the exact same price for XYZ. Or we’re definitely going to release everything all at once. And that doesn’t make sense to me because you look at something like — I’m just not a huge TV watcher, it’s one of my keys to productivity — but you know what I watch as soon as it comes out? I watch House of the Dragon because you know what? It’s so prevalent in my social media experience, in my group chat experience, that I feel like I’m missing out if I don’t watch it. And to your point, you just said it, there’s a social network effect around individual pieces of content, and Netflix is forgoing that for all their content because they have a hard-and-fast rule.

Now, I do think you can argue, like Squid Game for example, because it was an unknown IP, the binge model does help explode something into the popular conscious, because you can just watch it all at once and you can talk about it. But for known stuff, Stranger Things is always the go-to example, why wouldn’t you want a weekly release schedule? Why wouldn’t you want the Ringer having a dedicated page to write about every episode, and do reviews? And they’re forgoing all that, and it feels like they should have a more fine-grain approach so that some stuff is a binge-release approach, and some stuff is weekly.

EW: Or even a hybrid approach where you release the first three episodes or something, enough to get people into a cliffhanger, or something like that, and then go to weekly release. This is classically one of those hard things to A/B test, you can’t really A/B test on a show because no one’s going to want to be in the test that doesn’t get it.

Well, because every piece of content is different.

EW: Yeah. Yeah, no one’s going to want to be in the cohort that doesn’t get to watch it all at once. It’s hard, I think this is an easy win for them to extend the value of their content spend, and increase the network effect around some of their properties, which will open up more potential ancillary revenue streams related to that. I think with some of their movies, I really don’t think it would hurt them to throw the movie in a theater for like a month.

Theaters are valuable. It builds up perceived value. You see the advertisements for it, you see people talking about it, and then even if you don’t want to go to a theater, then you’re super excited when it comes on Netflix — that marketing spend pays off over time.

EW: I think it would make talent happier. And for a lot of people who just have Netflix on in the background all the time, when you finally do get to see, I don’t know, Gray Man or whatever on Netflix, you’re like, “Oh yeah, that was in theaters,” it just feels more valuable to you in some way.

Yeah, that’s right, it does.

The Scarcity of Time

So one of my long-term criticisms about Netflix is they’re well placed to the extent that content is a commodity. To your point, they have scale, they have great customer service, they have great technical infrastructure, all those sorts of pieces. But the problem is that they have this content issue, which is they have to pay for content. And if you’re going to pay for content, having the better content is obviously more valuable and interesting. And competitors like HBO or Disney or even Apple TV+ because they don’t care about how much they pay, can have sort better content. So you have this problem on one flank, which is competitors with better content.

I think the real challenge though is that if that was the only world of content, Netflix would still be fine by being the commodity provider. But on the other side is UGC — user-generated content — and that is the true commodity content. UGC platforms get all the content for free. Now, there is certainly an aspect where they can have differentiated content, like TikTok has the best short form video, there’s just no question about it. And my argument is actually Instagram should welcome more TikTok video because they need to leverage their demand base, not their supplier side. But they don’t have to pay for it, and so they just have a fundamental structural advantage relative to Netflix.

One of the things I thought was really interesting about Facebook’s results, is everyone has this assumption that TikTok is eating away at Facebook engagement, is taking away from Facebook usage. Facebook’s apps are up, they’re up in users, and they’re up in engagement. What TikTok has done is dramatically expand the pie for user-generated content, and that time spent on UGC is coming from somewhere, and the obvious place it’s coming from is scripted content, is paid-for content. And I think that’s probably a very underrated aspect of what’s happening to Netflix. I mean, it’s very enjoyable on a Saturday night when you don’t feel guilty about it, to watch TikTok for two hours. It’s incredibly compelling and relaxing. And that’s Netflix time that’s not being realized.

EW: It’s funny, you remember how Netflix used to say that our number one competitor is Fortnite, and then it later said it was sleep, and all these things. I think the general form of the argument holds, which is that one of the things that the Internet has done, and I think the smartphone is a huge part of this, is it has collapsed the barriers between different mediums of content, and made them all accessible to you at all times. And I always say that when I grew up, all these different mediums were, to me, spatially segregated. Meaning that I listened to the radio in the car because the cassette player or the CD player was there. I played video games in my bedroom because my computer was there. I watched TV in the family room because that’s where the TV and the VCR were. And I saw movies in theaters because that’s where we went to see movies.
I think if you grew up in those industries and you were running those businesses, you could easily become very complacent about competition from other mediums. I’ve seen this when I’ve gone and done talks at various entertainment companies, or the sports leagues, and everything. You’ll bring up something like a video game like Fortnite, or you bring up something like Twitch, and they’re like, “Well, oh, that’s different, that’s not the space we’re in. We don’t do social features and we don’t worry about it.”

