@WhySpace Your idea of looking at Wikipedia's revision history is great! This turned out very useful for me. It looks hard to build an accurate parametric model, especially because there's an important selection effect which limits what we can draw from the data: we expect the proportion of people in the data to increase with age; i.e., a higher percentage of worldwide 116-year olds are expected to be in the data than that of 111-year olds. The oldest people are more likely to be reported, but the selection effect persists even otherwise! Consider the 1...
As of today (2018-12-24), the DJIA closed at 21792.2, so the question resolves positive if it rises by 14.7% before it drops by 8.2%, relative to today's value. Historical closing values since the DJIA's launch in 1896 are [officially available](https://us.spindices.com/indices/equity/dow-jones-industrial-average) ([direct link to `.xls`](https://us.spindices.com/idsexport/file.xls?selectedModule=validationReport&p_index_id_arr=1720081&p_frequency=D&downloadFlag=Y&selectedReportId=2000&resultsFileTemplate=0&resultsFileType=103&fromSession=N&force_downlo...
@Anthony @erik.bjareholt Okay, sounds good. \--- Not directly related to the question, but let me share the following, which touches on a few aspects of the cryptocurrency markets. ####Recent cryptocurrency distributions. Two cryptocurrencies named Byteball (unit: GB; not to be confused with Bytecoin) and Stellar Lumens (unit: XLM) are both recently distributing some of their coins to Bitcoin (unit: BTC) holders based on snapshots of the Bitcoin blockchain at pre-specified times (action is needed to receive the coins). Byteball just held their 6th r...

Does it count if the individual 'wakes up' as a digitally emulated brain, with the original brain and body discarded? I think this falls within the spirit of the question.

Some disjointed comments: I really like this question and am anticipating the outcome. The RECAP case docket for the Elsevier v. Sci-Hub lawsuit can be [accessed on the Internet Archive](https://ia600408.us.archive.org/35/items/gov.uscourts.nysd.442951/gov.uscourts.nysd.442951.docket.html), among other places. I enjoy reading court transcripts and here there are two transcripts of phone conferences with Alexandra Elbakyan (from 2015). Looks like Elsevier will gain default judgment because the defendants haven't appeared. There's also the ACS lawsuit men...
The Bitcoin fork took place with seemingly little fuss, giving us the two descendant chains "Bitcoin" (BTC) and "Bitcoin Cash" (BCH). Bitcoin is expected to have another hard fork in November as prescribed by the [SegWit2x agreement](https://medium.com/@DCGco/bitcoin-scaling-agreement-at-consensus-2017-133521fe9a77), and it looks like that will be another contentious fork. As of writing, Bitcoin Cash has a minority of the hash rate and is trading at 0.15–0.20 BTC per BCH. I want to comment on the hash rate distribution. Bitcoin and Bitcoin Cash currentl...
@Anthony Wow, no kidding… In a [recent interview](https://www.ooom.com/digital/sergio-canavero-a-revolution-in-medicine/2/): > [Canavero]: […]I have good news for them [[Alcor](https://en.wikipedia.org/wiki/Alcor_Life_Extension_Foundation)]. > [Interviewer]: Namely? > [Canavero]: We will try to bring the first of the company’s patients back to life, not in 100 years. As soon as the first human head transplant has taken place, i.e., no later than in 2018, we will be able to attempt to reawaken the first frozen head. Oh dear, I'm pretty sure that's not...
@John Thanks for the links! There was a [recent paper](https://dl.dropboxusercontent.com/u/238511/papers/2016_solomon.pdf) ([supplement here](https://dl.dropboxusercontent.com/u/238511/papers/2016_solomon_2.pdf)) in Science by Solomon et al. that summarizes a model fit to past data (they didn't give projections), from which they infer that the ozone hole is recovering. Here's a plot of the data (note that 1995 is absent from the data!), with the 5-year average plotted as the yellow line: [figure 1](https://v-yu.com/media/metaculus/q347/plot1.png). Solo...
I have two suggestions for the resolution criteria: 1. At every point in time, for resolution, use the descendant blockchains of bitcoin and ethereum with the largest market cap. Rationale: Bitcoin may soon have a contentious chain fork, nominally due to disagreement over the solution to its scaling problem. Ethereum has already had a contentious chain fork in 2016 in the aftermath of the hack of [The DAO](https://en.wikipedia.org/wiki/The_DAO_(organization)), leading to "ethereum" and "ethereum classic". (There are [low-volume futures markets](htt...
@WhySpace, James, Anthony: It seems hard to change the current scoring system to incorporate the ideas under discussion. I suggest using other social incentives to achieve these ideals. I notice that upvoting has just been implemented, which IMO is great for encouraging users to post relevant news that they find. (I was able to upvote one of my own comments. Maybe modify the system to remove/disallow self-upvotes?) Perhaps we can add further features to the commenting system to make it a more prominent and easy-to-use social feature, which I hope will e...

