Good morning,

On Monday I wrote a brief takedown of an article that Apple PR had been passing around on background (not to me, for the record) questioning the impact of ATT; Eric Seufert wrote a much more comprehensive response on Twitter that really digs into some of the nitty gritty.

I’ll be talking more with Seufert about the current state of digital advertising tomorrow.

On to the update:

Three Trends Follow-Up

Yesterday’s Article about Instagram, TikTok, and the Three Trends was, in case you are curious, the one I was working on last Monday when I took a personal day; I was hung up a bit on the fact that the themes were very similar to the ones I explored in DALL-E, the Metaverse, and Zero Marginal Content, but I think the big difference is that while that Article focused on costs, this Article was really about user behavior.

To that end, I think this bit from [yesterday’s article] is worth exploring a bit more:

Lessin’s five steps:

1. The Pre-Internet ‘People Magazine’ Era
2. Content from ‘your friends’ kills People Magazine
3. Kardashians/Professional ‘friends’ kill real friends
4. Algorithmic everyone kills Kardashians
5. Next is pure-AI content which beats ‘algorithmic everyone’

This is a meta observation and, to make a cheap play on words, the first reason why it made sense for Facebook to change its name: Facebook the app is eternally stuck on Step 2 in terms of entertainment (the app has evolved to become much more of a utility, with a focus on groups, marketplace, etc.). It’s Instagram that is barreling forward.

In retrospect, and to CEO Mark Zuckerberg’s credit, Facebook saw this coming, and the evidence is the FTC consent decree the company agreed to back in 2011. From the complaint:

On approximately November 19, 2009, Facebook changed its privacy policy to designate certain user information as “publicly available” (“PAI”). On approximately December 8, 2009, Facebook began implementing the changes referenced in its new policy (“the December Privacy Changes”) to make public in new ways certain information that users previously had provided…

The Privacy Wizard did not disclose adequately that users no longer could restrict access to their newly-designated PAI via their Profile Privacy Settings, Friends’ App Settings, or Search Privacy Settings, or that their existing choices to restrict access to such information via these settings would be overridden. For example, the Wizard did not disclose that a user’s existing choice to share his or her Friend List with “Only Friends” would be overridden, and that this information would be made accessible to the public.

What was happening here is that Facebook was trying to evolve its network from Step 2 to Step 3: instead of Facebook content being constrained to your contacts, content would be available to everyone. I do think the FTC’s complaint was reasonable, to be clear; when I say that Facebook’s actions were to Zuckerberg’s credit, I mean that in a purely strategic sense: the company was well ahead of the trend in terms of how content was evolving, but in its rush to get there it changed the rules on users. Instagram never faced this constraint, because posts were default public from the beginning.

This also provides an interesting framework to examine how and why social networks have stalled out. Facebook, as I noted, got stuck on Step 2, and ultimately has evolved to be a utility. Twitter, meanwhile, has always been public, but, to take another of the three trends, remains text based. Yes, you can post photos and videos, but the Twitter value proposition is very much tied into the density of text.

Snap is an interesting case in this regard: the company’s core functionality is very much tied into Step 2, but the company has layered on Step 3 and is investing in Step 4. One of the major challenges facing Snap, though (and I’m stealing from tomorrow’s interview here), is that shifting to AI-driven recommendations is extremely computationally expensive, and Snap, famously, has outsourced all of its infrastructure. This has mostly worked out in terms of the core service; Snap said in their Investor Letter last quarter:

Adjusted gross margin was 61% in Q2, an increase of approximately 5 percentage points year-over-year. Infrastructure cost per DAU was $0.58 in Q2, down from $0.62 the prior year, and equal to the record low as a public company of $0.58 that we reported in the prior quarter. We believe the progress we have made in driving down our unit costs over time provides validation for our asset-light infrastructure model, which has, in turn, enabled us to hold CapEx investment at less than 2% of revenue over the trailing 12 months.

The problem for Snap is that to the extent that AI capabilities become critical not only for content but also advertising the greater the burden their reliance on cloud providers will become. To put it another way, while I am generally a huge advocate of using public clouds — the flexibility, scalability, and engineering focus it enables is almost always worth the higher cost — there may be an exception for companies that are extremely AI-dependent, and there is a strong argument that this is exactly where “social networks” (I put that in quotes because the entire point is that reality is moving beyond social networks) are headed.

The Question of “Cool”

Let me put a disclaimer up front: I am the last person in the world qualified to provide commentary on what is “cool” and what is not; I make my living writing business analysis and get excited about buying new equipment for my home office.

That noted, one of the implications of yesterday’s Article is that there may be a more nuanced view of the argument that social networking is ultimately dependent on the “cool” factor; John Gruber articulated the “cool” thesis last week on Daring Fireball:

To wit, that maybe Facebook’s problem isn’t really ATT, but rather that Facebook’s entire business, from when it first started selling ads through today, has been about offering the hot new thing to the ever-desirable “young demographic”. Let’s call them “20-somethings”, or, if you prefer, “18-to-34-year-olds”. That product, at one point, was Facebook itself — a.k.a. the blue app. If you’re old enough, you will remember that it was, originally, hard to get a Facebook account. It was like an exclusive club. You had to be a student at Harvard. Then you had to be a student at any Ivy League college. Then you just needed a .edu email address. Eventually, of course, you just had to be breathing. But make no mistake, the Facebook of Aaron Sorkin’s 2010 movie The Social Network was the hot new thing for the 20-something demographic.

