The first poll:

https://manifold.markets/SimoneRomeo/have-we-achieved-agi-already?r=U2ltb25lUm9tZW8

There should be people that believe that the current approach will yield results and those those that believe that a key discovery that will take necessary time to develop and then time after to develop AGI.

Either could be correct so I think the proper distribution should have two peaks. and isn't a straight up bell curve.

But people expect to see a normal bell curve and will bet to "Normalise" incorrectly.

@PaulDwyer I agree it shouldn't be a normal bell curve, but there's no obvious year for a second peak, so a twin peaked distribution doesn't make sense. There should be a left of median peak topping out sometime before 2030, and then a fat rightward "tail" putting nontrivial probability on ~every year before 2050.

This market currently says my expected value is 129, but the actual sale value (adding up the buckets) is 53. Anyone know if that's a bug on these new numerical markets, or a "feature" where "Expected value" actually means "value if you could theoretically sell your shares without shifting the market"?

If the poll is replicated many times, how do you plan to aggregate the results?

@TimothyJohnson5c16 what do you mean? As far as I saw all the polls were on the probability of this event happening on specific dates, this is the first time that Manifold allows us to create numerical markets. I'm not planning to aggregate results personally, I'm planning to let users bet on the specific year

@SimoneRomeo I see, so you mean every year you'll re-run the poll to see if we've reached the 75% threshold?

@TimothyJohnson5c16 ah, I see your point. I don't expect to run it every year but when there are outstanding breakthroughs. Also, anyone could run such poll and I would close the market accordingly