Ten years ago, Twitter was already messing with its users and decided to forbid third parties from using its API. At the time, I was already writing about the notion of critical mass for social networks. I was also trying to introduce the idea of a dilution point for the value of such network effects, reached when you get too many connections.
Here are the core ideas written in 2012 (I will let you decide ten years later how pertinent they still even though, at the time, I thought that a Twitter competitor would arise – since then, I learned my lesson):
(...) Once the threshold of critical mass has been exceeded and that of the mass that I call "virtuous" reached, you risk reaching a dilution point for the value of the network effect you're building. Too few connections and there's no value; too many connections and the value starts to disappear.
An example of a dilution point being exceeded or at risk of being exceeded is evident with Twitter. What is the relevance of this site (which I have been using with pleasure for quite a long time) when in your feed you have hundreds of people you do not know (but whom you follow out of politeness or Pavlovian reflex), who disseminate uninteresting information and who, when there is the slightest fascinating information, will end up broadcasting it in a loop for 24 hours to the detriment of everything else?
Too many tweets kill the added value of Twitter.
(...) Sites like Linkedin are faced with the same problem. What is the residual value of these connections when you have several hundred professional contacts on the site (because Linkedin needed to facilitate excessive connections to reach a critical mass)? The network effect of a former work colleague working in a company that you are looking to approach will be very different from the effect obtained from the colleague of the colleague of someone you met for just a few minutes at a trade show! This is quite the dilution point for the network effect.
As a social network, when too many people will end up discrediting what you want to do? And how to estimate this notion of "too much"?
Back in 2022, we could argue that if we had less than five hundred connections each on Twitter, most of the issues of the platform would magically disappear. The chances of having trolls retweeting ineptitudes or being bombarded by lame product offers would be meager because you'd have to curate your network. Now five hundred might be too low to make Twitter an exciting platform. Not enough critical mass to have surprising and delightful information popping up unexpectedly into your timeline. But then what would be the line that, until you cross it, allows you to increase the value of the network effect and that, past it, decreases it?
You might find an answer in the business model itself: the more pro your social network usage is, the more you could extend your network, and the more you'd have to pay.
If we let go of Twitter (yes, please!), this discussion is critical for any other network effect-based business model.
Consider Airbnb and how one of its core values is about providing unique home-away experiences. There is a point at which if you have too many apartments in a city center, you hit a quality wall. A point at which many crappy offers will creep up, diminishing the trust of your guest customers in what the site offers. Worse, hosts inevitably get so pro at Twitter that they refurbish an entire downtown building in standardized sleeping units with connected key box access to minimize friction. Congratulations, you reinvented the hotel.
For so many internet businesses invented ten or fifteen years ago, the core metric was not just growth but GROWTH!! The ones that survived, fueled at the time by cheap VC money, are now facing a cliff. Growth metrics have become toxic, and, in any case, cheap money has disappeared. Despite so much superficial enthusiasm, sites like Mastodon or Post.news ends up falling into the same growth trap.
We'll see what the future holds for network effect businesses, but I'm not sure anyone will learn the lesson. VCs will still want maximum reach and then disinvest with a massive premium before adverse side effects appear.
Let's have this discussion again in 2032? 🙃