Raymond Arnold

Secular Solstice guy

[Raymond Arnold @Raemon777]

@QiaochuYuan A thing that felt a little sad about the trailer was that it seemed it wasn't filmed in the "found footage" kinda vibe of the online shorts, which was a major source of the tension.

[Raymond Arnold @Raemon777]

It's a work in progress but I think it's got the essentials to be pretty fun for awhile. I played it a few hours over the past few days and noticeably improved at shooting-asteroids-in-time-with-the-beat.

I have aspirations of achieving some Cool Indie Game Vibes as I layer some more stuff in.

[Raymond Arnold @Raemon777]

I have made a Rhythm!Asteroids game!

If you like Asteroids*, but thought "what if this was somehow more like the scene from 2001 where the ship gracefully docks with the space station?" over and over again, have I got a game for you!

It is called "Pulsar."

[Raymond Arnold @Raemon777]

(Before making this game, I did not know whether Pulsars just spun faster than normal stars, or, like, fast enough to realistically be the metronome for a celestial orchestral symphony. The answer is, they totally can! They spin anywhere from "once every couple seconds" to "hundreds of times per second." Wowzers!)

[Raymond Arnold @Raemon777]

@Actuallykeltan the answer is "the internet"

[Raymond Arnold @Raemon777]

Hullo! Yeah that's still like my favorite class of experience and it's hard to try to cause it to happen to myself! Elizabeth keeps an eye out for movies that are good to go in blind. But there's not all that many movies that particularly reward that (and, most of them are Surprise Horror movies which I am also not that into)

[Raymond Arnold @Raemon777]

@QiaochuYuan you found this out after starting?

I am curious to hear like an after-the-fact-livetweet of your experience

[Raymond Arnold @Raemon777]

@diviacaroline Looking at this, I find myself wanting a big upper-2-floors, with columns supporting some kind of overhang that gives you a bit more "backyard/deck" vibe.

[Raymond Arnold @Raemon777]

@htmx_org I'm confused why field-level permissions don't just solve the graphql problem. The post just says "it's complicated" without explaining why.

(I use graphql with field-level permissions. Seems fine? What's the issue?)

[Raymond Arnold @Raemon777]

(from https://t.co/RrXuUqLnqf)

[Raymond Arnold @Raemon777]

wh... what are the units for the test-time-compute?

[Raymond Arnold @Raemon777]

@QiaochuYuan @WomanCorn I would call this "recursive improvement."

[Raymond Arnold @Raemon777]

@jamespayor @RichardMCNgo Yeah that does make sense for the straightforward reasons.

I think the reason it was on my mind is that there's a lot of things people might "think" are hope, but, actually are complicated kinds of social pressure that you mistake as hope.

[Raymond Arnold @Raemon777]

@jamespayor @RichardMCNgo I actually am not sure whether you meant hope to be a good or bad sign. I think you originally meant "good sign, if it's a particular flavor of hope"

[Raymond Arnold @Raemon777]

I agree with the "real curiosity/obsession" but something feels off about leaving it at that. (Scott Garrabrant said something like "go where the important problems are, then marry for love." I feel like that could be somewhat better operationalized)

Just pursuing curiosity doesn't feel that useful. I think it's tractable to cultivate curiosity in relevant domains. This still requires some good understanding/taste about what the relevant domains are.

[Raymond Arnold @Raemon777]

@RatOrthodox @adrusi I feel like they have circumambulated concepts that are not exactly this but triangulate it from different directions.

[Raymond Arnold @Raemon777]

@curiousgustaf It did a few weeks ago but I do notice it works less well now

[Raymond Arnold @Raemon777]

"Sometimes the AI just makes stuff up" is a problem I don't really expect to go away.

In the nearterm, AI is going to keep occasionally hallucinating, or misinterpreting information. Eventually, AI will be powerful enough we need to be worried if it's presenting misleading information on purpose.

There might be a nice window where the AI is powerful enough to not make things up but non-agentic enough that we don't have to worry about deliberate manipulation. But, even then, interpreting data is tricky.

I'm worried about this for my own use, but, I'm more worried about this on the global scale. I'm worried about people trusting things AI made up, and I'm worried about the internet proliferating with slop that makes it harder to even find original statements that are a human's real testimony.

An approach that might help is to make AI reports more "Verification centric." i.e. designed from the ground up to make it as easy and frictionless to verify as possible.