Open Thread 249
This is the weekly visible open thread. Post about anything you want, ask random questions, whatever. ACX has an unofficial subreddit, Discord, and bulletin board, and in-person meetups around the world. 95% of content is free, but for the remaining 5% you can subscribe here. In other news:
1: In the comments of Moderation Is Different From Censorship, some people asked if my moderation policy here was hypocritical. I answered that I would be happy to implement some kind of see-all-banned-posters policy here if it were possible, but I’m limited to Substack’s features, and that means making a blanket decision about whether everyone should see a certain person or not.
2: Related - the following people have been banned (links go to representative bad comment): Jason MacGuire, Timothy Buckmeister II, Les_Bergers_Des_Photons, PEG, Ludex, Bernard Gress, Anti-Homo-Genius, Embrace Christ, LutherFischerKennedyKaczynski, Thad, Bobby Bigdick, Lazarus (I assume he’ll return), Bokra, …, and Ian Duncan. The following people have received new warnings: Angus, G Retriever, Descriptor, Impassionata, Chaz Gibson, Jay Rollins, Machine Interface, Machine Interface again, IICS, Essex, KillerBee, Kamran, Calion, Cosimo Giusti. Thanks to everyone who reports comments using the “Report Comment” button hidden in the (…) sign below each comment.
3: An especially common offense this time around was “low-content high-temperature comments”, like responding to someone with “this sucks” or “I can’t believe Republicans are so dumb”. It’s entirely possible that something sucks or that Republicans are so dumb, and you have the right to make that argument here! But please actually make the argument instead of just saying an insult; otherwise you will be banned.
4: Related: I’m a little unhappy with Open Thread quality lately. Hopefully the latest purge will improve things a little, but I’m going to experiment with having Wednesdays be Challenge Mode Open Threads. In a Challenge Mode Open Thread, I‘ll delete all comments below maybe what the 33rd percentile comment in a non-Challenge-Mode thread would be. You won’t get banned for making these comments (unless they’re so egregious they would get banned usually), but your comment will get deleted. I’ll see how this goes and decide how long to continue the experiment for. This Open Thread today is not Challenge Mode.
5: Comments of the week: Steven claims an alternate mode of entering jhana different from anything I’d ever heard before; I’m tempted to disbelieve, but two readers say it works. The Genealogian gives genealogy advice. And LoreSnacks says we can’t blame the Santa Clara water district guy for OKCupid - sorry, Santa Clara Water District guy (though he still seems potentially bad in other ways).
6: A reader who asks to remain anonymous writes about one of the Buddhist sex scandals I mentioned in last week’s Highlights post:
While I know very little about Culadasa's life, I did listen to a couple of podcast interviews with him. I am fairly sure it was a podcast named Guru Viking (episodes 13 and 78). In the latter episode, to the best of my recollection, Culadasa argues that his "scandal" was not sexual in nature, but instead involved giving some money to support a sex worker (or former sex worker, not sure?) and that he didn't have a sexual/romantic relationship with her. His story is that he was separated from his (later ex) wife and that their finances were essentially separate and they were effectively in an open relationship at this point. His ex-wife used the allegations as a way of extracting more from him during their divorce and getting control of the income from his organization. I don't know if his story is true, but I found his perspective fairly believable and, if true, would indicate that his scandal was not driven by sexual desire/behavior. Sadly, since his death, the truth of his claims will largely be hard to resolve.
7: Effective altruist charity evaluator GiveWell (mostly handles the global health and poverty side of things) is hiring new research analysts:
8: And Redwood Research, an AI alignment organization I’ll be writing more about shortly, is looking for applicants for its upcoming interpretability program. They write:
Redwood Research is running a large collaborative research sprint for interpreting behaviors of transformer language models. The program is paid, and takes place in Berkeley during Dec/Jan (depending on your availability). Previous interpretability experience is not required, though will be useful for doing advanced research. More information is provided here.