For the last decade — the 2010s — that hot new thing for the 20-something demographic was Instagram, which Mark Zuckerberg had the remarkable prescience to acquire in 2012 — such early days for Instagram that the acquisition came just weeks after Instagram had launched its Android app.

Now, though, the hot new thing for the 20-something demographic is TikTok. Duct taping TikTok features onto Instagram can’t stem the tide of a generational sea-change like this. Facebook itself remains humongous, and so too will Instagram. They’re not going anywhere. But MTV is still around too. The top-rated cable TV networks in the U.S. are unchanged in decades: Fox News, MSNBC, and CNN.

My theory is that Facebook’s blue app and Instagram are evolving into something akin to those networks in terms of demographic appeal. Stable, huge, but not growing and not for the youth. I could be wrong. But I don’t think I am. Internet content networks grow until they stop growing, and then they lose their advertising demographic — and investor — appeal.

I’m not disputing that there is something to this thesis: the history of not just consumer tech but consumer products generally is that young people want their own thing, and that thing’s number one quality is that it is not the old thing. That almost certainly applies to Facebook’s various properties.

At the same time, I think there is a technological determinism argument that is under-appreciated here. Every social network — if you can even call TikTok a social network — is a product of the technological capabilities at the time, and there is certainly a case to be made that those capabilities are an essential factor in what breaks through.

Go back to MySpace: the idea of non-technical users being able to create their own pages wasn’t entirely unique — remember GeoCities? — but providing a means to link them together was. Facebook took the concept and added the News Feed (which was, interestingly enough, layered on top of a certain rigidity in terms of profile layout). Snap leveraged increased bandwidth and computer capacity to create a service that was more dynamic and video-centric. TikTok, meanwhile, is all about recommendations from across the network, which was Facebook’s blindspot.

This last one is particularly interesting in terms of this thesis: had Instagram created Reels before TikTok, would TikTok have ever achieved the dominant position it does today? To put it another way, it TikTok so popular because it is something new, or because Facebook/Instagram was stuck in a previous paradigm that was defined by what was computationally possible when they came along?

TikTok and the Sinicization of the Internet

Twitter user @rpc_tweets points out that my attributing Autoplay to YouTube ignores how Stories works:

To be perfectly honest, I totally forgot about this; as I noted in a reply I usually tap through Stories like it’s my job.

This, though, touches on the interaction model points I noted in yesterday’s Article. Part of the brilliance of the Stories format is the feeling that you need to catch up on everything before it disappears, out of loyalty to your social network if anything else; TikTok/Reels, on the other hand, feels like a concentrated dopamine provider, where every swipe provides the promise of just one more video that gives you the hit you’re looking for.

I have to be honest: the more I use Reels, the less interest I have in Stories; they seem so boring in comparison. At the same time, this emphasizes what a high-stakes gamble this all is for Instagram: there is a big difference between “I ought to” versus “I want to”; spending ever more time on content tailored to me is, if you step back and think about it, the exact sort of activity I ought to feel guilty about. Instagram has, thanks to its social network foundation, retained some sort of the old “connecting” DNA that Facebook used to talk a lot about, but the more it tries to ape TikTok the more it is nothing but pure entertainment that, if I’m honest with myself, I feel a bit guilty about indulging.

One more point: I’ve made the case that one of the reasons for TikTok’s breakthrough was the fact that the company started out in China, where centralized control of content was a feature, not a bug. This aligns with a point Michael Mignano made in the post I linked to yesterday:

But just as massively as social media platforms have grown and changed the way we all consume content, they have also wreaked havoc for platform companies, the internet, and more broadly, our world. Built-in distribution for content to social networks has meant that people can share and spread problematic content just as easily as they spread good-natured content. If a bad actor wants to share problematic content on social media, the content can spread fast because of the guaranteed distribution to the person’s network of friends…

The above problems with social media in turn generate massive costs for platforms, in the form of gigantic moderation teams made of tens of thousands of people, severe damage to platforms’ brands, and openings for competition to find more efficient means for distributing content. And no platform has been better at exploiting the weaknesses of social media than TikTok, the platform which popularized algorithmic content distribution and gave birth to what I call, recommendation media…

Since a platform is in control of what content gets served to who and when, there’s no expectation that a creator’s social network is guaranteed to see their content. Therefore, platforms can also choose what not to program, and there’s little creators can do or say to counteract this. Long gone are the days where a creator can complain about being de-platformed or shadow banned because their followers aren’t seeing their content; in recommendation media, the algorithm is understood to be the final decision maker about what gains traction and what doesn’t. This gives platforms far more leverage to hide unwanted content and therefore reduce the need for massive moderation teams. It’s not that these teams are no longer needed; they’re simply not needed to the same scale as in social media because distribution for certain types of content can be eliminated from a platform without changing the underlying structure of content distribution.

Mignano paints this as a positive, and it’s a reasonable point; at the same time, it is a striking example that the more that content moderation becomes the expectation, the more that the online environment becomes an inferior version of the Chinese Internet.

---

This Update will be available as a podcast later today. To receive it in your podcast player, visit Stratechery.

The Stratechery 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 subscriber, and have a great day!