@Anthony Have you considered doing a Kaggle-esque prize for improving the metaculus prediction algorithm? I bet you could significantly improve the existing one w/o it taking much more time from your team.

We need a separate question estimating the likelihood a plurality of organizers are negative utilitarians in order to estimate whether COVID causes them to cancel the conference or invite more people.

My compliments to the organizers of this prize series - I've seen many discussions on the internet point to the Metaculus questions on the Coronavirus, and use it to help ground discussion in quantifiable estimates. Obviously the key question of accuracy will be known later, but I think it's a great showcase of how centralized aggregations of predictions can be useful.

The AlphaStar Final bot has replays that shows its opponents ranking. I'm working through the replays and scoring the ones where it played a grandmaster. - We're going to resolve based on the total performance of AlphaStar across all three races. - Per the question resolution details we'll only evaluate games played as part of the Blizzard online experiment, so games played later in 2019 won't count and we can resolve the question now. It's likely this will resolve in the negative since in total AlphaStar won less than 80% of these 90 games, but it's ...

To be clear while showcasing the usefulness of Metaculus is great, the possibility of a global pandemic is not great. I am not pro pandemic.

EAG sent out an email saying they’ve appointed a board of advisors to help determine whether they should continue to hold the event as planned, postpone, or cancel.

— edited by ghabs

The question resolves negatively

When playing against grandmasters:

The average win rate of AlphaStar across all three races: 55.25%

  • Protoss Win Rate: 78.57%
  • Terran Win Rate: 33.33%
  • Zerg Win Rate: 53.85%

Detailed match by match scoring

@Tamay I think it should resolve as ambiguous to avoid confounding estimations. I might write a separate binary question that asks Will Libra launch by XX date.

Wuhan already fulfills the criteria with a population of 11 M and at least one report of food shortages, though there seem to be disagreements about whether there are food shortages in Wuhan.

Experiment! I'm crowdsourcing suggestions for a sub-question that would be highly informative for this question. E.G. what question should we forecast that would reveal a lot of information about whether Brexit will happen by Oct. 31st? You can vote or submit your own questions here:


@Jgalt It was a brutal exchange. According to the prediction markets this debate cut his chance at the nomination in half.

@casens They also need to take into account known-unknowns and unknown-unknowns. For instance there's concern that post-viral fatigue and other long term debilitations are worse w/ the coronavirus, but it's currently unknown.

@steven0461 The costs to the ties within the community would be much higher if an EA event led to more infections. Something like "leadership didn't prioritize the communities safety" and/or an implicit association that EA conferences lead to bad things happening. The later is my model of why many other conferences have already been cancelled (1, 2)

On at least 125 occasions, the President has acted without prior express military authorization from Congress

While it seems possible that the US will initiate additional military action against Iran this year (20%) it's almost certainly not the case (7%) that the president would ask for and receive congressional approval.

Bill Barr was vehemently opposed to Snowden receiving a pardon - his departure might be over upcoming Trump pardons.

This is awesome, thanks @Max

@DanielFilan@Anthony Smarkets has political markets on the Democratic primaries, but they are lower volume than the PredictIt markets. For example only £469 has been traded on the California outcome.

I'm estimating a 75% chance that Biden wins the SC primary. If he does win the primary, I don't expect him to drop out. In the scenario where Biden is still running on Super Tuesday, I'm assigning equal weights to Bloomberg, Biden, or Buttigieg winning the moderate lane in any given state, and slightly lower odds to Klobuchar. ~27.5% for Bloomberg in this scenario. If Biden loses SC, or underperforms enough that the media can call it a loss, I expect Bloomberg to have much better odds of winning the moderate lane (40%-60%). Roughly averaged together ...

In a situation where no candidate receives a majority of delegates and we enter a brokered convention, I expect Sander's to be at a significant disadvantage (< 25% chance of securing the nomination). Given the 538 odds of 1 in 3 for Sanders, and 2 in 5 for no clear majority, I estimate low 40% odds.