mclane wrote: ↑Sat Mar 21, 2020 8:00 pm
Chess is NOT all about tactics.
In most positions there is absolutely no tactics.
Chess is nothing but tactics. Just because one position has a quiet move at the root, says nothing about what happens deeper into the tree. Ever won a game without either winning material or mating your opponent?
Positional play? This is what humans call a move when they are not sure it is the best move. It is based on intuition, experience, etc. Against a computer, it fails miserably because they actually calculate to find the best move, rather than using experience to guess.
Strategy? That is what you do when you can't calculate deeply enough to actually determine the best move. There would be no strategy if you could see to the endgame. IE do you have a good strategy for tic-tac-toe? Or can you simply calculate the entire tree and play the best move possible without intuition, positional play or strategy? Chess is the same thing.
bob wrote: ↑Sun Mar 22, 2020 2:16 am
Chess is nothing but tactics. Just because one position has a quiet move at the root, says nothing about what happens deeper into the tree. Ever won a game without either winning material or mating your opponent?
Positional play? This is what humans call a move when they are not sure it is the best move. It is based on intuition, experience, etc. Against a computer, it fails miserably because they actually calculate to find the best move, rather than using experience to guess.
I can't agree to that. You can match two engines and see very different kind of wins. If one engine's score suddenly jumped up while the other remained mostly the same, until later where the second realizes it's losing, it was a tactical win. If one engine's score slowly rose over a very long series of moves and the opposing engine agrees it's getting worse and worse until eventually the score gets so high that the game can't be saved, it's a positional win.
These are all human concepts and human words but something very different is happening here on those kinds of games.
What I don't get is what mclane wants? An engine that on the first case misses the tactical move and plays a worse move like a human? If search alone finds the win you don't even need eval.
Dann Corbit wrote: ↑Sun Mar 22, 2020 12:26 am
At some point, computers will see 100 plies ahead. No matter how you look at it, that's strategic.
It's not strategy or tactics. It's calculation.
Both strategy and tactics are calculation.
For computers, there is no strategy or tactics. It's just calculation. Strategy and tactics are meaningful concepts only for human chess thinking.
Humans have transported their human chess thinking into the programs.
The evaluation and search ideas are, in fact, human thinking. Strategic ideas like controlling the center, creating open files, making walls of pawns for defensive purposes, breaking walls of pawns for offensive purposes, retaining two bishops for a little extra punch, etc, are transferred into these chess programs from the minds of the programmers into the code. These chess programs even "know" that an edge pawn is worth less at the start of the game and more at the end. These chess programs "understand" complicated material imbalance. They know how to build Alekhine's gun. They attempt to undermine and go for deflection, and just about everything else a GM does.
Sure, it is just calculation, but that is also true when a GM does it,
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.
Let me also say that these chess programs play really beautiful chess. I remember Komodo's two queen sacrifice moves from TCEC that were stunning, and I have seen many more memorable and spine-tingling moves made by chess programs.
I think that not only are chess programs able to play chess better than humans, they can also equal the beauty of human chess. Now, I admit that beauty is in the eye of the beholder and what I consider beautiful another person might not. But programs like Stockfish and LC0 and Komodo can play chess that is actually inspiring (for me)
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.
Dann Corbit wrote: ↑Sun Mar 22, 2020 12:26 am
At some point, computers will see 100 plies ahead. No matter how you look at it, that's strategic.
It's not strategy or tactics. It's calculation.
Both strategy and tactics are calculation.
For computers, there is no strategy or tactics. It's just calculation. Strategy and tactics are meaningful concepts only for human chess thinking.
Then, what do you call Lc0 playing with 1-playout being very strong ? There is no calculation, atleast in the sense of lookahed search, and is solely due to strategic concepts (not tactics) that it is able to do so. Planning involves a lookahed search, so MCTS, alpha-beta are the planning methods.
Lc0 without planning (i.e. 1-playout) is playing without knowing the tactics ( the actual path taken ) to reach the desired outcome.
So strategy, tactics, planning are not "fuzzy" concepts in AI rather are well-defined terms. Having said that, I find the OP's claim that there is no engine that plans quite laughable, when there are tons of alpha-beta and NN engines that do exactly just that ...
jp wrote: ↑Sun Mar 22, 2020 9:56 am
...
For computers, there is no strategy or tactics. It's just calculation. Strategy and tactics are meaningful concepts only for human chess thinking.
I agree. Moreover what is tactics play, what is strategical play is only the question of definitions.
Look ahead with millions of NPS is not planning for me.
It is also not knowledge for me.
A plan needs no search depth of 30-40.
It needs no AB search at all.
A plan is a concept.
Where to manoeuvre the pieces and WHY to do so.
A plan is a thesis.
And the search tree could find out if the thesis Works.
But it can be build without proof that it works or not.
Chess engines have plenty of evaluation categories and give malus and bonus and sum up them into a value.
But this is not a plan.
This is knowledge and the knowledge can be put into consideration of the position.
A plan can be a mate attack or misusing a positional weakness. It can be important to bring a piece into a certain position or to build a certain pawn structure or destroy it.
A plan is NOT the main line the AB engine has found out.
Of course this CAN be a method to do it and it is a method that is used in most AB programs.
I do believe that programs like LC0 can play a better strategical chess even if they use very few resources then the usual AB engines.
This is because normal AB programs rely mainly on many NPS and huge search depth. Something that is IMO not important for the thing I am talking about.
Normal AB chess Engines misuse resources in a very Heavy way.
It’s been done because these resources are there and we have plenty of them. But they lead to a loss of planning because the programmers rely on search instead on creating a plan. The search replaces the plan and substitutes it.
IMO it would make more sense to stop this waste of resources and substitute search tree with planning.
What seems like a fairy tale today may be reality tomorrow.
Here we have a fairy tale of the day after tomorrow....
jp wrote: ↑Sun Mar 22, 2020 9:56 am
...
For computers, there is no strategy or tactics. It's just calculation. Strategy and tactics are meaningful concepts only for human chess thinking.
I agree. Moreover what is tactics play, what is strategical play is only the question of definitions.
In my mind, tactics are search/lookahead, and strategy/planning is a shortcut that humans and computers combine with limited lookahead because we are just not capable of calculating the tactics of chess through to the end from the starting position. Human brains are hard wired to find patterns, so planning "feels" more like a natural solution to a lot of folk. Ultimately, the only perfect chess we have so far is tablebases, which are calculated by search. Solved games are solved because we were able to look ahead far enough to solve them, and IMO planning and patterns are not going to be enough to solve the game.
Another important thing to keep in mind is that humans do quite a lot of computations themselves; it's just that we're consciously aware of very few of them.
With engines, we're aware of all that they have to do to play high-level chess, which can make it seem like a more drastic difference than it really is.
I'm of course not saying there is no difference in the type or scale of computations between human players and engines; there certainly are differences.
It's just that the "Humans are consciously aware of very few computations while playing chess, therefore the human way of doing chess must be very special compared to those computation-heavy bean counters." inference is perhaps not so convincing.
