LessWrong 2.0 Reader

Oh, weird. I always thought "ETA" means "Edited To Add". 

ETA = edit time addition

I should probably not use this term, I think I picked up this habit from some other people on LW.

I agree with ~all of your subpoints but it seems like we disagree in terms of the overall appraisal.

Thanks for explaining your overall reasoning though. Also big +1 that the internal deployment stuff is scary. I don’t think either lab has told me what protections they’re going to use for internally deploying dangerous (~ASL-4) systems, but the fact that Anthropic treats internal deployment like external deployment is a good sign. OpenAI at least acknowledges that internal deployment can be dangerous through its distinction between high risk (can be internally deployed) and critical risk (cannot be), but I agree that the thresholds are too high, particularly for model autonomy.

"ETA" commonly is short for "estimated time of arrival". I understand you are using it to mean "edited" but I don't quite know what it is short for, and also it seems like using this is just confusing for people in general.

Who is updating? I haven't seen anyone change their mind yet.

Wasn't edited, based on my memory.

Yeah, that's probably the rationale

I interpreted the comment as being more general than this. (As in, if someone does something that works out very badly, they should be forced to resign.)

Upon rereading the comment, it reads as less generic than my original interpretation. I'm not sure if I just misread the comment or if it was edited. (Would be nice to see the original version if actually edited.)

(Edit: Also, you shouldn't interpret my comment as an endorsement or agreement with the the rest of the content of Ben's comment.)

Sticking to multiples of three does have a minor advantage of aligning itself with things that there are already number-words for; "thousand", "million", "billion" etc. 

So for those who don't work with the notation often, they might find it easier to recognise and mentally translate 20e9 as "20 billion", rather than having to think through the implications of 2e10

I agree the lower bound for output isn't very tight. I'd be very interested to hear other simple rules of thumb you could use to provide a tighter one.

I'll add a note to the section on input tokens that since they don't require KV cache, it's possible to get much closer to the upper bound.