@silly They don't always. But this way, the ability to predict months into the future is tested, whereas when they don't close until they resolve, they are not testing this long term prediction to the same extent as people will know they can update later. In many real world situations, long term predictions are needed for actions which cannot be changed easily later (e.g. business deals).

@(Jgalt) The online reaction to this footage seems very favorable to the defendant. The main thing, as I see it, is that the famous I can't breathe line was actually uttered many times before he was on the ground, before the knee on his neck. My understanding of the popular media narrative is that police essentially caused him to suffocate to death by the knee placement, while he was saying he couldn't breathe. But we now see that he was saying that all along even when not restrained much at all. Clearly, he was saying this for some other reason, presuma...

As the question maker, I intended for this to resolve positively with a pardon but I guess I got the legal details incorrect. Lesson to self is to explicitly handle edge cases...

@silly I hereby add a 100 US bounty for this feature, meaning I will donate 100 US to Metaculus whenever it is online and devs can share it as they deem fair.

@casens Good point. Yes, I had in a mind a healthy-looking newborn. We could stipulate it needs to survive at least some time span. We could say a year, that's a pretty conservative threshold, maybe 3 months is more sensible. From what I can tell, elephants have a life expectancy of 40-60 years in the wild. Apparently, captive elephants don't live very long, mean of 17 years only!, which is odd. https://www.nationalgeographic.com/animals/ar…

— edited by emilowk

New poll, same results: https://nypost.com/2021/03/09/andrew-yang-tops-democratic-candidates-for-nyc-mayor-poll/ Andrew Yang is the leading contender among a crowded field of candidates in the Democratic primary race for mayor, according to a new poll. Yang was the top pick to lead the Big Apple for 32 percent of the 644 likely Democratic voters who responded to a WPIX-TV/NewsNation/Emerson College poll that was released Monday. Behind Yang, Brooklyn Borough President Eric Adams was the second top candidate, according to the poll, garnering 19 percent...
Not remotely close for Italy in July poll: [How likely are the 'big four' European economies to vote to leave the EU if Brexit works out?](https://www.euronews.com/2020/08/10/nearly-half-of-italians-would-support-leaving-the-eu-if-brexit-is-successful-according-to): 43% stay vs. 31% leave (26% undecided). [Screenshot-from-2020-10-14-23-10-30.png](https://postimg.cc/3kP6mKck) Seems a lot of people estimated higher probability here based on the Italian result, but since it is now clearly no longer looking like that, my 3% probability seems more accurate....

Ability to compare own performance to other users who answered specific questions. For instance, if I have 5 resolved questions, I want to know my comparative performance on these questions, not the 1000s of other questions other users have answered (not strictly comparable to mine, only asymptotically under some conditions). Similarly, would be interesting to see performance on questions only from some time period.

For relative performance, nothing really beats being shown the distribution of accuracies, so one can see one's centile.

I made a formal model to project this, [seen here](https://twitter.com/KirkegaardEmil/status/1291433844088897536) ([details on rpubs.com](https://rpubs.com/EmilOWK/HDI_timesries)). The trends for countries are very close to linear, and if we give each country their own slope, we get a well fitting model. If we project with this to 2030, China will be at 0.89 HDI with a quite small prediction interval. Most of the uncertainty probably comes from the question of how nonlinear the Chinese growth pattern will be, and it could be tapering off soon, and might ...

US actuarial table to forecast death within a year. Data from all Americans and from 2016 data, so roughly applicable here.

If we look up at 58 woman, the chance of death in a year is 0.6%. Since we are forecasting about 0.5 year, this then becomes about 0.3%. The current median forecast is 90% chance of survival, meaning 10% chance of death. So the forecast death chance is 10/.03 = 33.3x relative risk! Seems like this is too much added risk. I was at 90% but I will move to 95%.

[For the rankings/high scoreboard](https://www.metaculus.com/rankings/?timeframe=resolved), there is already functionality in place to limit the time span of the questions, thus controlling for time spent. It would be nice if there was a column with average points per question. Right now, the top points earners in some period are people who spent a lot of time predicting on questions. Using the average score gain in some time period is an easy solution to this confounder. It would also be nice if it showed the number of resolved questions per person (as ...

Filter by Pending status for authored questions. I often submit questions, unfortunately, some are stuck in Pending forever, and can be hard to find. I found that "Open" includes pending, despite them not being open, so one can use "authored" and "open" to find them provided one wants to scroll a lot.

@(D0TheMath) I believe IA allows people to request to delete archives and they comply with these (e.g., [see this guide](https://www.joshualowcock.com/tips-tricks/how-to-delete-your-site-from-the-internet-archive-wayback-machine-archive-org/)). The other site does not do this. This is also why, I think, that site is censored by Twitter, Facebook and some other sites (that's why it keep changing its domain name). So I guess it means that Andrew Glidden filed a request to IA to remove this tweet from their archives. I don't know how to update my predictio...
@(Jgalt) This kind of thing was also seen in several Nordic protests. In Denmark, we intercepted some from Sweden and Germany at least once to prevent the inflow of a large set of left-wing protesters adding to a local riot or violent demonstration. There seems to be a small cluster of extremist protestors in many regions, and they travel around to keep attending the latest events. Here's a newspaper report from 2006 where 90 foreigners were arrested by Danish police, and 15 were banned re-entry for 12 months. https://www.avisen.dk/15-udenlandske-demon...
With one day to go before closing, I am reviewing evidence and making a final prediction. My priors: 1) Trump admin. is pretty incompetent, and has said many things they never did, meaning this could be another example, 2) Trump admin. does not like China, so maybe they will make good on this one threat, 3) Trump admin. does not have that much time left, and will presumably shift attention to election in November, 4) Trump admin. will probably face PR issues if they move forwards, but maybe they don't care. So, all in all, lukewarm and undecided. Googlin...
Given that the sharing private question feature is now live, I would like to suggest the follow up expansion feature: organizations. I am pretty sure I have seen this feature on a competitor's site, https://www.maby.app/pricing/. At least, I think so. I think of the feature like this: - Allow users to create organizations. - Organizations are invisible to public search. - Organization admins can invite users to their organizations. If they accept, all organization private questions are shown to them. - Organizations have their own admins/moderators and...

@Jgalt Only the actual homicide charge counts. Good question though.

@moderators @silly For the purpose of this question, when was "Actually this went live a few days ago but just now getting around to updating this thread"?

— edited by emilowk

@dan I think the way forward here is to have 3 categories of questions:

  1. Public ranked. These are questions submitted to the mods for use in the public forecasting tournament.
  2. Public unranked. These are questions that are public, but which mods don't interfere with, and the points are not counted towards the tournament rankings.
  3. Private unranked. Same as above, but only visible to self + specified users.