I was emailing with a friend recently, who asked:

"On the web, in order to build a platform you first need a hit app. Do you think this dynamic is different in blockchain?"

It's a great question, and one I have been thinking about a lot lately.  First, let's unpack the idea that the way to make a "platform" on the web is by starting with a hit app.  This has certainly been the case with Amazon, Google, Twitter, Facebook, Salesforce and others -- where a very successful application enabled either the consumer scale (FB and Twitter), enterprise scale (Salesforce) or infrastructure scale (Google and Amazon) for others to build on. Identity is a particularly interesting example.  To the extent that Google, Facebook and Twitter are the identity standards on the web today, those were obviously a second-order results from hit applications.   So you essentially have the infrastructure or platform layer "falling out" of, or layering-under, the application layer. But there are counter-examples, most of which are lower-level services and developer tools: Stripe, Twilio, MongoDB, etc.  These are platforms first, by their nature -- they exist only to have things built on top of them, and were able to achieve scale with that approach. And thinking back to the creation of the web itself, I think it's fair to say it was platform-first.  The Internet protocol stack (TCP/IP, HTTP, SMTP, etc) is horizontal infrastructure, designed in a layered model that anticipated continued development by others (though perhaps you could argue that email was the app that drove development of that platform). So, looking at the blockchain, what does it mean to build a platform?  How will the important infrastructure features (e.g., identity, reputation, personal data, payments, etc) come to be?  Will they be achieved platform-first, or will they be a second order result from something at the application layer? In other words, will platforms on the blockchain be built bottom-up (like the original web protocols), or top-down (like the commercial web)? In a lot of ways, the blockchain is like the original web: protocol-based and open source.  Implemented by a group of collaborating peers.  Dueling protocol design, white papers, and RFCs.  Layered. In other ways, the blockchain is like the commercial web: hyper growth fueled by powerful economic incentives.  Economies of scale and network effects. I don't have a clear answer, and maybe it will be a combination of both. But it's really interesting to think about.