The problem is from the consumer standpoint, you now, and this is kind of what you’ve talked about in the past with the newspaper forcing a bundle of news on you. Like with the newspaper, once the internet came along, if you’re like, “If all you care about is sports and the funny pages”, now you could consume that all the time.
With entertainment, let’s say you’re like, “Hey, I only like entertainment where I can play along with my friends. And Fortnite is the best option. I can’t do that on Netflix or HBO Max”. Those companies will argue, “Well, that’s not the business we’re in”. But that’s too bad, that’s just the way the modern world is.

Yeah. With the disappearance of geographic-driven differentiation, which I have always talked about with regards to newspapers, but you’re talking about it in a very micro sense, like even within the home, which is a phenomenal insight, and also the cost of distribution going to zero, the only scarce resource in the world of digital content is time. And so by definition, anything that’s a full screen experience, which you’ve talked about in the context of TikTok, right, that’s a video that’s taking over your entire phone. It’s not in a feed, it’s your whole phone is that experience. This is the zero sum game part. It is zero sum for every other entertainment experience in the world.

EW: Absolutely. And I think it’s very hard because you also have this, I call it “hindsight blindness”. People say hindsight is 20/20 but I think if you look at traditional media companies, sports leagues, all of these, they tend to be run by older executives who just don’t live and breathe younger mediums. They don’t feel the threat of the competition in the way that if you’re a kid and you go to high school today, and all your friends are on TikTok and you’re talking about the latest meme.

The water cooler conversation is around TikTok, it’s not around the sports game.

EW: Yeah. It’s very different, and it’s just very hard for them to react to something like that. And so we’re in a situation in the entertainment world now where you compete with the strongest aspect of entertainment from every competitor in the world. If one competitor is best at social experiences, that’s the bar for what the social experience is. And if one of your competitors is best at doing whatever, super premium IP like a Disney, that’s the bar for what super premium IP is now. And that’s just a very difficult thing, but that’s the nature of the internet and the democratizing effect.

Right, why would I read my local sports columnist, if I can just read Bill Simmons, right? Yeah, it’s great if you’re at the head of the curve because the whole world’s your addressable market. It’s terrible if you’re not because the whole world is your competition.

EW: This is the squeeze that you feel. On the one side, you’re like, okay, because of this wealth of entertainment, the value of any incremental piece of content, the median piece of content goes down, and at the same time, the bar for competition goes up. And it’s tough because when you have a lot of execs who’ve also been running businesses for a long time, they get themselves anchored on unit values for their content that just won’t sustain over time. And when they go out to market, they’re like, “Well, this is always been worth X, this is what I want.”

It was very hard for me when I was at Hulu. I dealt a lot with people on the studio side, and I could just feel it. We were living in different eras. And specific, I think, to the entertainment business is the nature of it, is that a lot of the execs there were optimizing, rightfully so, for the incentives of their career, which were short term in nature. When we were at Hulu, we’d complain a lot, for example, when our parent companies would license content to Netflix, but for them they’re like, “Look, I don’t own Hulu stock. And so if I can get this deal done and the licenses show to Netflix for the next few years, I’ll be gone by the time this deal’s done.” So these are all the kind of weird micro incentives that make the streaming space kind of a fascinating thing to study right now.

The other thing is cost structures. All of these entities have cost structures that support geographic domination. I think maybe the go-to example going forward is going to be sports. You have this entire structure that’s predicated on volume, particularly when it comes to baseball, basketball and hockey. Those are the challenges where it started out you got most of your revenue from attendance, so the more games the better. And so they have these super long seasons. Then you added on RSNs where the value was the churn contribution or marginal churn contribution, in that you had to have a RSN to prevent churn, and so they could increase price. But an RSN is valuable through volume because they have lots of games and lots of volume.

Meanwhile you have the NFL, which if it was possible, I’m sure there would be an 80 game NFL season. It’s just not physically possible. The human body would fall apart. And it turned out that set them up so much better for this role of abundance because there’s a scarcity of NFL games. There just aren’t that many. The importance is much higher. And so their valuations are going through the roofl, and basketball in part, they’re holding on because ESPN and Fox are betting on the sports bundle sort of strategy. And so they’re retaining value. College football’s retaining value. But you see something like Bally Sports is going into bankruptcy. The whole RSN part, it’s falling out.

However, you have the players who get 50% of revenue who have the expectation going forward of ever increasing salaries. Your cost base is going up. And of course the NBA should transition to a much shorter season, should transition to feature games that are much more importance. But they just fundamentally can’t because of their revenue model that’s in place.

EW: And it makes their next national TV contract renewal negotiation in 2025. Really a fascinating thing to study.

I think they’ll get a lot for this one because Discovery is over the barrel because they still need the cable bundle revenue, and so they’ll pay up. But I think the one after that is this is going to be good.

