That seems like it could be a good idea. A few more questions though:
How would writing the question help to convince people? Would it not only be convincing in 5-10 years’ time if some of the predictions turn out to be accurate? Or, do you think if consensus on a Metaculus question that prediction X will occur is in and of itself convincing for rationalists?
I’m still a little uncertain about the practical benefit of writing questions, in helping to advance the technologies.
How would writing the question help to convince people? Would it not only be convincing in 5-10 years’ time if some of the predictions turn out to be accurate? Or, do you think if consensus on a Metaculus question that prediction X will occur is in and of itself convincing for rationalists?
I would personally find a consensus on Metaculus pretty convincing (at least, conditional on there being a significant amount of predictions for the question). I find it hard to gauge other people’s expertise and how much to defer to them, especially when I just see their point of view. Aggregating many people’s predictions is much more persuasive to me, and many of the top Metaculus predictors seem to have good epistemics.
That seems like it could be a good idea. A few more questions though:
How would writing the question help to convince people? Would it not only be convincing in 5-10 years’ time if some of the predictions turn out to be accurate? Or, do you think if consensus on a Metaculus question that prediction X will occur is in and of itself convincing for rationalists?
I’m still a little uncertain about the practical benefit of writing questions, in helping to advance the technologies.
I would personally find a consensus on Metaculus pretty convincing (at least, conditional on there being a significant amount of predictions for the question). I find it hard to gauge other people’s expertise and how much to defer to them, especially when I just see their point of view. Aggregating many people’s predictions is much more persuasive to me, and many of the top Metaculus predictors seem to have good epistemics.
Great—have you seen the existing Metaculus questions on anti-aging by Matthew Barnett and others?
Ooh, no. That’s super interesting, thanks!