How fast are the chess engines?

Originally posted by cashthetrash
And of course the games are recorded on RHP in your game history, both finished and ongoing. And that is a database correct. So in effect it is very easy to copy the moves. So in theory couldn't you go an entire game? And if an opponent was using an engine unknown to you, and you coppied their moves, they would be cheating and you wouldn't? That seems l ...[text shortened]... se between database and engine use. I want to make sure I positivly understand this correctly.

And how would two people repeat a single game? They would both need to be following that specific game without deviation from begining to end. And why would they do that? If the original game ends in a draw, why bother playing on, just agree the draw. If either white or black won, why would the other side follow the game? It really doesn't make any sense.

Use a database to explore opening theory. Beyond that, play your own game. That way you run no danger of a high correlation to engine play.

Originally posted by Gatecrasher
And how would two people repeat a single game? They would both need to be following that specific game without deviation from begining to end. And why would they do that? If the original game ends in a draw, why bother playing on, just agree the draw. If either white or black won, why would the other side follow the game? It really doesn't make any s ...[text shortened]... yond that, play your own game. That way you run no danger of a high correlation to engine play.

I would play for a draw and see if I get a chance to win. Why leave a game if I cannot lose? I have my own style of play what ever that is. I do not like following a game if my side lost it. In following games I like to follow the top players like Fischer. It does get me in trouble at such a time that the game does not follow the game I am following. I do not know what the player's plan was. One time a player told me that I missed a trap that most players get caught in. I told him that so far he is playing Kasparov.

Originally posted by Gatecrasher
And how would two people repeat a single game? They would both need to be following that specific game without deviation from begining to end. And why would they do that? If the original game ends in a draw, why bother playing on, just agree the draw. If either white or black won, why would the other side follow the game? It really doesn't make any s ...[text shortened]... yond that, play your own game. That way you run no danger of a high correlation to engine play.

Two might not, but one can. And sometimes people do things for experimental purposes only to test prove/disprove a theory with no evil intent intended. Questions without answers can be maddening at times. Perhaps I should pm you a message with more details of my theory. You may already know the answer.

Originally posted by General Putzer
fritz, shredder, etc., need only 10 seconds a move to crush anybody below expert level. Even 5 would be plenty.

My Fritz8 at 15 seconds a move and at the very low computing power of 80 MB, found every one of Fischer's critical moves in his "Game of the Century" against Robert Byrne. You can't outplay the top engines in tactics, period.

Originally posted by cashthetrash
Two might not, but one can. And sometimes people do things for experimental purposes only to test prove/disprove a theory with no evil intent intended. Questions without answers can be maddening at times. Perhaps I should pm you a message with more details of my theory. You may already know the answer.

Thanks for your PM. Looking back at your posts, I now understand where you were heading... The situation you described has happened before at RHP and it is NOT legal.

One distinction that should be made: A database consists of completed games, not games in progress. And certainly not games you are currently playing. Specifically, you cannot take the moves of your (stronger) opponent in one active game, to play against another (stronger) opponent in another active game, such that your two opponents are effectively playing against each other. That contravenes the "third party assitance" clause. The motive: If your two opponents are stronger than you, you will gain rating points regardless of the outcome. That is cheating. Pure and simple.

Originally posted by Gatecrasher
Thanks for your PM. Looking back at your posts, I now understand where you were heading... The situation you described has happened before at RHP and it is NOT legal.

One distinction that should be made: A database consists of completed games, not games in progress. And certainly not games you are currently playing. Specifically, you cannot take the m ...[text shortened]... you, you will gain rating points regardless of the outcome. That is cheating. Pure and simple.

Thanks for the clarification. I am glad you explained it as you did. I wasn't aware of this happening before and I thought it might be wise to bring it to your attention. Your earlier statment had me confused. Thanks.

Originally posted by exigentsky
Very true for OTB. However, on RHP since both players can use databases to aid them in the opening, I would say the chance of going past 10 moves of theory is actually quite high.

In just a few moves, we are talking about literally millions, billions, and numbers exponentially higher than that, number of possible games. If you don't understand this, lets for sake of argument, assume that there are only 3 moves that guarantee a strong game with 3 replies each. Type 3*3 in a calculator 20 times to see how many possible games there are after 10 moves.

It is possible that two players follow the same theory up in to the tens and even twenty or thirty moves, but it means that both players are literally following an earlier played game that should be theoretically even for both sides, and in practice *at least* one side should want a win. As soon as one player decides to do just one move differently, its a completely new game and the previous theory gets re-written.

The idea that someone could generate a database of opening moves to be large enough to practically guarantee a player a win is laughable. If that was the case, why wouldn't the makers of fritz, shredder, et al. just generate the database themselves? And even in some other dimension of time when we can generate this kind of database, just imagine trying to read it with human eyes. It would take you years just to follow one variation.

Originally posted by bosintang
In just a few moves, we are talking about literally millions, billions, and numbers exponentially higher than that, number of possible games. If you don't understand this, lets for sake of argument, assume that there are only 3 moves that guarantee a strong game with 3 replies each. Type 3*3 in a calculator 20 times to see how many possible games there a gine trying to read it with human eyes. It would take you years just to follow one variation.

Exigentsky is right. Anyone who plays the main lines in openings like the Ruy Lopez, King's Indian, Gruenfeld, Sicilian Dragon, etc, will know that opening theory usually lasts for at least 15 to 20 moves. Beyond that, anyone who has a good sized database should be able to find games to follow further, if they choose.

I had a look at the games that I'm playing at the moment. In one game I've reached the 19th move and both sides are still in opening theory. It might not be for another 10 or so moves before one of us deviates.

In fact, strong OTB players who play those sort of openings will usually memorise at least 15 moves of opening theory. A Grandmaster will memorise much further.

