The Elusive Interoperable Internet: Why Twitter Killed its API
Imagine what we could have built together...
Of Birds, Crocodiles and Frogs
On January 13, 2022, a handful of popular third-party Twitter apps silently stopped working.
TweetBot and Twitterific might be little known to your typical Twitter user. But for many of the media-types that dominate Twitter, they were essential tools. And the developers of those apps had no idea what was happening. Some assumed it must be a temporary outage.
When word came, the news was not good. Apps that replicated any part of the Twitter experience would no longer be tolerated.
In a way, the move makes a lot of sense. Twitter is an advertising business. When eyeballs are not on Twitter apps, Twitter loses money. But the move reads as ungrateful for a company that, in many ways, has its third-party developers to thank for its growth.
Twitter’s first mobile apps came from third-party developers. The Twitter bird came from Twitterific, as did the word “tweet” and the invention of replies/comments. Twitter’s cultural impact and its technical development relied heavily on the work of independent developers.
But those developers forgot the lesson of The Crocodile and the Frog.
“There once was a frog who longed to cross a fast-flowing river filled with snakes. He hopped along the bank, searching for a fallen log or narrow point, but found nothing. After many days, a crocodile appeared. He listened sympathetically to the frog's story, then offered him a ride.
"You'd eat me, or leave me in the water where the snakes would eat me," the frog said. The crocodile insisted his motives were benign. "If you're worried, ride on my tail."
That sounded reasonable, so the frog hopped on the crocodile's tail. But it soon became apparent that the crocodile needed its tail as a rudder. "Hop onto my back so I can steer properly," the crocodile said. The frog moved up and they made good progress. "Why don't you hop on my head?" the crocodile asked. "The view is better."
This made the frog nervous. The crocodile dismissed his fears. "Surely you don't think I'd eat you when we're halfway across." So the frog climbed to the top of the crocodile's head.
Then the crocodile complained that something was tickling his nose. "Would you check?" he asked. "I'm afraid I'll sneeze and send you flying,"
The frog hopped onto the crocodile's snout and, in one gulp, he was gone.”
Twitter’s decision puts an exclamation point on a truth little acknowledged, but quietly understood by startups – that the giants of Silicon Valley may offer you passage across the river, but they also won’t hesitate to gulp you down if it serves them.
Thanks for reading Charterless! Subscribe for free to receive new posts and support my work.
Web2: Create, Exchange, and Extract
Web 2.0 wasn’t supposed to be this way.
Tim O’Reilly articulated the promise of the new era when he explained that while Web1 was built on delivering services to millions of users, Web2 was about leveraging the web’s capacity for collective intelligence.
Google succeeded because it leveraged links created by users for its PageRank. eBay aggregated the listings of thousands of sellers and their rankings/reviews. Social Media was starting to do the same as we shared photos, blog posts, and gradually everything else online. The secret to the new internet was collaborating with your users and other developers to build an ecosystem.
The playbook that emerged - for Facebook, for Google, for YouTube, for Amazon’s Web Services and yes, for Twitter - was straightforward, if hard to execute.
It worked something like this:
The new company built a new tool with some kind of novel functionality. It used that novelty to attract users’ attention and data. The company then turned around and licensed access to that attention or data via an API. This enabled other developers or content creators to generate novelty on the hosting platform, which would in turn, generate more attention and data while growing the new feature.
New startups offered novelty to established businesses in exchange for that business’s attention and data-stores. These virtuous networks of exchange powered the exponential growth of many new products – like Zynga and Instagram (which grew on the back of Facebook) and, more recently Substack and Clubhouse (which grew on the back of Twitter).
But, there was a time-bomb lurking just out of sight that threatened to end this infinite game of trade. That would be monetization.
Growth was positive sum – many platforms could succeed together. But value extraction was zero sum. If I give attention to your product, I can’t turn it into profits myself.
Suddenly, the assets collected by the major platforms – attention, data and functionality — were not goods-for-exchange. They were leverage to extract rents. These assets could no longer be given away for free or next-to-nothing in a virtuous exchange.
Attention had to be traded for ad dollars. Data and functionality had to be traded for developers’ dollars. Facebook shut down Open Graph. Twitter restricted access to its firehose API. And Amazon did the same with its Marketplace APIs.
The builders and the VCs of the era paid attention. When it became clear that any major platform could and would shut down a business that depended on their services, it became impossible to invest in the developers building on these platforms.
The Web2 model is a tragedy of game theory. Because a firm needs to eventually generate monopoly profits, they have to eventually restrict access to the very same open-assets that helped them build their empire. And once firms demonstrate that this is the inevitable path, no one will bother to build on newer platforms. The result is a cycle of under-investment, under-innovation and stagnation. Rather than a composable web, we end up with a stagnant one.
