Curious how varying time would effect confidence

I wonder if @jonsimon, @anon and the other people who think this is near zero also think that /BoltonBailey/chess-with-queen-odds-solved-by-204 should be near zero. Is it a question of good heuristics being present, or is the game tree complexity the only thing that matters?

@BoltonBailey id best against us finding any structure that makes brute force unnecessary and also against brute force being available in the insanely short window of 16 years. I don't know about chess with queen odds, depends on state space size and if it seems like there could be some structure to it.

@anon What chance do you think there is of AGI before 2040? Most of the worlds in which this resolves yes IMO involve AGI

@dominic AGI means a system with human-or-better capabilities in a wide range of fields. It doesn't mean "solves all NP hard problems".

@jonsimon Definitely wouldn't solve all NP-hard problems. But solving chess to me seems in the ballpark of things that are hard for humans but a superintelligent AI would help a lot with.

@dominic general to human intelligence maybe dyson swarm or really effective fusion almost certainly not

@anon actually maybe fusion but I doubt we'll control enough energy transistors to do it by 2040 in any case even w fusion being somewhat viable before then

@anon *energy or transistors

/MartinRandall/when-will-7x7-go-be-solved

Pretty sure this should be easier than Chess.

Truly wild that this is over 1 percent

Checkers and Connect4 weren't ultra-weakly solved before they were weakly solved

@JonathanRay yes, because crafting a uw solution is more unconventional than brute force approaches. It requires a large amount of formal construction that humans would not waste their time on.

My bet is that that proofing engines are going to get much better at solving these large spacial problem in order to solve more important problem cases seen in physics/medicine.

A byproduct of these advances would be uncovering a method of solving chess UWly.

The frustrating thing about markets like this is you're mostly trying to predict what people think "solved" means, rather than about whether it will actually happen. Probability of it actually happening is near zero.

But I'm not going to lock up my mana for 15 years assuming that people will realize that.

@jonsimon the meaning of "solved" is extremely well defined in this case?

curious if anil has object-level opinions here or not

@jacksonpolack imo an ultra-weak solution is more forgiving than you'd expect.

@jacksonpolack My belief is that an ultra-weak solution will be crafted by a proof engine instead of a chess engine.

Do you think AI will be pivotal here or no?

Its essential.
Could be a mix of something like DeepMind's AlphaGeometry and ATP (Automated Theorem Proving), though every 2/4 years or so there are new tactics/discoveries that change the ecosystem up, so who know.

This is wild

@Joshua why are we doing this lol

I am committed to ensuring accurate pricing and have the balance to back that up.

@jacksonpolack I just want to beat NFL after they tried to screw me over by making the market I made most of my month’s profit in unranked this morning

gotcha

@june + I think this is 40% odds and has only been held down because of Jonathan Ray’s willingness to commit more to a 2040 market than the rest of Manifold is, but if not for the NFL thing I wouldn’t be buying this much

well you're wrong about that

@june yeah you're wrong about that. I think this is very unlikely to happen.