I don't understand why you believe Lc0 (which is a fascinating chess-playing entity) is using "strategy". It's doing something like image processing on chess diagrams, and "discovering" what "patterns" lead to a higher probability of winning. Is that strategizing or planning?mclane wrote: ↑Mon Mar 23, 2020 9:43 pm I do not disagree. You are right. The AI science moved away. But I still think chess engines are stupid. Although they have 3400 ELO they are not playing chess but using brute force search to „solve“ it.
LC0 is the evidence for me that it is possible to build strategy over tactics and still play decent human chess.
It’s IMO a shame that a NN has to teach humans that human engines are stupid.
LC0 showed that Stockfish is an idiot.
When will the chess programmers write an engine that plans ?
Moderators: hgm, Rebel, chrisw
-
- Posts: 6442
- Joined: Tue Jan 09, 2007 12:31 am
- Location: PA USA
- Full name: Louis Zulli
Re: When will the chess programmers write an engine that plans ?
-
- Posts: 4556
- Joined: Tue Jul 03, 2007 4:30 am
Re: When will the chess programmers write an engine that plans ?
Suppose that someone manages to create software that "plans" in the way you want, and it manages to play the same moves of Magnus Carslen. Then what? We already had access to his moves, we wouldn't have anything that we don't have.
The world is full of machines that are capable of imitating human plans and human chess play perfectly, those machines are called "humans", creating an artificial one would be a waste of time, just ask someone on the streets to make a chess move.
What would be useful is creating something that plays chess on a never seen before level that defeats everything else seen before, and for that Stockfish and Leela are on the right path.
-
- Posts: 2526
- Joined: Mon Feb 08, 2016 12:43 am
- Full name: Brendan J Norman
Re: When will the chess programmers write an engine that plans ?
Look Uly, it's very simple:Ovyron wrote: ↑Mon Mar 23, 2020 10:22 pmI never said improved that much, I just said that I improved some hundreds of elo points, and now can reach winning positions against 2300 guys now and then (just yesterday I reached a winning position against a 2100 guy WITH EASE on a 5 0 game, and then I proceeded to lose it with some tactic that I missed...)BrendanJNorman wrote: ↑Mon Mar 23, 2020 11:21 am If you're saying you gained 800 Elo in one year, yet you aren't willing to prove it, your story is paper-thin bro.
Regardless of how many points, you are saying that after having a peak rating of 1600 or so for more than a decade, you have now jumped hundreds of Elo points in a single year and (on your mysterious NEW liChess account) can now EASILY handle 2100-2300 guys.
I'm willing to believe it, but the fact that you won't even share your liChess account so as to put skepticism to rest is pretty damning.
I'll even go first...here you are: https://lichess.org/@/ChessNCognac_com/perf/blitz
That's my blitz profile.
Where is yours?
I'll be impressed (legitimately) even if you even got up to 1800, but the way you're behaving is impossible to believe.
I'm pretty sure anybody in my position would be the same.
-
- Posts: 381
- Joined: Thu Nov 22, 2012 12:20 am
- Location: New Hampshire
Re: When will the chess programmers write an engine that plans ?
How do Lc0 and Stockfish differ in regards to solving mate problems?
Say there is a mate in 10 puzzle, is Lc0 really going to strategize or plan?
Say there is a mate in 10 puzzle, is Lc0 really going to strategize or plan?
-
- Posts: 12541
- Joined: Wed Mar 08, 2006 8:57 pm
- Location: Redmond, WA USA
Re: When will the chess programmers write an engine that plans ?
Real artificial intelligence will take a lot more horsepower even than our GPUs.
https://www.quora.com/If-the-human-brai ... mputer-Why
"If the human brain were a computer, it could perform 38 thousand trillion operations per second. The world’s most powerful supercomputer, BlueGene, can manage only .002% of that." But, we cannot perform like a supercomputer. Why?"
So consider:
38,000,000,000,000,000,000 is 38 exaflops. The world's most powerful computer is .2 exaflops.
So if you want actual understanding (e.g. Deeper Blue has literally no idea that it won) you will have a very long wait.
If you just want really good chess, we have that today. And like Inigo Montoya, "I hate waiting."
https://www.quora.com/If-the-human-brai ... mputer-Why
"If the human brain were a computer, it could perform 38 thousand trillion operations per second. The world’s most powerful supercomputer, BlueGene, can manage only .002% of that." But, we cannot perform like a supercomputer. Why?"
So consider:
38,000,000,000,000,000,000 is 38 exaflops. The world's most powerful computer is .2 exaflops.
So if you want actual understanding (e.g. Deeper Blue has literally no idea that it won) you will have a very long wait.
If you just want really good chess, we have that today. And like Inigo Montoya, "I hate waiting."
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
-
- Posts: 4556
- Joined: Tue Jul 03, 2007 4:30 am
Re: When will the chess programmers write an engine that plans ?
What if it was on rapid? Suddenly only blitz ratings matter?BrendanJNorman wrote: ↑Mon Mar 23, 2020 10:52 pm I'll be impressed (legitimately) even if you even got up to 1800, but the way you're behaving is impossible to believe.
-
- Posts: 2526
- Joined: Mon Feb 08, 2016 12:43 am
- Full name: Brendan J Norman
Re: When will the chess programmers write an engine that plans ?
Ovyron wrote: ↑Mon Mar 23, 2020 11:01 pmWhat if it was on rapid? Suddenly only blitz ratings matter?BrendanJNorman wrote: ↑Mon Mar 23, 2020 10:52 pm I'll be impressed (legitimately) even if you even got up to 1800, but the way you're behaving is impossible to believe.
-
- Posts: 1470
- Joined: Mon Apr 23, 2018 7:54 am
Re: When will the chess programmers write an engine that plans ?
Yes. If SF is an idiot, then Leela is also an idiot.zullil wrote: ↑Mon Mar 23, 2020 10:32 pmI don't understand why you believe Lc0 (which is a fascinating chess-playing entity) is using "strategy". It's doing something like image processing on chess diagrams, and "discovering" what "patterns" lead to a higher probability of winning. Is that strategizing or planning?mclane wrote: ↑Mon Mar 23, 2020 9:43 pm I do not disagree. You are right. The AI science moved away. But I still think chess engines are stupid. Although they have 3400 ELO they are not playing chess but using brute force search to „solve“ it.
LC0 is the evidence for me that it is possible to build strategy over tactics and still play decent human chess.
It’s IMO a shame that a NN has to teach humans that human engines are stupid.
LC0 showed that Stockfish is an idiot.
mclane said he wants no computation, and Leela is doing massive computation.
But clearly mclane does not just want really good chess. He wants something else, or something more.Dann Corbit wrote: ↑Mon Mar 23, 2020 11:00 pm If you just want really good chess, we have that today.
-
- Posts: 2526
- Joined: Mon Feb 08, 2016 12:43 am
- Full name: Brendan J Norman
-
- Posts: 6442
- Joined: Tue Jan 09, 2007 12:31 am
- Location: PA USA
- Full name: Louis Zulli
Re: When will the chess programmers write an engine that plans ?
If an engine were sentient, and wanted to play chess well, it would likely install Stockfish and Lc0.BrendanJNorman wrote: ↑Mon Mar 23, 2020 11:29 pmIt sounds like he wants the engine to be sentient, which as Dan said, will not be possible for a LOOOOONG time.