This dynamic even threatens AI startups, the buzziest space of our present moment. OpenAI has out-innovated large companies to debut some of the most impressive tech we’ve seen in a decade. But businesses that want to build on this platform are not venture investable. After all, if OpenAI owns the platform, and they can raise API prices or restrict access at any time, how will a new business ever be able to extract sufficient profits to justify venture returns?
In contrast, the open protocols continually mint new Unicorns. Humble email continues to offer a substrate for startups like Superhuman, MailChimp and Substack some fifty years after the original development of electronic mail.
The Third Way
If Web3 works, and that is a big if, it will be because its approach to building platform technology offers a way out of this negative Nash equilibrium. Web3 provides a way for developers of new technologies to credibly promise not to change the rules of the game when they need to make money later.
This is, of course, what confuses so many commentators about Web3. It is not a technical solution to a technical problem. It is a technical solution to the social problem of credible commitments.
Twitter’s API can do anything that a blockchain can do and it can do it faster. What that API can’t do is make credible commitments not to revoke access or to raise prices.
Smart contracts are APIs. But they are immutable once deployed to a blockchain. This means that an API that works a certain way, at a certain price point will have to work that way forever.
This allows developers to make open-ended commitments to the people who build on their technology. And that means that businesses built on these platforms are long-term viable for investment. It opens the door for applications that compose these platforms together, not because of novel technology, but because of novel mechanisms for trust.
This is, of course, just a lot of buzzwords. So let’s get concrete about the types of businesses that could be built on Web3 platforms.
Clients/Aggregators A reliable Twitter API with blockchain encoded access-points would allow the development of better Twitter clients – much like it did with Twitter’s first mobile app. They could also breakdown the barriers between different services altogether. Today, apps are hacking ways to interoperate with existing networks – like Texts.com, but these depend on somewhat illicit backdoors. The dream of a singular inbox that can solve our app fatigue has been intractable in the dynamic world of Web2. But Web3 protocols and trustable smart contracts might finally make it come true.
Composable “Custom Games” In 2003, Blizzard released the third title in its popular Warcraft series: Warcraft 3. The game was a smash hit, becoming the fastest selling video game of all time. Warcraft 3 was a simple real-time strategy game – build a city, build an army, go to war. But what made Warcraft 3 iconic was its Custom Games feature. It enabled hobbyists to easily design custom maps, set new rules and otherwise build totally new games – some of which created entirely new gaming genres: Tower Defense and DotA.
But Blizzard also forbade developers from selling or profiting off of these custom games. So rather than spawning a sustainable ecosystem of new games-within-the-game, companies copied these game concepts wholesale and launched them as independent titles. It was a loss for players (who invested time and energy into the early games on Warcraft) and for the developers who designed them.
But it’s not all great news.
Web3 solves one of the Web2 platform problems with credible commitments. But it does so in a way that limits the rents that entrepreneurs can extract. And while that might be good for users, it’s not great for speculators or would-be Mark Zuckerbergs.
There lies the paradox at the heart of Web3: millions of speculators are betting on ecosystems that can only realize their paper valuations if they, too, can extract monopoly profits. But the Utopian promise of Web3 is predicated on building in a way that will prevent that extraction.
Web2 turned away from its idealistic promises when it was time to make money. Web3 has slammed the door on this Web2 escape path. But it doesn’t have a viable alternative for empire builders.
One path, of course, is to own the substrate on which the new technologies are used – that’s why there are so many blockchains like Ethereum, Solana, Polkadot, Aptos, Sui, etc. competing for shares of the pie.
But for app developers, the path is more circumscribed. For all the money made on Web3 speculation, it remains a mostly unsolved problem. There are a few concepts gaining currency with builders:
The Hyperstructure Model - Popularized by Jacob Horne of Zora, “Hyperstructures” are a model of value extraction wherein ownership is valuable because a community could activate a fee-switch. The desire of other users and stakeholders to stop them from doing so gives share ownership value.
The Support/Convenience Model. Many open source software products profit by offering a version of their platform with tech support for enterprise customers or hosting or a better UX. For example, Word Press is a free and open platform for building websites. WordPress.com, however, offers hosting, and a friendlier UI for fees.
Of course, Web3 could also follow an altogether different path than Web2. Maybe Web3 businesses will be uninvestable precisely because they have succeeded at preventing rent extraction. If that’s the case, and Web3 succeeds anyway, it will be all the more remarkable for creating an internet where independent developers, rather than VCs and unicorns dominate.
After all, the only trustable crocodile is one whose mouth has been bound shut.
Thanks for reading Charterless! Subscribe for free to receive new posts and support my work.