Happy to see the community prediction matching my prediction (2%). The US has been around about 244 years and has spent 4 of those in civil war. That gives me a baseline rate of around 1.6%. The only two things that would make me update upwards are 1) recent protests and 2) election crisis. And the protests themselves at their peak this year still weren't close to civil-war level. On the other hand, compared to when we actually had a civil war, the nation is way more connected, way wealthier, and there is no broad support for secession. At the end of...

@emejec wrote:

Trump will win other con tester hands down.

"Hands down"



It seems to me the chance is greater that the RBG replacement will hurt/help Trump more than his debate appearance. And since the timelines may coincide, it seems like the bigger factor in deciding this question.

Hopefully my high school knowledge of physics doesn't embarrass me, but I want to note that in the gaps in our physical knowledge lie many possibilities: It's possible that spacetime is huge/infinite and there are "big bangs" that are happening throughout it that just haven't reached our observiball. So it's unknown whether total energy is conserved in the universe. Moreover, it's possible that even if we have heat death of all currently present matter, sometime in the future there will be another big bang in the same area of spacetime. Who knows what ...
  • Predict US GDP X years from now (e.g. 5).

  • Predict the local minimum of the US unemployment rate when we do reach it (within a 5 year range, say).

  • Predict first company to reach $1T nominal market cap.

  • Predict when % of people living in extreme poverty will hit 3%.

  • Predict global mean surface temperature in 2025, 2050

  • Predict US debt in 2028

  • Will Trump approval rating ever climb up to 48%

Starting with Q1 of 2018 and working backwards by quarter, here is their subscriber data from shareholder letters: 125 117.58 109.25 103.95 98.75 93.8 86.74 83.18 81.5 74.76 69.17 65.55 62.27 I did not include data going farther back than q1 15. Linear extrapolation for 17 more quarters suggests 210 million subscribers. Exponential extrapolation gives about 330. Quadratic is about 255. Which type of trendline is better? It seems that I can predict 2018 data with data up to 2016 better with an exponential line than a linear one. However, th...

@randallburns Unclear that it's disinformation specifically, but it's certainly overblown. 9 ballots were found, 7 for Trump.

9% Metaculus prediction?? :O Should I be freaking out? :O

PA is veering hard toward Biden in the counting. Biden is likely to win it soon


@KungfuStat wrote:

He got money, which is everything for control

Not actually true. The connection between money and reelections is tenuous at best. He also doesn't even have the most money.

More than 2000 predictions on this question? Is someone trying to game the system?

— edited by Fruo

Maybe I'm overconfident in a Biden win atm. Start with the known current map at https://abcnews.go.com/Politics/2020-Electoral-Interactive-Map and incorporate info from 538 on what remains to be counted: https://fivethirtyeight.com/live-blog/2020-election-results-coverage/ Nevada and Michigan go to Biden. NC likely goes to Trump. That leaves GA, PA, and AZ. Trump needs all of them to win. In GA, Trump leads by 80k votes, with 250k votes left to count. As long as he gets a 90/160 split (36% of the vote), he wins (doesn't seem wildly impossible). For A...

36% chance of a major earthquake in a 3-year span? Unlikely...

Feature idea: tournaments (?) where everyone who participates has to give predictions for a certain set of events. Right now, my Brier score is really good because I only estimate for questions that I have a decent model for with high confidence. It means it's not a good metric to compare me to someone who has a worse score but has predicted on hundreds of questions. Tournaments where everyone predicts the same set of questions make for a more fair comparison :)

@Anthony Agreed with all others. $10B is critical to the question.

LA Times (ranked 21 on the list of media outlets) covers it:


The question makes no mention of verification...

@Uncle Jeff wrote:

The best chance at a positive resolution will be on April 1.

Then we'll really have a hearty debate whether a joke article really is one that claims there is "reasonable evidence" for the existence of aliens :P

Add a swiss army knife of statistical calculation tools to aid in estimation. Things like a binomial calculator would be great. They could exist behind a button floating on the bottom right of the screen.

Other than that, I think the primary thing would be more activity on the website, more questions to predict on - so just getting this website in front of more enthusiastic and numerate people. Perhaps the rationalist/EA communities would be good for this.

If I could split the percentage here, I'd drop from 2% to 1.7%.