EW: The one after that is going to be very hard. I absolutely agree with you that the NBA is caught in this trap. Honestly, if we went to, I don’t know, a 50-something game season or something like that, maybe you play every other team twice. Each game is a little more meaningful in the modern entertainment environment. That’s probably a much better thing in the long run. But like you said, these near term incentives make them so that they’re structurally sort of trapped.

And players by definition can’t have the long term outlook because their careers are short.

EW: Right, yeah, exactly. That RSN flip is such a crazy one. I’m from Chicago, I’m a Cubs fan. And the Cubs tried to do what the Dodgers and the Yankees had done, and get back their local TV rights and they just got caught on the timing. By the time they were able to do that, the RSNs were reeling and no longer had the leverage that they once did, and so the rights were worth a lot less. And I think the NBA also has this challenge that the NFL doesn’t, in that the NBA gave away local TV rights to the local teams who got a lot of money for it and sold them off.

As for the NFL, I don’t know whether it was through luck or thoughtfulness, retained those and was able to package them together into NFL Sunday ticket and NBA and Major League Baseball are trying now to buy those rights back. Or maybe they’ll have to buy the RSNs. I know they’re trying to contemplate that, but it’s tough when you have deals on that kind of time scale and the Internet comes along and the world changes on a faster time scale.

It’s a real challenge.

TikTok and Instagram

One of the things that you wrote since the last time you’ve been here are these three articles about TikTok, and I think one of the really important insights is that Facebook didn’t see that TikTok solved the cold start problem, not by spamming and trying to get content and then pushing content from people you knew, but realizing that there’s content all over the network and now compute power is such that we can sort of push it from anywhere. There’s a question here about is US social stagnant and thus a foreign entity could break through, although I think the role of censorship is actually under-appreciated, where TikTok is obviously an outgrowth of ByteDance in China, where they had to be very careful about what was spreading, and it was better to have a centralized content store because then you could censor it more easily as opposed to peer-to-peer sort of sharing. It turned out though, that that was actually the way to build a new network. TikTok’s called a social network, but the social aspects don’t really matter in the traditional sense, in a sort of a Facebook sense, it’s an interest network. It is a UGC network.

I’m curious though, sort going forward, it’s been really fascinating to see Facebook respond to that. I feel like in retrospect, Stories saved Instagram in many respects because what’s interesting about Stories is it became the dominant aspect of Instagram while only taking up 20% of the screen space. And so that lets them retain that part of Instagram and then repurpose the feed to build up their Reels product. But I mean are they going to pull it off? Are they going to be able to mix and match your regular social with interest stuff?

EW: It’s a good question. I don’t know the answer for sure. We were talking earlier about streaming and how prices are sticky. Well, there’s a way in which brand is sticky also, especially when it comes to a social network. You saw the backlash to Adam [Mosseri]’s announcement about a number of changes to Instagram, and they had to roll some of them back. And look, it’s understandable why he was pushing for some of those things. But it goes to show that the brand of a social network is half owned by your users and half owned by you. I could argue all brands are like that, but in social it’s even more salient because users have come to have a certain expectation for how something works. And one of the odd things about western social is if we roll back to the beginning and trace the evolution, there’s one sort of dominant architecture that has come to be prevalent in all of western social media. That’s the feed, the singular feed in which all content is pushed. And that feed is governed typically by an engagement-based algorithm. And the companies monetize through ads that they place in that feed.

The other fateful decision, I would argue on top of that, is that western social media companies chose to be fairly promiscuous with what they put in the feed. Nowadays you see stuff from people you don’t follow. If you go back to the earliest days of social in the west, the feed was deterministic. It was just reverse chronological stuff from people you followed. Contrast that to today, western social media is probabilistic, the distribution is unknown. You put something out there, you have no idea who’s going to see it, whether your friends who follow you will even see it. That has put users and these companies into a weird relationship that’s somewhat almost like a game that is semi-adversarial, and I think it’s hard to under overstate just what a sea change that was for western social. That choice was made. And I think it’s sort of inextricable from advertising as the dominant business.

I was going to talk about this way back in the context of Netflix, there’s a bit where advertising goes hand in hand with an overabundance of content that you have to sort.

EW: So we live in a world where distribution is probabilistic based on engagement. And the way I like to think about all social media services, and especially I thought this after TikTok came along and I was studying it, is I think about every social media network as an evolutionary system. I went back and read Origin of Species and a bunch of books on evolution. There are three elements to any evolutionary system. One is you need variation. In nature, that’s handled by genetic mutation. But in the case of social media, variation is generated by just every user trying different things. The second thing you need is selection. So in nature, that’s sort of natural selection, survival of the fittest, and in western social media, it’s generated by the algorithm. And the algorithm is taking inputs on engagement and then choosing what gets distribution and what gets no distribution. The final thing you need is amplification. So in nature, amplification is handled by passing on your genes to your offspring. But in social media, it’s governed again by the algorithm. The algorithm can choose to show whatever you make to a whole lot of people, not just people who follow you.

