What’s an API good for, anyway?
Last week, Instagram unceremoniously locked down their API, ending the ability to build a third party client. This shouldn’t have come as surprise to anyone: the API has become a platform “Hail Mary” for free developer labor, becoming only less effective each time a platform has gobbled up its developer community’s innovations (hey, Twitter!)
Flickr founder Caterina Fake’s prediction in 2006 that APIs would become “BizDev 2.0”– allowing code to act as legal agreement, ended complicating custom deals, and enabling permission-less innovation– has failed to pass.
Looking back over the past decade, we’ve seen popular APIs wither away:
- LinkedIn locked down their APIs in 2015, because the applications on top became a threat to their business
- Netflix shut down their open APIs in 2014, with few interesting applications
- Twitter shut down many API users in 2011, when Twitter realized they wanted to capture that value themselves
The thinking for releasing a public API goes something like this:
“Why don’t we release an API? We can’t possibly imagine all the amazing things developers will create on top of our platform (for free!)”
But what happens next is one of two things:
- An innovative use case is created, which is eventually shut down by the platform because it (a) enables the bootstraping of a competitor or (b) is so valuable that platform wants to capture the value itself.
- Not much happens because either (a) the applications are boring because it’s hard to create something innovative without give-and-take between the API’s designer and its user. Or (b) developers know that interesting applications get shutdown/copied so few invest in the platform.
So when do public APIs really work?
When the API is the delivery mechanism for the business. For Stripe, Twilio and AWS, the API is their business. A “self-service” API defined not only the go-to-market strategy but the core product of each company. It isn’t possible to create a service on top of their APIs that harms the company except outright fraud.
These three services (Stripe, Twilio, AWS) are exceptions: most open APIs benefit neither platform nor developer. Even for the API companies, the API often ends up being long-tail marketing: it takes the business development team to unlock the big dollars.
Most of the time, for most businesses, business development is still bizdev 1.0 and developers relations are (depreciated).
Thanks to Jeff Weinstein, Justin Overdorff, Jess Peterson and others for their comments.
1) Google Maps is a more complicated example, although it fits alongside with Stripe, Twilio and AWS: their business is providing and selling mapping data. What makes it complicated it that they happen to have the dominant client too, and have given away their data for years in order to dissuade competitors. No longer, and we see the rise of platforms like Mapzen and Mapbox.
2) APIs are still valuable for lots of other reasons not mentioned here.