One of the most persistent cycles in the history of computing is the oscillation between centralization and decentralization. This cycle becomes entrenched because each has distinct advantages and disadvantages: centralized solutions often yield better economies of scale, allowing for higher quality artifacts — but they can also easily become stifled, reluctant to innovate for fear of disrupting the good thing they have going for themselves. Decentralized ones, by contrast, can be a mess but they democratize innovation, where a “worse” solution by some metric is in fact vastly preferred because it is much better by another metric (e.g., cost or convenience).
With the collapse of Twitter, we are seeing this oscillation in spectacular fashion in a new domain: social networking. In our Oxide and Friends discussion with Kris Nóva, Nóva made the observation that the decentralization of the Fediverse allows for people to experiment with different ways of doing things by running their own instance — “the control and freedom to build your own community” — and that she particularly loved the influx of folks with large scale internet experience into the much smaller scale of running a Mastodon instance. This observation was a bit of an aha moment for me, because it made me realize the stunning resemblance that Mastodon bears to a much earlier revolution: the homebrew computing movement of the late 1970s.
Homebrew computing — as epitomized by the Homebrew Computer Club in Silicon Valley — consisted of hobbyists designing and assembling their own computers. But like Nóva and her fellow Hachydermians, these hobbyists were in fact experienced engineers; their “homebrew” was just them building for themselves, for fun. Yes, homebrew computing was fragmented and incoherent; the participants at the time were not possibly thinking that the machines that they developed would form the basis (for better and for ill!) for all future computing. But by the late 1970s, the time for radical decentralization of compute had indisputably arrived — and its spirit would burn so bright that it would endure long after the homebrew became commercial, as viscerally expressed in Apple’s iconic 1984 Superbowl ad.
So it is now with social networking, where centralization is frankly long past its “Disrupt By” date. The progression here now seems painfully clear in hindsight: advertising-based models demand engagement — and algorithms will naturally observe that when we are enraged, we stay engaged. If follows that when algorithms are promoting content that enrages us, the social network will — either implicitly or explicitly — select for those that are most divisive. Not only has this has encouraged us all to be more divisive (viz. “hot take” entering the lexicon), it has embraced people who are divisive to their core, elevating them to heights that frankly would have been impossible without centralized social networking. The net effect of all of this is that while we may find this engaging, we certainly don’t find it to be enjoyable: I (like many, I suspect) have resented the attention of mine that it has held — like an extraordinarily bad reality TV program that I can’t stop myself from bingewatching. So while it is tempting for us to look at the destruction of Twitter as a catalyzing event, in truth I think it’s merely an accelerant: the model has been broken for a long time, it’s just now being speedrun to the logical extreme. (I would also add that GenZ may have already figured this out — and that my own kids view the destruction of Twitter the way I might view a collapse of LinkedIn: with ironic detachment and some sense of disbelief that anyone could really care that much.)
What does homebrew decentralized social networking mean in practice? As he has so many times before, Tim Bray has an excellent piece on this. Part of what makes decentralized social networking so appealing is that there is no algorithm deciding engagement. Rather, the stuff that’s promoted — boosted, in Mastodon’s parlance — comes purely from folks that you follow. That is, you only see a strict timeline. The effects of this are both surprising and refreshing. I have said that it’s harder for things to go viral on Mastodon, but my own experience is in fact more nuanced: things can go viral because boosting is a lighter operation that retweeting — and when they do go, they seem to go further and longer. Take, for example, Tim’s post on his blog entry: as I write this, that has 295 favorites and 595 boosts. Tim has 7.1K followers on Mastodon; you would never see numbers like that on Twitter, where likes will essentially always dwarf retweets.
All of this points to an even larger decentralization: that of the social networks themselves. In the last few years, some began looking at the ratio of their followers to the number that they follow, seeking to maximize it as a show of social dominance. If it needs to be said, there is a really easy way to do this: just unfollow everyone! This act feels innocent (if juvenile), but it is in fact deeply isolating: it is opting to be fed exclusively by the algorithm, losing any tether to genuine social connection. It is perhaps unsurprising that the people that I saw do this rapidly descended into dark, bizarre thinking, where they saw the world as increasingly conspiratorial and contentious. Being stingy with follows also nullifies one of the important values of social networking: hearing entirely new voices. Speaking for myself personally, this was especially important after #MeToo in 2017 and George Floyd in 2020, when new people that I followed elevated yet more voices that needed to be heard, leading to more new follows — and more new voices in a virtuous cycle. In decentralized social networking, this virtue is elevated, if by default: there is no algorithm, so if you insist on following no one, you will see… nothing. (As my mother was fond of saying when I found an empty mailbox as a kid: “If you want to receive a letter, you need to send a letter!”) There is something beautiful in this, and I would like to believe that a world in which people need to simply follow more people — to listen more! — will foster more temperance and less divisiveness. (And perhaps even the occasional apology?)
All of this brings us back to another point that Nóva made: “any large change has been uncomfortable and unexpected.” This was true of the homebrew computing movement where, important as these machines historically were, they were not easy to use! We can expect some discomfort in our future, but we can also expect fruitful experimentation — and surely some important interactions that centralized social networking never would have allowed!
Finally, a programming note: Adam and I were early adopters of Twitter Spaces, turning it into our Oxide and Friends podcast. We, like Tim, are leaving, and will be moving to Discord. Fittingly, for our first Discord recording tomorrow (Monday November 28th, 5p Pacific), we will be joined by Tim to get his perspective; join us!