What was interesting to me about TikTok when it came along was that TikTok was the first social network where you could have zero followers and get uncapped distribution. If you created an account, had no followers, made an amazing TikTok, hundreds of millions of people could be shown that thing tomorrow. Whereas with western social media, yes you could go viral with a tweet or an Instagram post, but if you didn’t have any followers, it was pretty hard to. There was effectively a cap on your distribution. And so what we have with TikTok coming along and sort of really forcing western social media companies to reevaluate whether they were even built correctly and what was the job that they were set out to do.

I wrote back in 2015 that Facebook had a real problem. I went back to that Paul Krugman quote where he said that the Internet wasn’t going to be interesting, because people don’t have very much to say. And I’m like, he was actually right. The vast majority of people aren’t very interesting. And my thing on Facebook was, look, Facebook needs to lean into being a portal for entertainment, because you’re going to be capped on interesting content by your network.

Now what I got wrong is I assumed that the interesting stuff would come from professional media organizations. What TikTok realized is no, UGC is where that’s going to come from, we’re just going to harvest it from across the network and put it in there. And I love your framing of this evolutionary aspect, because look, if the algorithmic is so all powerful, why are we even limiting ourselves? We should lean all the way into it. Network shouldn’t matter at all.

EW: This is the challenge, because I think a lot of people did come to Facebook and other services for connecting with friends and close people, and I still use the Internet for that quite a bit.

People don’t churn out of Facebook. They use it less. Snapchat I think is another great example where Instagram completely kneecapped Snapchat’s growth, but users the chat bit didn’t leave. Chat is incredibly sticky. You see this with messaging apps as well.

EW: So you could argue that the way to think about it is that let’s separate out deterministic and probabilistic distribution for communicating with our friends and family, messaging apps have maybe a better solution. The distribution is deterministic. If you need to text your wife that you’re running late and can she pick up the kids from school, you’re not going to go through the Facebook newsfeed and pray that she sees the story. You’re like, “I need this to go through.” And you could argue that Facebook with its portfolio should say, “Okay, Messenger and WhatsApp are going to handle the type of deterministic communication that is core to social and that probabilistic distribution and systems like TikTok, which are sort of closed loop evolutionary systems to try to force the fittest entertainment to the top are the right way to treat that type of content”, which is just like, “Hey, I just want to find something really entertaining that fits my interests.”

And I think what’s an interesting challenge for Instagram is that they are sort of fighting those two jobs, and they’re trying to determine what it is. And they’re always trying to have it a little bit in both ways. It’s hard, like you said, nowadays with my friends, it’s mostly, we just do Stories because we know that at least it’ll be at the top of the Instagram app and you can click through the bubbles, and we no longer really post stories in the feed because it’s a very low chance that anyone’s going to see it when you’re competing with posts from Kim Kardashian and whatever other famous influencers. I think that’s maybe one way within Facebook to resolve this inherent tension.

Well, here’s the question. I wrote a piece a few years ago, Social Networking 2.0, that basically argued that we’re at the end of the road as far as public social networks go, where socials are based on the context of people you know. Because for lots of reasons, Facebook and Twitter are transitioning to broadcast, that’s the real value of them. And anything personal you want…What was Google Social network called?

EW: Circles.

Circles.

EW: Google+.

Google+. Yeah. I’m going in circles. So you know what I was talking about right?

EW: Yeah, yeah. Right. Right.

Circles actually really was a fundamentally correct insight. We want different circles of communication. Within me, contains multitudes, within me, contains multiple social networks, all of which I want to have different expectations for, different experiences with. And the public social networks would become places where you would initiate or find content to put into your private social networks. And messaging is sort where that is. Again, that article suffered for not including TikTok to a sufficient degree, but TikTok feels like taking that concept to its logical extreme, which is the public stuff should have no network aspects attached to it at all. And then the network stuff should not have any algorithmic stuff attached to it. It’s the same thing we talked about with content: there’s a bifurcation between high end content and just filler. Similarly, there’s a bifurcation and social between broadcast, purely public and private. And Facebook’s stuck in the middle, just like Netflix is.

EW: Yeah, exactly. There’s another way in which TikTok takes things to the logical and natural extreme when it comes to entertainment, which is that even if you do have a gazillion followers on TikTok, you are only as good as your next TikTok. So you’re like Charli D’Amelio, you might have a hundred million followers on TikTok. If you make a boring TikTok, TikTok is going to bury that thing. They are not going to show it to anybody.

It’s remorseless.