I can say it's mostly not my fault since I've been MIA for a while. ;)

… but I probably would have made things even worse if I submitted predictions.

Looks like this resolves negative by a large margin. They modified the scoring so that the total score is now out of 40 instead of the former 120. They used to score on three criteria: relevance, correctness, and grammar. This year they scored on "how human their responses are", and the champion bot (Mitsuku, who is now a 3-consecutive-year champion) scored only 27/40 = 67.5%, compared to last year's 90%.

I'd like to see a reverse test where humans are scored on "how robotic their responses are"!

@(zarock8y) Based on the resolution text, I think the question should resolve positive as long as Sci-Hub is accessible on August 1, 2018, including the case where it is accessible only as a Tor hidden service. \--- I'd actually like to see Sci-Hub shut down their online domain and keep only their Tor hidden service, so that it remains freely accessible to anyone who needs it (i.e., almost everyone in the world, since only a tiny minority has non-exorbitant access), but those of us who use it purely as a convenience over using our institutional access ...

With 2 citations so far, this is equivocal (36%) under a Poisson model (that assumes a constant rate) after updating from a pretty uninformative prior for the rate parameter. Here's a plot, showing a wide predicted distribution.

Edit 2017-05-01: Switch away from Dropbox links.

— edited by v

People with clocks set behind UTC should see $1167 million, and those at or ahead of UTC should see $926 million. I think the $926 million figure is the appropriate one since UTC should be considered the default timezone, and that figure is the data point corresponding to 12am UTC on 2017-11-15. (Download the CSV for the historical data.) At the current time of 2017-11-20 2am, the last date in the data is 2017-11-18, UTC, so I think the volume they give for 2017-11-15 12am corresponds to the volume on the 15th instead of the 14th, so using that volume ...
@Anthony re: FAQ. That's reasonable; I take back my suggestion for a change. If I understand correctly: 1. Future questions should try to specify a resolution based on reports (i.e., time of first credible and publicly-available knowledge, even if obscure), not acts. 2. But even when the resolution is based on reports (as above), if the act has already occurred without reports before question opening, then the question resolves ambiguous. Are these correct? Re: resolution for this question. To be pedantic, if we resolve based on the September 15 launc...

This is not exactly related to the question, but the operator of one of the Ponzi schemes on Eve Online in 2006 (scams involving the in-game currency are allowed) wrote a first-person account that I really enjoyed reading.

@Englisc That's true. In fact, it's a common theme among my predictions that I rely on base rates*, historical data, and numerical data, and discount other things like current events and news. I think there is great value in these other considerations! It's just that I am generally much less confident in my judgments on them.

* "Base rate" sounds objective, but I usually feel that the choice of reference class leaves a lot of room for subjective judgment of the sort that I feel more confident about.

With the exception of Lee Sedol defeating AlphaGo in one game, the Brexit is so far the most unexpected result on Metaculus in the sense that every other resolved question with (median prediction <= 32%) or (68% <= median prediction) has resolved in the predicted direction.