Platform Moats

Your superior technology does not matter as much as you think.

Once upon a time I was invited to speak at a conference which was very different from the conferences I used to attend or even speak at. The strangest thing was that most of the people who organized and attended were twice my age and more. Luckily, like any good conference, the day before the event, there was some sort of social party where speakers got to meet each other as well as the organizers and the advisory board. This is how I met John Perry Barlow or Alan Kay, among others

“A meshwork of green ropes against a red background” by Clint Adair on Unsplash

When I described Superfeedr (which was young at the time!), I told that our API was “just” a Pub/Sub implementation over HTTP. I was really surprised by the reaction of these “old-timers”. It’s hard to remember their exact words but this was something along the lines of :

Wow! eventually everything is being implemented over HTTP, despite it being such a crappy protocol!

The next hour or so was spent telling me about some of history of the early days of the web. Apparently, at the time, a very large portion of the tech community (whatever that was!) was extremely dismissive of HTTP and the web: there was no way this would scale to even being considered a viable option for the information highways that people where starting to think about. It was a good thing dozens of application specific protocols existed, from FTP to SMTP through NTP: all of these were better suited than HTTP to get actual things done!


Being French, I grew up with some kind of weird fascination for the Minitel: this strange “computer” which my parents were using to look up other people’s phone numbers or to order clothes (3615 “La Redoute”). Recently my friend Tariq KRIM shared a tweet which shows a screenshot of a very official report ordered by France’s Prime Minister about the Internet (in ‘94’)

For those who do not speak French, this has gems like “Internet does not include any kind of security system” or “Message transmission is not guaranteed” or “there is not even a full directory of people or services”… etc, ending with “The worldwide revenue of all internet services is barely a 12th of that of Minitel”.

The author was dismissing the internet because of its technologies: it was unsecured, non deterministic and chaotic. We all know what happened next.


If you’ve followed me for a little while, you know that I have only ever owned Apple computers. There was no doubt in my mind, but also that of many my peers, professors and friends that the Mac was a better technological platform. From its Motorola chips all the way to the OS, thru the SmallTalk network stack, people argued everywhere that the Mac had a superior stack than the Wintel environment… and yet, it it was not for the emergence of the web, which made both its superiority and difference irrelevant, it is possible that Apple would have disappeared before the early 2000s.


I was about to write about Betamax, Playstation 3, Lisp, and many more superior technologies which failed in front of lesser, but massively adopted technologies, but I want to go back to the web.

HTTP and HTML were indeed incredibly fragile and limited technologies. So much so that many things we take for granted today started as “hacks” or “layer 2” technologies on top of of HTTP and HTML. For example, cookies were invented to circumvent the “stateless” nature of HTTP. JavaScript, which is now the language in which everything is rewritten started mostly has a hack to make these static HTML documents a bit more dynamic. Similarly, Ajax was definitely a hack to go around the impossibility to create interfaces which would self refresh and reduce the latency associated with the infinite dance of clicks and full document reload.

An area where this is the most fascinating is around video. Early, the web was often compared to a better TV system, one that users could control more. But if you ask anyone around the web from the mid 90's to the mid 2000’s, they’d tell you that there was no way the web would replace the TV streaming protocols… and yet, today, Netflix, Youtube, Hulu are now only using web based technologies. Data is transferred using HTTP and served inside HTML documents with the <video> tag.

Don’t get me wrong, to get there, people invented dozens of approaches at all levels because the web was so limited: from RealNetworks protocols to Flash, through complete stacks like Joost. All of these were MUCH better than what the web stack had to offer… and yet, here we are now, all watching HD videos from our web browsers without thinking about it.


In retrospect, what is clear is that the limits of the protocol shaped the types of applications designed on top of then: the streaming web was not really a thing before the explosion of Youtube (ten years in!). However, Wikipedia and its “book like” metaphor was made possible from the beginning.

The relationship between a platform and the apps that are built on it is symbiotic: they need each other more than one needs the other. The web platform needed some successful apps for people to subscribe to their dialup provider — and I am maybe too young to remember what was exactly the app that people used in the mid 90’s. But it is also obvious that those who assumed they could replace the platform by providing a better technical approach to help build them their specific app were eventually crushed by the network effects generated by the platform they tried to avoid and replace.

So the challenge for app developers becomes to identify what are the true characteristics or the real innovations of the platforms on which they’re building, rather than try to coerce the platform into doing things it cannot do, until these things become possible.

This is the reason why I believe that eventually we may see better social networks using blockchains, but given the scale that these would require, for now this is not only undesirable but also impossible.

“A castle on an island in Scotland with a bridge and a mountain in the background” by Sorin Tudorut on Unsplash

If you read this and it felt like a giant subtweet, you are right, I am talking here about all of the “Layer 1” smart contract platforms which are better than Ethereum. I have no doubt that they are incredibly designed by talented researchers and engineers and that, indeed, they allow for significantly better scalability, throughput, latency, security or even deployment costs…

I also understand that several ÐApp developers such as CryptoKitties or Kin may be frustrated by the limited capabilities of the current Ethereum blockchain. I am also worried that it won’t support Unlock at scale and I am already struggling with extravagant gas prices.

But what I fail to see, is how any of these more advanced Layer 1 is going to compete with the network effects around the Ethereum ecosystem. At this point, everyone knows that Ethereum is limited and yet, this is where everybody starts to build, this is the platform for which dev tools are being developed, and the platforms which has already too many wallet applications… Each of these are compounding network effects.

Every day spent optimizing transactions per second on another Layer 1 platform sees hundreds of new engineers embracing Solidity, Web3.js or Metamask and making the limited Ethereum as the only platform of choice.

In closing, I want to point at this tweet by Denis Nazarov, which, I believe threads around the same idea (despite him being such an EOS fanboy!):