search

quantified-uncertainty / potential-projects

2 stars 0 forks source link

In-house forecasting team #8

Open OAGr opened 2 years ago

OAGr commented 2 years ago

Have 2-4 people, in-house, doing forecasting of key things. This would be used for forecasting with low weights/stars. It would hopefully speed up work by a lot, because right now we need to wait on Metaculus for most things (this requires waiting for them to post the questions, which could take months, then hoping/waiting for forecasters to forecast on them)

uvafan commented 2 years ago

I like this idea overall (it could even spin off into a competitor to Good Judgment superforecasters if it goes well). Why do you propose using it for forecasting with low weights/stars?

OAGr commented 2 years ago

They seem more neglected. There's a decent bar for metaculus questions.

uvafan commented 2 years ago

My independent impression pretty strongly disagrees with this or I'm misunderstanding you. Would be good to chat about this too

OAGr commented 2 years ago

My fairly strong impression is that lots and lots of interesting stuff isn't estimated. I think this has been a major oversight of academia. For example, think of "microcovid, but for many other things". This might be a longer discussion though.

NunoSempere commented 2 years ago

I think that Eli disagreed with Metaculus being a decent bar, not with it being more neglected.

NunoSempere commented 2 years ago

I agree this proposal would just straightforwardly be good. I think the main bottleneck is actually QURI funds here.