EW: That’s the way it has to be. Because in evolution, if you are choosing for the fittest entertainment and you’re fighting all these competitors in the entertainment space, you have no choice but to do that. Like you said, as everybody who’s ever had to screen submissions to a film festival or an essay contest or anything, most things are not that great. You need kind of a filter, a sort of weeding out process to prevent people from being inundated with high noise, like lousy content because that drives people away. That’s naturally actually what Facebook and all these companies have been trying to do with algorithms in the first place. It’s just that ultimately, I think there was a mismatch between that architecture of the feed and the nature of interaction that was happening in the early days of social.

There was a path dependent sort of thing, where you just end up on the wrong place. They were artificially gating themselves by prioritizing content in your network.

EW: And I think it took TikTok coming along — the reason why I wrote all these pieces about TikTok is because a lot of people wrote me and they’re like, I don’t watch TikTok. I don’t know why you’re writing about TikTok. And I said, “Look, even if you don’t watch TikTok and you don’t find it entertaining, totally fine. What’s interesting is structurally the challenge that they presented to western social media.” They were probably the first interesting thing to come along product-wise in social for a while. And so I was very interested in just breaking it down to see how it worked, because it really pointed out some of the mistakes that western social media companies had made.

Now they’re starting to incorporate some of that. And the question is, will Facebook and Instagram and these other companies sort of take the right lessons from those challenges, and will they be able to rationalize those things? And going back to our conversation on streaming media, it may be that also that advertising being the main source of monetization is a good model for monetizing UGC entertainment-based content, but maybe not optimal for the deterministic types of social communication that we want. And one of the ways I’ve been frustrated with western social is that I think I wish it was more useful as social infrastructure, and I think that it would be if advertising wasn’t the dominant business model there.

Twitter

Well, it’s interesting to tie this into the Twitter saga. I think one of the interesting things about Twitter is that it is fundamentally text based. And I think that is a very sharp differentiator in both directions. I think that’s why Twitter’s growth was fundamentally capped, because most people prefer visual. But for the people who prefer text, Twitter is irreplaceable, number one. And number two, no one is going to build an alternative to Twitter because it’s a bad business because it’s capped. And so one of my takes, I wrote this when Jack Dorsey stepped down, I’m like, look, Twitter, I think could be the one and only social network that actually is subscription-driven because it’s so unique. Its value is so high to the people that want it.

Moreover, its growth prospects are so fundamentally limited by that same differentiation, such that I think advertising’s never going to work on Twitter. It’s never going to really work at the scale that’s necessary. And it’s interesting because obviously that debate’s sort of playing out big picture right now. I mean, I don’t know. It’s very contrarian. I had so many people that replied to me after that article saying, “This is the stupidest thing ever. You can never have a social network that’s not free.” And I just feel like Twitter is actually so highly differentiated that it’s the one that possibly can. I mean, what do you think?

EW: I would agree that I think Twitter’s one of those examples of a large niche business that the internet enables…

Yes. That’s a great way to put it. It’s a niche. That’s a great way to put it.

EW: Exactly right. Yeah. Invisible Asymptotes was an essay I wrote a long time ago, and I wrote that Twitter’s invisible asymptote really was that it was this short text-based network and most people actually just don’t care for that. That’s fundamentally a limiter. I tried to get people on Twitter and they’re like, “What is this? This is terrible. I would never use this thing.” And I was like, “Yeah, I get it. I get why it’s not for you and why I open it every day.”

Multiple times per day.

EW: Yeah. That’s always been its challenge, is that at scale it just might not be a great business. And look, not all businesses are meant to be great businesses. You can build a great product and it can be a terrible business and that may ultimately be what Elon has bought. And it’s funny because a lot of people are like, “He’s going to bring back free speech and all of that.” But also that might be a decision that also makes it a worse place for a lot of people. So there are these challenges where the principles that he wants to bring to Twitter may just make it even worse as a business in the long run. It remains to be seen. But I do agree with you that I think it is capped as a business and I don’t know that there is an easy way.

It’s definitely capped for sure. The question is that cap, is that the flip side of the coin of it being so indispensable to its users? Because that’s the whole thing, right? If you’re a niche product, you have to maximize your revenue by increasing your average revenue per user. If you’re a wide scale product, you want to increase — you said this at the very beginning, right? There are two ways of profitability and it’s just sort of been accepted as a rule that every social network, because the network effect is so important and the value increases with the number of users, they have to be free. There is no way around this.

I just wonder if Twitter is the exception that proves the rule. It’s so unique. I would argue that the people who are attached to Twitter that open it multiple times per day are more attached to that service than any users of any other social network are attached to that service. If that’s the case, monetizing them by showing them promoted tweets sounds like a terrible strategy because you’re not harvesting nearly enough of the value you’re providing.

