The recent Gordon Seidoh Worley/Said Achmiz blowup and the subsequent threads (1, 2) it spawned, along my own involvement in them, got me thinking a bit about this site, on a more nostalgic/meta level.
To be clear, I continue to endorse my belief that Said is right about most of the issues he identifies, about the epistemic standards of this site being low, and about the ever-present risk that absent consistent and pointed (reasonable) criticism, comment sections and the site culture will inevitably devolve into happy death spirals over applause lights.
And yet... lukeprog hasn't been seriously active on this site for 7 years, Wei Dai hasn't written a post in over a year (even as he engages in productive discussions here occasionally), Turntrout mostly spends his time away from LW, Quintin Pope spends all his time away from LW, Roko comments much less than he used to more than a decade ago, Eliezer and Scott write occasional comments once every 3 months or so, Richard Ngo has slowed down his pace of posting considerably, gwern posts here very infrequently (and when he does, it's usually just linking to other places), Duncan Sabien famously doesn't spend time here anymore, lsusr said an official goodbye (edit: it was an April Fool's joke) months ago...
While speculating about the private or subconscious beliefs of others is rightly frowned upon here in general, I will say I do suspect some of the moderator pushback to Said comes from the (IMO correct) observation that... LW is just missing something, something that Said contributed, at least a bit, to pushing away in the aggregate (even if any one given action of his was by itself worthwhile from a cost/benefit perspective). Something that every single one of these authors used to provide in the past, something that used to prevent "the project of thinking more clearly [from falling] by the wayside", something which resulted in "questions left in the articles for commenters to answer", something that's a bit hard to fully pin down...
Back in 2009, Eliezer wrote "Rationality: Common Interest of Many Causes" and talked about "Raising the Sanity Waterline" in broader society. He then wrote HPMOR; later on, he wrote Inadequate Equilibria, both of which were flawed but worthwhile books. Scott started the SSC and talked about everything, from science and rationality to politics and economics to medicine and social dynamics. There was a certain... vibe, for lack of a better term, connected with all this. It's what spawned the original LW wave of support for CFAR, from people who were not sufficiently plugged into the social dynamics on the ground to realize that was apparently never what CFAR was supposed to be about. It's what got people hopeful about The Martial Art of Rationality, a sense that a rationality dojo is possible. It's what's embodied in one of the best and most emblematic comments ever written on this site, namely gwern's pointed and comprehensive takedown of Eliezer's FAQ on technological unemployment. It's a sense of curiosity embodied in the virtue of scholarship. It's covering a breadth of topics for the sake of discussing them and becoming stronger and more knowledgeable.
Now, it's mostly just AI. But honestly, it's not even the long conversations or (somewhat) shorter debates about AI that used to generate and propagate tremendous insights. It's... different. Even when it's not AI, the conversation feels... stilted, lacking in a way. The engagement feels low, it feels off; the comment section isn't producing totally different yet insightful models of the problems discussed in posts and spawning long and fruitful conversations anymore. I'm not sure what to really make of it.
There are some who buck this trend. Viliam, Steve Byrnes, jefftk, Sarah Constantin... I'm sure I'm missing some names. But it's just not what it used to be overall.
Discord (like Twitter) gives you fairly powerful search primitives, to a greater extent than most people ever notice. You can filter by user, date-ranges, that sort of thing... It was written by nerds for nerds, originally, and it shows.
It seems worth noting that 𝕏 search has been broken for quite a while, and shows no sign of improvement.