POLL: What is more important?

[xshat]

Chess is chess, and when the day comes that chess is solved, all chess engines will be clones of each other. So what does it matter if people take certain ideas from other engines or not; as long as it makes chess stronger then it's one step closer towards solving chess.

There hasn't yet been one author smart enough to create a completely unique engine that can never lose, so shouldn't all the best programmers just work together to achieve this end? Or is it more fun to watch the drama unfold and people bicker over who stole who's ideas and source code, and have chess development remain stagnant due to opposition. Chess is a competitive game, but does solving chess also have to be so competitive?

[Uly]

Chess is chess, and when the day comes that chess is solved, all chess engines will be clones of each other.

Not true, solving chess still leaves the problem of finding the best moves against an opponent that doesn't have access to the solution.

For instance, chess is a draw, so imagine that in the opening position 1.e4, 1.d4, 1.c4, 1.Nf3 and 1.Nc3 are optimal, which one do you play? Against an opponent without the solution the first four choices are better than the fifth, otherwise I can draw the solution by playing (as black):

1. Nc3 Nc6 2. Nf3 Nb8 3. Ng1 Nc6 4. Nf3 Nb8 5. Ng1 Nc6 {Draw by threefold repetition}

Because since a perfect game is a draw, playing that isn't different, right? Wrong! The task of the engine is now to lead the game into complex positions and hope that I blunder and go into a lost game. This isn't an easy job as the engine knows that all the drawing moves are always at my disposal, so it'd require opponent modeling (what works against one player doesn't against another).

Having chess solved would only give the engines a lot of more wide space for style, since a great percentage of the moves will draw, the engines will differ in what move they will play, and I won't be surprised if the engine's styles are more different than they are now, since it's possible an engine could play 1.f3 and beat most opponents without the solution.

But solving the game is so out of the horizon that it seems irrelevant now, it'll probably require a new approach and breakthrough that in retrospective the last years of computer chess would have no effect on finding the solution.

What matters is the present, and in the present engines' origins are important.

[xshat]

Chess is chess, and when the day comes that chess is solved, all chess engines will be clones of each other.

Not true, solving chess still leaves the problem of finding the best moves against an opponent that doesn't have access to the solution.

For instance, chess is a draw, so imagine that in the opening position 1.e4, 1.d4, 1.c4, 1.Nf3 and 1.Nc3 are optimal, which one do you play? Against an opponent without the solution the first four choices are better than the fifth, otherwise I can draw the solution by playing (as black):

...

But solving the game is so out of the horizon that it seems irrelevant now, it'll probably require a new approach and breakthrough that in retrospective the last years of computer chess would have no effect on finding the solution.

What matters is the present, and in the present engines' origins are important.

Solving chess would create a 32 man tablebases, in which every move would be known by the engine, hence they would all be clones since they would always make the perfect move. It may take a while but to say it is too far off may or may not be certain.

Do you provide any reason to say that engine origins are more important than engine development?

[Uly]

Solving chess would create a 32 man tablebases, in which every move would be known by the engine, hence they would all be clones since they would always make the perfect move.

You missed the point, at a given position there are several playable moves, there is no "THE" perfect moves, there are many that are perfect.

See for yourself, we already have 6men solved:

http://www.k4it.de/index.php?topic=egtb&lang=en

Go there and paste this FEN:

3rk3/3p4/8/8/8/8/3P4/3RK3 w - - 0 1

3rk3/3p4/8/8/8/8/3P4/3RK3 w - - 0 1

Now, assume this is the opening position, you'll see all possible moves are perfect and lead to the same result (draw).

So an engine may play d4, another Rc1, as they're optimal moves, this kills the "all engines will be clones in the end" argument.

You still haven't provided a reason for programmers to stop what they're doing to make an engine that never loses a game.

[Marek]

Solving chess would create a 32 man tablebases, in which every move would be known by the engine, hence they would all be clones since they would always make the perfect move.

