I predict 5% by July, 25% by October, 50% by December, 75% by March 2023, and 95% chance by July 2023.

There's a small chance that good Arc performance + an Ethereum crash would mean a quick resolution, and next gen-GPUs should be out by March. Among the 3 GPUs, the 3080Ti is the most likely to drop below $1100, since that price was created by Nvidia during the shortage to capture more of the price increases.

What does completely mean? Surely the government will still have access to the worldwide Internet even if the citenzry don't. Is North Korea "completely" disconnected from the global internet?

I'd forecast 1% for a complete disconnect but perhaps 35% that unauthorized ordinary citizens will be completely unable to access IPs outside Russia.

— edited by moreheadm

Funnily enough, my comment and perhaps other predictors seemed to spur new interest in the question, some of which share my optimism. My prediction roughly lines up with "twice as long as it's already been" heuristic. I predict a 25% chance of a solution in 25 years, 50% chance of a solution in 50 years, and 75% chance of a solution in 100 years. Part of the reason P vs NP is so difficult is because probably P != NP but we have few tools for proving lower bounds. A few mathematical breakthroughs and a solution might be close. Plus, CS research will be ...

What if less than two national space agencies exist in 2200 (but humans are not extinct)? For example, governments could be dissolved after anarchism/communism is implemented globally or nations could be wholly replaced by privately owned businesses or humanity could be unified under one government with only one space agency or the Earth could be a fief of another spacefaring civilization or (probably most likely) space exploration could be completely privatized.

— edited by moreheadm

As feedback for future contests, I'd like to see the questions available for a longer time before opening. It'd also be cool if Metaculus added the ability to make a prediction before the contest that activated as soon as the contest started.

@johnnycaffeine It's conceivable that there be large mining settlements on the moon that are nontheless not self-sufficient (because they primarily get their energy/food through shipments from Earth).

@kokotajlod Your question is about the ability of AI code generation to produce a deep learning program. My interpretation of the original question is that it intends to forecast the ability of AI code generation to produce a program producing program. In this case, the AI code generator would produce a program with similar function to itself. The extra level of indirection is necessary to capture that ability.

As suggested by the question, an AI system with that ability might be able to (or be close to being able to) improve itself.

The question is about 25,000 troops but the text says 100,000.

This is extremely unlikely now, with Russian troops confirmed to be (or soon to be) in Donbass. Ukraine won't implement Minsk II since Russia won't concede the border at this point. Things would have to change drastically for it to resolve positively, though an ambiguous resolution is possible depending how "annexation" is interpreted.

I'd expect Breyer to retire before 2023, assuming the Democrats win GA runoffs, which they are very likely to do at the time of commenting.

Here's net approval ratings after 180 days for the other presidents 538 has listed. - Trump: -16.9 - Obama: +17.6 - W. Bush: +19.5 - Clinton: -0.3 - H.W. Bush: +49.2 - Reagan: +28.1 - Carter: +40.1 - Ford: -3.0 - Nixon: +36.1 - Johnson: +65.1 - Kennedy: +58.0 - Eisenhower: +54.0 - Truman: +84.0 There is no president which had a net approval rating below +5 before 180 days which rose above +5 by day 180. Personally, I think it's unlikely that the apparent pattern of double digit net approval for the first 6 months will continue this year. I can't belie...

Can you make the range begin sooner? While it seems likely the shortage will exist into 2023, I think we should at least allow predictors to determine that.

@nhuvelle The market cap of Bitcoin is 0.1% of world wealth. Bitcoin's not going to grow in the next 10 years the same way it did in the past 10, simply because it limited by the total wealth of the world.

Say China and the U.S. fight a war against each other and then the U.S. captures Beijing (possibly before any of the "World War" criteria are met). Would the U.S. and China then be considered to be on the same side? I like the idea of defining China based on de facto control but I think the time "de facto control" is measured should be clearly stated (is it when the conflict starts or when one of the "World War" criteria are met). As a side note, I think the United States should be defined similarly, i.e. "de facto control of Washington D.C." *— edite...

@JackW By 2050, China may have completely independent competitive semiconductor manufacturing and TSMC might otherwise be in a less dominant position.

I ran a regression on the last 50 years of yearly temperature data. I then used that to project the temperature in 2022 and used that as the mean to calculate the inverse normal of 1.025 (with std.dev equal to the RMSE of the regression), which came out to 0.815, corresponding to a probability of 18.5%. I've lowered my prediction to 15% to account for the real science (rather than my primitive hackery) linked by @steven0461 and because I suspect there might be some small correlation with 2021's temperature.

— edited by moreheadm

@Matthew_Barnett I would be interested in the same question except in regards to the U.S. and India. India is interesting because they have a large economy, large population, and nuclear weapons. Also, their geopolitical alignment is perhaps less clear than that of western Europe. If you want to compose it that would be great, but I could if you can't, but maybe in a week or so.

@(kievalet) Do you think Donald Trump is unlikely to run or do you think he's unlikely to win the nomination if he does run? I doubt there's a scandal short of arrest and conviction that could really sink him and that would be incredibly risky political. And I don't see what an alternative candidate could offer the Republican base that would convince them to turn away from Trump. Obviously, Trump's approval ratings show that independents are inclined to distrust him, but the support from the core party is there. Say DeSantis runs and Donald Trump runs. ...
Does anyone have sources that shed light on the question of whether Ukraine is violating the ceasefire? Their official line (which Western media is not very critical of) seems to be that [they are not returning fire](https://abcnews.go.com/US/wireStory/putin-support-security-council-ukraine-83037165). That's extremely hard to believe for me, at least because it's hard to believe they wouldn't return fire after being fired upon. Plus, [the OSCE report](https://www.osce.org/files/2022-02-22%20Daily%20Report_ENG.pdf?itok=63057) seems to include some incid...