EW: Right. The open question to me about Twitter is, let’s say you charge a subscription fee for some set of users and you bring back a lot of accounts that had gotten shut out. Is there an equilibrium that you can attain where it’s still good enough for enough people to be viable? Because it is a kind of delicate balance. What I’ve never been able to figure out is, I have this two-by-two that I use to analyze social. The X-axis goes from people you know to people you don’t know on the left. And then the Y-axis goes from convergent worldviews to divergent worldviews at the bottom. I’d argue that western social media, the biggest thing it did was increase the surface area of the bottom left quadrant, which is interactions between people who disagree with each other and don’t know each other.

Yep. Which means people actually get the filter bubble totally wrong. You actually see way more opinions that you disagree with today than you ever did previously. And that makes people really mad.

EW: Yeah. So the upper right quadrant, it’s easy to build a business, people you know, you share interests. The upper left quadrant, people you don’t know, but you share interests. It’s like Reddit and things like that, still works. A lot of Twitter’s like that. The bottom right quadrant, I call the racist uncle quadrant. It’s people you know but you see them at Thanksgiving once a year and they have some racist views, but it’s like, all right, once a year, they’re family, so you put up with it. And the bottom left quadrant is interesting because I would argue Twitter has probably more of that than almost any other social network.

Part of it is that Twitter chose at a number of points to just amplify distribution randomly. So when I look through my Twitter feed today, I get a lot of tweets from people I don’t follow. A person I follow liked this thing, or they just decided they were going to show it to me because they thought I was interested.

Twitter, by following the example of other social networks, has made mistake after mistake. And I think to your point, you talked about the probabilistic versus deterministic, Twitter is at its highest value when it’s extremely deterministic and you follow the right people. Right now, getting someone to follow the right people is really, really difficult. And I think you wrote this in one of your articles, the problem is conflating the person with the interest, where people follow because they’re interested in something, not because they want to follow a person. And so they’re like, Twitter fell in that division. But yeah, I mean so many things that work for other social networks just don’t make sense for what Twitter is.

EW: Yeah, right. My sense in just toggling back and forth between the, I forgot what the two different feed sorts are for the Twitter algorithm, but as I toggle back and forth…

Yes, it’s like home versus latest or something.

EW: Yeah.

It’s basically deterministic versus probabilistic.

EW: Right. I think that they have tried with their probabilistic algorithm to lean into a little bit of what TikTok does, and I just don’t think they have enough good signal to do it well. But the question to me is just, let’s say you increase the surface area, the bottom left a lot, and Twitter does that, especially for politics, is there some selection bias such that the people who are left are the ones who just secretly enjoy that kind of conflict?

Of course that’s the case.

EW: And that drama.

Yes, without question.

EW: And if that’s the case, how many are left if you go really deep on that? Because one of the things that I was just reading all these evolutionary theory books, and the one thing about evolution is you get what you select for. That’s one of the most powerful things about evolution is that if you select for something, you get it in the long run. It doesn’t even take many generations for that to become dominant. I think Twitter’s at a point where their algorithm and their users have been battling each other for so long that they’ve battled to kind of an equilibrium. It’s kind of how in gaming they have this theory of the meta, the dominant tactic. I think the meta of Twitter has settled into this place.

It’s the dunk. The meta is the dunk.

EW: Yeah. It’s just the pleasure of going and just seeing the people that have terrible takes that you disagree with and dunking on people and telling weird jokes that go viral, a lot of which are jokes about other people. Maybe that is kind of the local maximum that they won’t be able to get off of. And maybe that’s what they’re destined to be.

Which is, by the way, a terrible product for advertising.

EW: Yeah, exactly.

This was my whole point when I wrote this article. To the extent your social media is entertainment, it’s a lean-back experience. You’re just sort of scrolling through, you’re relaxing, you’re on a Saturday night, maybe I’ll watch Netflix, maybe I’ll scroll through TikTok for a while. That’s a phenomenal posture for advertising because you’re at ease, you’re not riled up, you’re sort of open to new experiences. That’s the whole reason you’re surfing TikTok or Instagram is because, hey, what’s going to show up? Maybe if something shows up is a product I never thought about. That’s why Facebook works so well from an advertising perspective.

Twitter’s the opposite. Twitter, you’re leaning in. Part of it is it’s text. You have to concentrate to read text. I know for sure, if I want to go viral, the more meta, the better. Small M meta. Where you make passing references to some bit of Internet lore that’s not explicitly referenced in the tweet. If you have to explain your tweet, it’s not going to go viral. You want it to be inscrutable to someone from the outside. That’s what people love. And then they’ll retweet it and it’ll go crazy. That is heavy mental engagement. You’re super focused and when you’re mentally engaged, ads don’t work on you because you’re just going to scroll right past it.

