I think it’d be useful to be able to express probabilities in odds form. I have trouble explaining probabilities to regular people, but I can often explain odds easily in terms of money or in terms of people (e.g. “20:1 means that out of 21 people, 20 do X while 1 does not do X”).
pargui
Karma: 0
At the heart of this discussion seems to be the issue of whether it makes sense to miss out on good “investment” opportunities just because you can’t quite measure the impact.
I’m an outsider so make of this what you will but it seems to me that EA is bound to have a measurability bias to it because, if not, then it starts to look like old school charity
Obviously what you’d really want to do is give a lower bound of how much benefit society might be getting and see if the numbers work out when comparing to alternative uses
This discussion is going nowhere without that last kind of analysis imo, and even then it might just be endless but I’d rather see someone try
Reminds me of the predictability (or not) of Black Swans, aka Tetlock v. Taleb. Also Tetlock’s point that nothing is truly unique: you can usually learn at least something from similar cases/reference classes. (I know @Eliezer Yudkowsky isn’t saying you can’t learn anything at all). So the question is how much can you learn beforehand If frontier-lab safety people think they can learn a great deal from model to model, that would be important evidence for me (do they?).
Conversely, if the claim is that transfer from previous systems to ASI is necessarily too weak, or that one future step is crucially different from all previous steps, that needs an argument beyond just asserting it and is very suspect from the forecasting experience. The prior should be against the qualitative/unpredictable/uncontrollable sudden jump.
Tbf, I may very well missing a lot of context and maybe that argument has been made elsewhere.