No; for what is the "perfect move" in a position which has several drawing moves but no winning ones? You may have the 32-man tablebases, but what about your opponent? Let's assume the starting position is dead equal, with several drawing moves. Is it a good idea to play the Exchange Slav or the Exchange French? If your opponent has the 32-man tablebases then it makes no difference what is played - every game will be drawn (time and hardware permitting). But how do you know that he has them? Your own 32-man tablebases of themselves won't tell you how best to start against an opponent that doesn't have your tablebases but is otherwise a very strong player. So, for the fewest draws against imperfect opposition, the possessor of 32-man tablebases has to do some other analysis - and that gives scope to the programmers to produce different engines that will perform differently against imperfect opposition.

[xshat]

Solving chess would create a 32 man tablebases, in which every move would be known by the engine, hence they would all be clones since they would always make the perfect move.

You missed the point, at a given position there are several playable moves, there is no "THE" perfect moves, there are many that are perfect.

See for yourself, we already have 6men solved:

http://www.k4it.de/index.php?topic=egtb&lang=en

Go there and paste this FEN:

3rk3/3p4/8/8/8/8/3P4/3RK3 w - - 0 1

3rk3/3p4/8/8/8/8/3P4/3RK3 w - - 0 1

Now, assume this is the opening position, you'll see all possible moves are perfect and lead to the same result (draw).

So an engine may play d4, another Rc1, as they're optimal moves, this kills the "all engines will be clones in the end" argument.

You still haven't provided a reason for programmers to stop what they're doing to make an engine that never loses a game.

There is a perfect move if/when chess is solved. You're talking about the present, in which engines do not make perfect moves because chess has not been solved. Your conditions do not apply to what I'm talking about, therefore they do not "kill" any statements I made.

And I never told programmers to stop programming chess, you took what I said about chess evolving out of context.

[xshat]

Solving chess would create a 32 man tablebases, in which every move would be known by the engine, hence they would all be clones since they would always make the perfect move.

No; for what is the "perfect move" in a position which has several drawing moves but no winning ones? You may have the 32-man tablebases, but what about your opponent? Let's assume the starting position is dead equal, with several drawing moves. Is it a good idea to play the Exchange Slav or the Exchange French? If your opponent has the 32-man tablebases then it makes no difference what is played - every game will be drawn (time and hardware permitting). But how do you know that he has them? Your own 32-man tablebases of themselves won't tell you how best to start against an opponent that doesn't have your tablebases but is otherwise a very strong player. So, for the fewest draws against imperfect opposition, the possessor of 32-man tablebases has to do some other analysis - and that gives scope to the programmers to produce different engines that will perform differently against imperfect opposition.

If every game is always a draw, or always a win for white, then engines are going to make the same moves as that is what tablebases are meant for.

[Charles]

Why the heck would anyone want this:

Working towards creating stronger engines -- BY ANY MEANS NECESSARY?

This is absolute hogwash. So we should all steal commercial engines code put it on the web so we can ALL work create a stronger engine?

No way.. This is marxist gibberish!

There is a demand for stronger engines, we want stronger engines, and a few of us will work for it. However, we don't endorse lying, cheating , *killing* for it.

So Determining and defining which engines are clones, derivatives, or completely original is the only logical and moral choice

[Roger Brown]

SNIP

So Determining and defining which engines are clones, derivatives, or completely original is the only logical and moral choice

Hello Charles,

I respectfully submit that before we get to your point above - with which I agree - there has to be agreement as to what constitutes a clone, a derivative and a completely original chess engine.

Until then, this endless dance will continue.....

Later.

[jjh13]

Determining the individuals who were and are contributing to stronger chess engines is not very important in the bigger view of life and the universe. Creating the strongest chess engine possible (or "solving chess") is even less important. I do not see any reason NOT to give appreciation to the individuals who are actually contributed their time, money and efforts to the task.