EW: Yeah. And it doesn’t help that Twitter’s ads haven’t also been that interesting or great in their targeting over the years. It’s a rough product from that perspective. I think you’re right. The point you made I think is really important, is that, ironically, the fact that it is a bad business is kind of its moat.

That’s right. So if you charge for it, people will, honestly, I think just charging for extra features is dumb. Elon has nothing to lose. He overpaid by probably three or four X. I say, charge every user to use Twitter. I think everyone’s going to complain and say there is no way I’ll ever pay. And a massive, massive part of that base is going to come crawling back. Yes there are the people that dip in and dip out, you’re going to lose them, but at this point, everyone’s heard of Twitter. It’s not like you have a trial issue. Everyone’s trialed Twitter and the vast majority of people have decided that it sucks and they don’t want to be there. But the people that are there, they can’t break away, and there is no alternative, and there will never be an alternative.

EW: Yeah. I think it’s an interesting idea. I would love to see him try that experiment.

Well the funny thing is if he tries and it fails, the company’s gone. Right? And honestly, it might be great for society.

EW: Yeah, it might be. I think it’s just, we may look back on it as one of these weird anomalies in history. Let’s say Twitter goes away. We’ll just go back and, what? What was the idea in building this particular configuration that was just suboptimal?

If Twitter goes away, there will never be another Twitter. I think that’s the case. To try to build it again, it would be madness. I think you said this before, part of the value people get on Twitter is the conflict. They want to be in the same place as their enemies. But the problem is that can only be an emergent ecosystem. You’re never going to build that from the ground up. That had to emerge on Twitter and no other company would allow that to emerge going forward because it’s terrible for business.

The Problem With Search

EW: Yeah, it is really funny, I think, in this first era of social, or maybe this is the second era of social, we built so many things sort of bottom left quadrant. You remember all the weird anonymous social networks that were just instantly toxic. You look back at it and what’s funny is some of them are trying to make a comeback. I’m like, did you not learn the lesson the first time around that there’s no business to be built in this space? One of my favorite quotes is from the sociobiologist E.O. Wilson. He wrote this a while ago. He said something like, the chief problem of modernity is that we have godlike technology, medieval institutions and paleolithic emotions. Something to that effect.

Phenomenal quote.

EW: And yeah, it’s a great quote. And I have a version of that quote, which is slightly adapted for the internet era. The latter two parts are still the same. We still have the medieval institutions and the paleolithic emotions. But the first part is that we now have godlike search and distribution. And that, I think, is just one of the things that we’re still grappling with. If you read McLuhan or Neil Postman, they talk about what a huge sea change it was in the world when we went from print to TV as the dominant medium. I think if they were alive today, they would be so fascinated to delve into what happened now that we’ve gone from print to TV and then to the internet where everybody has, potentially, a megaphone that can reach the world, and where you can find anything, any piece of information.

And because the internet has been driven by these algorithms that select for engagement, it’s not just that when you search for something, you can find it. It’s that if you search on any belief you have, you will find the most charismatic engaging argument for that belief because it’s been filtered through page rank and social media algorithms. It’s had to fight its way through a bunch of services that govern distribution based on engagement-based algorithms.

So what comes out of it is super potent. I see this with memes. You’ve been around long enough, but you probably remember, like me, that memes when we were younger were just, the same meme would circulate. The one I liked was the fake commencement speech by Kurt Vonnegut. Someone would forward that to me via email once a year. And I was like, “Oh my God, we’re still on this meme.” It’s going around very slowly through AOL email addresses. Today, I get new memes sent to me every few hours and they’re so potent.

The selection pressure on them is so insane.

EW: Yeah.

Well I think there’s this bit because I talked a bit about before about Twitter being written culture and written culture was the dominant form of culture in the West for the last 500 years, and it transformed the world. I don’t know if you’ve read the book Weird, but this idea of the printing press, the downstream effects were just so massive and it fundamentally changed the way people thought and individualism and all these sorts of things. But the natural state of humanity is oral culture. It’s not being precise with your facts. It’s sort of passing down stories, passing down traditions and TV started it, but UGC, user-generated content, is all about the oral tradition. It’s the reemergence and that’s the natural state of humanity.

However, It’s a toxic mix with search because oral culture, it worked because once you said a word, it was gone forever. It wasn’t written down. Writing down the words changes them. This was Snapchat’s fundamental insight — the bit about stuff disappearing, that was the natural state for internet-based communication. We had this fundamental problem where computers, by default, save everything, but the zero marginal cost of distributing communications means we say stuff that ought never be saved, and I think a lot of the problems in society are driven by that mismatch.

EW: Yeah. The thing I always take from the E.O. Wilson quote is not just the three things, the godlike technology, the medieval institutions and the paleolithic emotions, but I think what he’s really saying there is that the cycle time of change is different for those three things. So for technology, the cycle time of change is very fast. For institutions a little bit less so. They can be pretty slow to change. But for people, people change the slowest of all. You can read a Shakespeare play and still be like, “Oh yeah, I recognize why a fellow is so jealous.”

