Is this three cities that are at any point under Russian control before June, or are the three cities required to be simultaneously under Russian control?

@Sylvain given that commenting is less central to the usage of this site model than predicting is, I'm curious why the policy is to ban people for those kinds of comments rather than just removing their ability to comment.

@citizen I think the resolution date a year from now is an error, everything for the tournament is ending ~nowish

@(AlyssaStevens) that website isn't the official US DEFCON level, though it is probably a pretty good proxy. from the site: "Remember, the current defcon level warning system status today is not recorded as an actual official change until the United States government officially releases the actual current official warning system level for the military to the public. The live levels on this site are what our intelligence analysts believe the level should currently be based on open source intelligence (OSINT)." "Question: Are the 'live' alert levels on...

it would have been helpful if there was a way to order the questions according to the ordering in Scott's form, to speed up the data entry from the spreadsheet page that I have for my predictions.

huh, wonder what happened to the metaculus prediction in late september?

is there a quality component at all? a friend complained that the MinION/nanopore sequencers have a relatively high error rate, and it seems plausible that there would be something on the market marketing itself as a portable genomics sequencer with an even higher error rate. it seems likely that something very cheap but sufficiently low quality wouldn't be useful. > However, ONT reads (henceforth referred to as nanopore reads/sequences) have a high error rate (10% vs. 0.1%) compared to paired-end short reads generated on the Illumina platform (hencefo...
I think this is more likely than not. Between FLOP/s per dollar falling and algorithmic improvements (, training a GPT-4 equivalent will get cheaper and cheaper every year. Right now I believe the cost was ~100M for the training run and 5x that for an organization that is capable of launching that training run ( (can't just take 100M and push a button that gives you a GPT-4 equivalent after the right ...

does Russia actually regularly officially announce its nuclear readiness levels? I didn't think the US does for defcon levels.

— edited by isajun

they don’t seem to be listing the total since even the anti-mobilization protests stated, which is when they seemed to have restarted the count. right now the site is only showing todays arrests.

@(Unwrapped8600) one consideration here is that there's potentially going to be ~big swings in points/ranking due to crossing the 30% threshold, and it seems very plausible there will be a .05% decrease in the share of the vote that will affect points dramatically but not trigger the "significantly different" clause for re-resolution ("If the definitive count is significantly different (more than 1pp), then the question will be re-resolved.") I think that that if the definitive count is lower than 30% we should use the definitive count even it's only sl...

Is this just a way of asking whether Labour or Conservatives will head the next government?

If you think that there's a 60% chance Labour will form the first government after the election, doesn't it make sense to put 60% yes on all Labour candidates, and if they do win, then it will resolve ambigous for all Labour members who aren't leaders of the Labour party, and and yes for whichever was leader.

I guess this is also getting at the likelihood of being replaced between the election and forming the government. Does that happen a lot in the UK?

was this performance at all times, or just resolve time?

@humongousboy How do you use those to work out the chance of default?