Our Screenshot Q&A with Eugene Wei on the current state of social media businesses. Eugene published an excellent analysis of Twitter's issues recently, titled How to Blow Up a Timeline. That inspired us to quiz him about the nature of social networks, what Twitter should have become, and how status translates between platforms. You can find Eugene on Twitter (for now) @eugenewei.

1. When you were at Amazon, why did Jeff Bezos make your team read Creation: Life and How to Make It by Steve Grand? Do you think it has relevance to Social Media?

He felt that the same lesson at the heart of creating synthetic life (which is the topic of the book) and the emergent complexity of nature should guide our Amazon Web Services design: complex things come from the most elegant primitives.

That is, elegantly designed primitives maximize combinatorial options for builders. I think he felt that was a substrate independent law. He wanted to make sure we didn’t over design any single web service in anticipation of what developers might want to use them for. Rather, design simply and let the developers figure out how to combine them. Think of Legos. Jeff wanted us to build AWS like the basic pieces, not like the custom pieces they sometimes offer for various brand collaboration sets.

Social media is similar in that we’ve started to see certain primitives repeated across many different services. A profile page. A post. The like button. The feed. Some algorithm. Blocking. Friending or following. An inbox. And so on. It’s hard to imagine a new social service that doesn’t incorporate at least a few of these primitives. It’s no longer sufficient to just reuse these, though. It’s about the specific way they’re mixed and matched and what emergent communities you can foster with them.

Just as more complex life forms emerged from simple primitives in Earth’s history, we see certain social services make phase shifts in complexity if they’re lucky to gather enough users and help them relate in specific ways.

2. Why do you say people underestimate the market part of product-market fit? Can you give an example?

Product-market fit works in both directions. The perspective of the builder is always toward what they can do to change their product, what features and functionality to add. But with a long enough lens on successful and failed technology launches, you see that things like market context really matters. If it didn’t, we wouldn’t see as many ideas that once failed become a huge success later on.

Are enabling technologies in place? Are cultural conditions conducive to spreading this idea? We tend to overstate how many startups create new markets rather than take share within an existing market.

I wrote a piece called Invisible Asymptotes whose main thesis is really that the market usually determines how large something can become. Those limits are often invisible to us because we don’t realize what market we’re playing in.

In the case of Twitter, they may have invented a new market in micro-blogging, but they also have run into the market reality that most people just don’t enjoy it, or at least they don’t enjoy Twitter’s version of it.

3. If you could build a new “golden-age” Twitter, what would it look like?

I’ve always felt Twitter stopped at one level of abstraction too low for mass appeal. From the onboarding to the syntax to the chaos of the timeline to the somewhat messy presentation of replies. Even trying to find communities of interest, assuming you figure out how to follow the right people, is too hard in my opinion.

I’d also be interested in shifting Twitter from what feels like a single global maximum—I think it’s at like 350M MAU’s—to a series of local maximums, each a community of interest, that sum up to a larger business, maybe 1B+ in MAU’s. You could lose some of that cozy feel of one shared timeline where everyone is screaming at each other, but I think that business will always hit a hard ceiling just because of the ambient disagreeability of it all.

Then on the private messaging side, I’d invest in really building out DM’s. That’s the private messaging framework. At some point it seemed as if they deprioritized DM’s, I’m not sure why.

Don’t miss future Screenshot Q&As. Sign up now.

4. If you chose the “right” monetization model for Twitter - what would that be?

As a pseudonymous network, Twitter won’t ever have the same personal information that Facebook or Instagram have. But Twitter is a pseudo-interest graph, and so ads targeted to various communities of interest have always seemed like the most appropriate path. That assumes Twitter can grow each of those communities, but that’s probably something they have to solve anyway.

I’m not against supplementing that with one or more subscription options, but the features of Twitter Verified obviously haven’t resonated with much of the user base, and it may actually be driving some users away. The only social services at scale that have relied on direct payments in the West seem to be games, and Twitter, while game-like in its nature, isn’t regarded as such when it comes to whether users would pay for it.

5. What is the right lesson to take from Twitter’s rise and fall?

Scaled social media services are complex adaptive systems. These are notoriously difficult to understand and manage. The algorithm in a social media service is like natural selection in evolutionary systems. You have to be very careful tinkering with it for a mature community.

6. What is legibility? How is it related to Social Media?

I first encountered it in James Scott’s Seeing Like a State. Faced with the challenge of governing a large mass of land and people with complex local arrangements, the state imposes standards that allow it to more easily govern top-down. Things like standardized language, population registers, standard weights and measures, more geometrically rigid urban design, these were all ways to simplify reality so the state would have a more synoptic view.

Just as the state needed legibility to impose taxation, social media needs some amount of legibility to produce useful content to its users and then to serve up and measure ad impressions in order to make money to survive as an ongoing concern.

Scott documents repeated examples of when imposed legibility overly simplifies reality. Social media is the same to me. The way it pushes us into very specific interactions is structurally coercive, and I think it’s been damaging to society.

7. As social media matures, how do you evaluate “status currency” across platforms? Are views on Instagram Reels, YouTube, TikTok all equivalent value?

Status is always relative, so there isn’t a fixed way to measure it. It’s only when we can translate status into something discrete like currency that we can have a conversation about how much a view is worth on, say, YouTube versus TikTok. Creators who are multi-platform are aware of that difference in ROI.

There are some structural differences. For example, since TikTok leans heavily on an algorithm to determine views for a video, your absolute follower count is less valuable than it might be on another platform where the follow graph is a stronger input to your success. On TikTok you’re only as good as your next video whereas on another platform your aggregate follower count might serve as a form of accumulated advantage.