3.3 KiB
Short one-sentence description of your proposed project
Write a twitter bot to record, resolve and tally predictions and bets.
Longer description of your proposed project
I want to expand on the https://twitter.com/metaforecast (@metaforecast) twitter bot, to also record, an later resolve, the bets and predictions people make. I plan to implement the Predict, Resolve and Tally interface which I've already designed (https://github.com/NunoSempere/PredictResolveTally.)
I feel that this has a small chance of raising the sanity water-line of Twitter by a small percentage (because people could then more easily make bets, or record predictions that other people make.)
On the one hand, I think this would be valuable, because it would make bets from experienced prediction market people (such as [private]), more accessible to the general public, because I could add them to metaforecast.org (as ACX forecasts (https://metaforecast.org/?query=%20&forecastingPlatforms=AstralCodexTen), Peter Wildeford forecasts (https://metaforecast.org/?query=%20&forecastingPlatforms=Peter%20Wildeford) or OpenPhil predictions (https://metaforecast.org/?query=%20&forecastingPlatforms=GiveWell/OpenPhilanthropy) already are)
On the other hand, this has the potential to hold pundits accountable. The original Alpha Pundits challenge (https://www.openphilanthropy.org/files/Grants/Tetlock/Revolutionizing_the_interviewing_of_alpha-pundits_nov_10_2015.pdf) had the problem that pundits didn't want to participate. But because today, pundits are on twitter, they can be made accountable by recording which of their tweets make statements about future events.
On the third hand, this can be used by regular people to tally their own bets. This method would just remove the trivial inconvenience of having to keep a spreadsheet, but it might also have a good effect because people might mimic other people making predictions in public.
However, I feel that this kind of project right now is not really worth my time, because my ability to make something go viral is quite minimal, and this kind of thing would only be valuable if people actually used it. Thus, the project would only be valuable if it is widely used. So this kind of thing would only make sense within something like ACX++ (as opposed to EA Funds, or practically any other funder.)
See also: https://sideways-view.com/2019/10/27/prediction-markets-for-internet-points/
Describe why you think you're qualified to work on this
I'm a fairly good programmer (https://github.com/NunoSempere/PredictResolveTally). I already have a twitter bot working (https://twitter.com/metaforecast/with_replies) in an adjacent capacity. Most importantly, I've thought a whole lot about what makes forecasting and predictions useful (https://forum.effectivealtruism.org/s/HXtZvHqsKwtAYP6Y7, https://forecasting.substack.com/, https://quantifieduncertainty.org/)
How much to ask for.
$10k, with the option to ask for more if it becomes widely used (like @threadreaderapp)
Estimate your probability of succeeding if you get the amount of money you asked for
95%: I create a functional bot. 50%: It gets used regularly by a small number of people. 10%: It reaches cultural fixation, like @threadreaderapp. <5%: It meaninfully improves Twitter discourse. <1%: It meaningfully improves American discourse.