Two strong correspondence players (who have access to plenty of opening books and massive databases) probably won't be looking for opening novelties before they reach move 20 at the earliest.

Originally posted by Gatecrasher
And how would two people repeat a single game? They would both need to be following that specific game without deviation from begining to end. And why would they do that? If the original game ends in a draw, why bother playing on, just agree the draw. If either white or black won, why would the other side follow the game? It really doesn't make any s ...[text shortened]... yond that, play your own game. That way you run no danger of a high correlation to engine play.

I did that once, mainly due to incompetence - Game 1424399 both of us were following the theory, and I think both of us were hoping the other would deviate first, my opponent had won the other game and didn't need more than a draw anyway. I'm not the first, this game occurs 4 times on my database. It got to the repetition at the end and I really didn't have anything better...

I'm experimenting with an opening to use, primarily OTB, but maybe here as well. I intend to use an engine to check/generate lines and construct some traps. Now, on the one hand this is perfectly legal, as there is no game in progress, essentially I'm gambling that it'll come up. On the other hand it is going to be fairly hard to prove that I did the analysis before the game started if I use it here. The odds are that I'll never get round to doing the work, and I can leave it for OTB only use, but that would be a bit of a shame, any comments?

Originally posted by DeepThought
I'm experimenting with an opening to use, primarily OTB, but maybe here as well. I intend to use an engine to check/generate lines and construct some traps. Now, on the one hand this is perfectly legal, as there is no game in progress, essentially I'm gambling that it'll come up. On the other hand it is going to be fairly hard to prove that I did the ...[text shortened]... work, and I can leave it for OTB only use, but that would be a bit of a shame, any comments?

Your research is fine. You might be lucky and find a few useful moves for your opening. When we check for engine use we exclude established opening lines, and check for engine match-ups over many games. No way on Earth you could do engine "research", move after move, game after game.

Originally posted by David Tebb
Anyone who plays the main lines in openings like the Ruy Lopez, King's Indian, Gruenfeld, Sicilian Dragon, etc, will know that opening theory usually lasts for at least 15 to 20 moves. Beyond that, anyone who has a good sized database should be able to find games to follow further, if they choose.

In fact, strong OTB players who play those sort of op ...[text shortened]... es) probably won't be looking for opening novelties before they reach move 20 at the earliest.

I replied to existengsky, but I really should've replied to the person who suggested generating a massive database of opening moves.

You're right, of course, that strong players can follow an opening line 15-20 moves deep, or go even deeper if they so choose, but they're doing just that, following a main line. They follow those lines because they give the best theoretical advantage, but if a player (wisely or not) chooses to play a novelty earlier, then both players still have a chess game to play, and they have to rely on their positional and tactical intuition rather than database theory and making automatic moves.

Or am I really underestimating top GMs here? Can top GMS really see all the way to mate if a player makes a novelty in the first twenty moves?

Originally posted by bosintang

Or am I really underestimating top GMs here? Can top GMS really see all the way to mate if a player makes a novelty in the first twenty moves?

Unless it's an exceptional position, nobody can see that far ahead.

What I meant to say is that it's very hard to find good novetlies in the early stages of main line, theoretical openings.

So if two Grandmasters were playing each other in a correspondence game, I doubt that they'd waste their energies looking for novelties before the twentieth move or so. If one of them does deviate from theory, the other player is very likely to benefit, either by immediately gaining equality (usually when playing Black) or with an increased advantage (usually White).

For instance in the very popular, main line Zaitsev Variation of the Ruy Lopez, hundreds, if not thousands, of games have been played in several ultra sharp lines. Any opening book that covers this opening will focus on alternatives at various critical points from about move 25 onwards.

Opening theoreticians have been analysing these lines for years, so it's extremely unlikely that a casual player will come up with a good move that hasn't already been played or hasn't already been analysed and dismissed by the theoreticians as inferior.

Originally posted by Gatecrasher
In practice, even running your engines for several hundred years, you may only be able to extract a few extra opening moves beyond the current large databases.

Going along those same lines...

There are 170 septillion ways to play the first 10 opening moves in chess. That's 170,000,000,000,000,000,000,000,000. That's a lot of positions for a chess engine to analyze. And that's just the first 10; it increases exponentially with each following move.

Originally posted by David Tebb

So if two Grandmasters were playing each other in a correspondence game, I doubt that they'd waste their energies looking for novelties before the twentieth move or so. If one of them does deviate from theory, the other player is very likely to benefit, either by immediately gaining equality (usually when playing Black) or with an increased advantage (usually White).

This what I was basically trying to say. An inferior (but not exceptionally so) move allows black to equalise or gives white a slightly stronger advantage, but that doesn't mean anything if the player behind the pieces does not understand why the other players move is inferior, and does not know how to capitalise on it.

Bringing this post back to the post I was responding to (not existengsky, but someone previously), that pondered if one could build a database to match the play of a grandmaster or a computer. Even if I, a 1500-rated player, had a ridiculously massive database of opening lines, its still very, very highly unlikely that I would beat you, a 2300-rated player, even if you did make a so-called "inferior" move within the first twenty moves. My database can help build me a strong opening position, but only in some mathematically improbable fantasy can it win my games for me.

Originally posted by bosintang
This what I was basically trying to say. An inferior (but not exceptionally so) move allows black to equalise or gives white a slightly stronger advantage, but that doesn't mean anything if the player behind the pieces does not understand why the other players move is inferior, and does not know how to capitalise on it.

Bringing this post back to the ...[text shortened]... ning position, but only in some mathematically improbable fantasy can it win my games for me.

Actually I have two games I consider won by database alone. Of course that's nothing out of 733 games played but it does happen every once in a while.

Game 1456758 is one and right now I can't find the other one in my archive.