This post has been updated as I’ve thought more about these distinctions.
My point in writing this post is both to note that there are internet-based models of organization and production other than the much-discussed “platforms,” and to distinguish a few different kinds of platforms. All of this is based just on my own reading and thinking. I’m sure others have better, more formal, and more considered definitions.
Different kinds of platforms
Participatory platforms: They offer users the ability to create something, to communicate, or similar — often in a fairly open-ended way. And they place few if any limits on those users’ activity or on who can join. Think Twitter or Tumblr. The scale and variety of users’ activity and the connections between users create something the platform owners could never have created on their own (for good or ill).
APIs: Application programming interfaces let other programmers communicate with your system, to access data or functions, or to do a variety of other things. They therefore let others build things using your product, data, or service. One version I’m counting within my loose version of APIs is operating systems which offer APIs so that people can build software “on top of” the OS. So think of Android and Windows as falling within this category, but also Facebook or Twitter at some points letting others build apps on top of their platforms or access pieces of their data. This is what people are getting at when they use the metaphor of a “platform” to generalize the idea of letting others “build on top” of whatever you offer.
Two-sided markets: These platforms match buyers and sellers. Think eBay, Uber, Airbnb.
Big companies like Google, Facebook, or Amazon fit multiple categories; they are platforms in multiple senses. Other platforms are platforms only in a single sense.
Sometimes, reading the business press, you’d think platforms were the only way the internet changed organization. But there are other models made possible by digital technology and the internet:
Peer production: Users come together to collaborate, typically towards a project that will be a commons — like Wikipedia or Linux. The collaboration is complex and may require elaborate rules and norms to govern interactions.
Crowdsourcing: Users offer distinct inputs — votes, predictions, code — that are collected by a central entity. Sometimes they’re aggregated (as in prediction tournaments) and sometimes the top entrant is rewarded (as in contests on Kaggle or TopCoder).
Data loops: A product gathers data which is used to improve the product, creating a positive feedback cycle. Netflix’s recommendation algorithm is one example, and here is a piece on this model.
Aggregators: They sit on top of a mountain of content and help their users find things. Participatory platforms often have to build aggregators — like the Facebook news feed — but aggregators needn’t be participatory platforms. Google is an aggregator, but unlike Facebook the content it navigates is not it’s own.
(I’ve only tried here to include models where the nature of organization and production is changed. Of course there are lots of ways to use the internet and digital technology to make existing production more efficient.)
The problem with thinking of platforms as the end-all be-all of the internet is — other than the confusion that comes from the multiple types — that it distracts from the real concept at the internet’s heart: networks. These are all networked models of organization and production, made possible by ubiquitous connections and the constant flow of information between them.