Chains of community
When I’m at an effective altruism event, people will often ask how I first got into EA. I sometimes tell them that before there was an effective altruism community, it was the rationality community, which used to be the Singularity community, which used to be the transhumanist community, which used to be the Extropian community, which I first heard about way back in 2003. Obviously, that’s a bit exaggerated. There are many people in EA who’ve joined recently and who wouldn’t identify as rationalists, and many rationalists who wouldn’t identify as Singularitarians, and so on. But these communities did all blend into each other. And I’m often surprised by meeting someone, and then learning that I’d been at an event or on a mailing list with them five or seven or ten years earlier, and I just hadn’t noticed at the time.
As a quick experiment, I went back through the archives of the SL4 mailing list, which was active from 2000 to around 2009. I made a list of everyone whom I had no idea were there at the time (so excluding, eg., Eliezer Yudkowsky and Ben Goertzel), but whose name I recognized from later projects. There turned out to be quite a few (dates are of first SL4 post): Adam Safron (2007), who worked with us at MetaMed Anders Sandberg (2001), now at FHI at Oxford Andreas Stuhlmueller (2004), now researching AI at Stanford, coauthor of agentmodels.org Austin Parish (2005), worked with us at MetaMed Benya Fallenstein (2009), a researcher at MIRI Edwin Evans (2001), chairman of the board at MIRI Gwern Branwen (2007), now independently researching cryptocurrencies Maksym Taran (2008), formerly at Quixey, now at Google Moshe Looks (2002), now my manager at Apprente Perry Metzger (2003), computer security researcher in New York Randal Koene (2001), neuroscientist and lead scientist at Kernel Shane Legg (2001), co-founder and chief scientist at DeepMind Stuart Armstrong (2008), AI researcher at FHI Rarely are groups, especially large and diffuse ones, created completely from scratch. They build on what came before. Conversely, how a group is run now can have effects many years down the road, sometimes long after the group itself dissolves.
One obvious implication, which I’ve been thinking about more lately, is that the cycle will continue and something will eventually replace Effective Altruism. Exactly what, or when, I think no one really knows. (I think there are some signs that it’s getting close now, which I’ll discuss more in another thread.) But just knowing that it’s likely to happen is, I’d propose, helpful in planning for what comes next.
Another implication is that I’m definitely worried about Tumblr. Most of the people above were students when they joined SL4, and since Tumblr is now popular with students, it may become the closest “equivalent”. Unfortunately, Tumblr’s features make it very difficult for strong communities to form there. The strongest communities are, of course, in physical reality, which has persistent identities (people remain the same), clustering (groups can form and exclude non-members), and high bandwidth (you can communicate a lot of stuff). Blogs, mailing lists, and the best subreddits have two and a half out of three; bandwidth is limited by text and by the high latency. But Tumblr has none of them. Identities are ephemeral, and constantly pop up and then vanish again. There are no subTumblrs, specialized groups, or other ways to create private spaces. And bandwidth is not only limited by text and latency, but by the threading system. In many ways, although it calls itself a blogging platform, I think Tumblr is more similar to 4chan than something like Wordpress or Squarespace. There aren’t really any groups unique to Tumblr yet, but when they get started, I’d expect them to (ironically) look like 4chan’s Gamergate: no leaders, manifesto, budget, or coherent agenda beyond causing chaos. That’s not a healthy environment, and it may have big consequences later on.