Jun 3, 2023·edited Jun 3, 2023Liked by Steven Sinofsky
Yours is a way more nuanced view, but FWIW I have always basically thought there are two types of valuable platforms – distribution platforms and infrastructure platforms. I actually think Ben Thompson's whole aggregator framework here is useful, as distribution platforms aggregate consumers and infrastructure platforms aggregate developers.
In a distribution platform, businesses fundamentally adopt the platform for distribution. You have built a product that has captured a valuable-enough audience that other businesses want to integrate with you to serve that audience. Examples:
- iOS. No one would care about the iOS platform if it had 100k users. I'm sure some people like Swift and Objective C, but the real reason you build iOS apps is because the iPhone has a billion+ users globally.
- Facebook (circa 2007-2012). Facebook was the largest, fastest-growing app in the world. We let people access that audience.
- Android, Windows, Xbox, Playstation ... most hardware products.
I had never thought of SAP but I think SAP and Salesforce are examples of this, too – they have enough customers out there that other businesses would rather just build on top of this and co-sell/distribution with them then start from scratch.
Infrastructure platforms in my mind are things that make a developers job and time-to-market substantially easier (e.g., AWS). I think the dividing line between product and platform here is hazier, but I think an infrastructure product starts to become a platform when a large part of the value of the platform comes from the fact that other developers use it and that connective tissue bleeds into all kinds of other systems.
The end user typically has no idea what infrastructure platform you're using and doesn't care (they don't know if you're using AWS vs. GCP or Stripe vs. Adyen and don't care).
Common mistakes I've seen:
1/ Trying to build a distribution platform from day one. Developers will only care about a distribution platform if you have distribution, which in almost all cases requires having a great product that lots of people are actively using (ideally 10s or 100s of millions). This is one of the reasons that not having an App Store was arguably a feature, not a bug, of iOS v1.
2/ Being confused about what kind of platform you are. Distribution platforms should aspire to good APIs and developer experience, but they should be clear-eyed about the fact that developers aren't picking them because of the quality of their API or technology but based on maximal relevant reach. A Twitter competitor isn't going to win because of a better API. MacOS wouldn't have beaten Windows in the late 90s with a 10x better API.
To your point about paradigm shifts, I think the fundamental reason they are vulnerable is that the most visible platforms are distribution platforms (they aggregate consumers) and paradigm shifts are almost ipso facto shifts in consumer behavior and aggregation. E.g., Linux and Windows are still exceptionally relevant as operating systems and the dominant environment in which software runs, but their importance as chokepoints for consumer demand are massively diminished.
Yours is a way more nuanced view, but FWIW I have always basically thought there are two types of valuable platforms – distribution platforms and infrastructure platforms. I actually think Ben Thompson's whole aggregator framework here is useful, as distribution platforms aggregate consumers and infrastructure platforms aggregate developers.
In a distribution platform, businesses fundamentally adopt the platform for distribution. You have built a product that has captured a valuable-enough audience that other businesses want to integrate with you to serve that audience. Examples:
- iOS. No one would care about the iOS platform if it had 100k users. I'm sure some people like Swift and Objective C, but the real reason you build iOS apps is because the iPhone has a billion+ users globally.
- Facebook (circa 2007-2012). Facebook was the largest, fastest-growing app in the world. We let people access that audience.
- Android, Windows, Xbox, Playstation ... most hardware products.
I had never thought of SAP but I think SAP and Salesforce are examples of this, too – they have enough customers out there that other businesses would rather just build on top of this and co-sell/distribution with them then start from scratch.
Infrastructure platforms in my mind are things that make a developers job and time-to-market substantially easier (e.g., AWS). I think the dividing line between product and platform here is hazier, but I think an infrastructure product starts to become a platform when a large part of the value of the platform comes from the fact that other developers use it and that connective tissue bleeds into all kinds of other systems.
The end user typically has no idea what infrastructure platform you're using and doesn't care (they don't know if you're using AWS vs. GCP or Stripe vs. Adyen and don't care).
Common mistakes I've seen:
1/ Trying to build a distribution platform from day one. Developers will only care about a distribution platform if you have distribution, which in almost all cases requires having a great product that lots of people are actively using (ideally 10s or 100s of millions). This is one of the reasons that not having an App Store was arguably a feature, not a bug, of iOS v1.
2/ Being confused about what kind of platform you are. Distribution platforms should aspire to good APIs and developer experience, but they should be clear-eyed about the fact that developers aren't picking them because of the quality of their API or technology but based on maximal relevant reach. A Twitter competitor isn't going to win because of a better API. MacOS wouldn't have beaten Windows in the late 90s with a 10x better API.
To your point about paradigm shifts, I think the fundamental reason they are vulnerable is that the most visible platforms are distribution platforms (they aggregate consumers) and paradigm shifts are almost ipso facto shifts in consumer behavior and aggregation. E.g., Linux and Windows are still exceptionally relevant as operating systems and the dominant environment in which software runs, but their importance as chokepoints for consumer demand are massively diminished.