It’s still very insightful about the human condition.

EW: You understand all these things. And what’s crazy is that we paired infinite memory and godlike distribution with people who are still the same crazy people that we always were. We saw with all the people getting canceled in that first era, it was just the mismatch. We just weren’t ready for that type of persistence, that type of distribution of everything we say. It’s really changed the epistemic environment of the world. Because to me, the reason I’m very concerned about social media is social media today is the way that we as a society talk to each other and see each other. That is the dominant medium now. And it shapes how we feel about our fellow citizen, about the world, about different cultures.

I don’t think anyone who works in those businesses set out to tinker with the way the world works necessarily in such a precise way, but the fact is that they really did. That’s the problem with the E.O. Wilson quote. He says the chief problem of a modernity, and it still holds today, is that technology moves faster than the people do and that the institutions do. That has a runaway effect that I just think we’re still grappling with. My hope is that now that we’ve had some time to look back on social media and everything and to see its effects, that we can build more smart the second time around.

The Group Chat Redemption

I think there is a human adaptation happening. To me, I actually feel I personally am kind of in a bit of a nirvana state as far as social media these days. And the answer to that is WhatsApp group chats with disappearing messages. So you get the bit where you get to hangout with people, very high trust environments, you can say what you want, you can do whatever. I’ve always been very online because of being in Taiwan. Twitter started out as the water cooler and everyone could talk. However, anyone who is trying to have personal conversations on Twitter these days is a madman. I stopped posting. It’s probably bad for my business, but I stopped tweeting actively from my account years ago, and I’m in such a better place and so much happier. But you also want that outlet.

The reason why people share stuff is because they want to share, that’s the human condition. This is where group chats are amazing. And the other thing is, it’s the Circles idea. You have different group chats for different topics where you complain about X, Y, Z. You can talk about sports, you can do all the different things. And honestly, it’s a real contributor to my personal happiness. Now, I have a very strong group of in-person friends that I think is a very important aspect as well. But it’s almost like what’s missing, and I think this is maybe one thing that, and I think Facebook has woken up to this and the value of this. They talk about one of the advantages of Reels is it’s super trivial to share it into your private messaging groups where more social networking is happening.

Maybe that’s the answer. It’s the answer for me personally, where you can harvest the zero distribution aspects of the internet and get that oral culture aspect that humans crave, but do it in a way that’s “safe” where it’s only the people you trust and the stuff disappears. The disappearing parts are really important.

EW: I’ve done the same myself, especially in the pandemic. I tweet less than I did in the past by a huge margin. Most of my interaction now is through sort of private chat groups. I think it’s just a better match, like we said before, of deterministic and probabilistic distribution. Use the right tool for the job. I also agree with you that one of the striking things when you look at human history and human communication is that written communication is really a minority of human history and oral, and in some cases visual communication, were more dominant for more of human history. And so you could argue that Twitter is even anomalous from that perspective as a social network.

Absolutely. I think there’s nothing like Twitter, there never will be again. It had to emerge at a particular moment in time where, just from a technical perspective, text was easier to distribute and have individualized feeds at scale. Facebook was text at the beginning too, obviously it exploded with photos, but there was a technical deterministic aspect to why Twitter could come up when it could and no one would ever invent it ever again.

EW: Yeah. Also the fact that text and alphanumeric data is easier to scan and search and there were a bunch of advantages that the Internet took advantage of. But the moment we could we shifted. So I remember when I first used the internet and I was using news groups and email, they invented the little emoticons because you quickly found that text doesn’t communicate tone of voice that well, and there’d be all these misunderstandings, so you’d put the little equal sign parentheses to do the smiley face. Then we got emoji. And of course we started using those like crazy because they brought in just higher emotional valence to our communications. And then we had the animated gifs. And then we had short video. The moment you give anybody those tools, they always opt in to start using them more and more because, tonally, they convey a lot more than text does. Text has always been kind of a narrow-band communications medium except to the very gifted. And so it will be interesting to see what happens with Twitter.

It’s going to be interesting. Eugene, this is by far, I think the longest interview I’ve ever done, but I didn’t want it to end, particularly once we got going on the social stuff. I think the takeaway is number one, this should have been two episodes, one on streaming and one on social. So number two, you’re going to have to come back more quickly in the future.

EW: I’d love to do so. Thanks.

---

This Daily Update Interview is also available as a podcast. To receive it in your podcast player, visit Stratechery.

The Daily Update is intended for a single recipient, but occasional forwarding is totally fine! If you would like to order multiple subscriptions for your team with a group discount (minimum 5), please contact me directly.

Thanks for being a supporter, and have a great day!