Computers can find cooks and solve the most intricate chess problems but can programs compose them? Even though humans cannot defeat computers anymore, I hope a place remains for chess composers.
I’ll say computers will never replace composers. Especially for expert (and higher) level problems. For one thing, as insane as most of those problems are, the key move is usually very subtle to a human. Not really sure a computer could ever really put “art” into the artfulness it takes to compose a really good problem.
I think when it comes specifically to endgames though, I think even with table bases, it would take a human to be able actually pick out positions that have a meaningful reason to be solved.
Computers can calculate out an endgame, but it takes the deft touch of a human to decide exactly what position to start at, in which a human would find reasonable joy in solving.
No, I’ve never actually solved an expert level problem yet. I don’t even try anymore. I might pick it up again sometime though. In any event, expert and harder level problems do have an insanity to them.
Buried in there is the time controls for solving various problems, specific to
World (European) Championship anyway. -I guess the world championship is always in Europe.
Have to solve 3 different 2 movers in 20 minutes
Have to solve 3 different 3 movers in 60 minutes
Plus other problems like 4 and 5 move mates, help mates, self mates, and endgame problems.
You can teach computers to research and discover whether a problem’s main idea has occurred in other published studies. You likely can get them to combine ideas in intricate ways that require a large amount of time and energy to solve. What problemists seek as a jewel or unblemished pearl are brand new ideas that have never been done before. How do you write an algorithm to do that?
I’m not really sure a computer could really “combine ideas in intricate ways”. Since programs can go through every possible combination for several ply in mere moments, how can someone create a program that can decide what kind of positions can give a human brain enjoyment for solving it?
Although I haven’t really solved any difficult expert level compositions, when you get to easier compositions, you end up with “themes” for most of the problems. There are only so many matting patterns available before you end up regurgitating the same mating ideas in different positions. -It’s excellent material for increasing one’s ability when playing games, no doubt. But the really hard compositions are more art than science.
When I say more art than science, I mean I stare at a composition, and spend several minutes trying to figure out that no matter what move I find, the opponent still seems to be able to finagle it’s way out of the mate. The subtle move needed to achieve a mate can be maddeningly difficult to find. I really think that sort of artfulness can’t be created with a chess program.
For those with an Android phone, check out “Chess Puzzles and tactics” by Kruber. It’s a pretty simple program, but they managed to pack over 130,000 expert 2 move problems.
-Arguable not all the “expert” problems are anywhere near expert level, a LOT of them are.
I looked up a few of the expert problems, and all could be found as human made compositions, some dating back to the late 1800’s. Too bad the program doesn’t have more features.
Most/all the non expert level problems run into various mating patterns that is good for chess players to learn.
The program is fairly simple though. You get random problems and not much else. Still, considering the sheer amount of problems, it’s still a worthy program.
Can someone check to see if it’s available for iPhones?
Computers already generate Sudoku and Kakuro puzzles. Virtually every Sudoku you see in any newspaper is computer-generated.
Never say never. If any computer wizard develops a desire to write a program that generates chess compositions, I’m sure it will happen. The flavor of such puzzles might be a bit different (and even more bizarre) than human-generated compositions, but that’s life.
Bill Smythe
Well, on my Mac, the chess utility that is included with the operating system keeps me busy.
My train of thought wandered off the track. In basketball, one player gets credit for the shot and another for the assist, i.e. passing the ball to the other player to shoot it. In hockey, up to two players can get an assist.
Well, I was thinking, if this was the rule in chess, one of the chessmen would get the checkmate (or sometimes two) while there are two types of assists: Prevent the king from capturing the checking piece, and prevent the king from moving to an empty square.
I do this because the chess utility, set for figuring three moves ahead, often repeats openings and follows the formula so rigidly that I know how to checkmate in 10 moves. The next level is too difficult and the machine takes too long. It is only supposed to spend one second per move but it often takes more. So instead of rushing right to the checkmate, I find various ways to capture all the opposing chessmen except the king, being careful not to checkmate while there are still opposing chessmen and being careful not to stalemate. The machine lets me take back moves, but not if I checkmate the king or produce a draw. If the machine checkmates me, I can take back moves.
There could be a scoring system whereby you get more points if you checkmate with the bishop assisted by two rooks than if you checkmate with the queen and rook.
Several months ago, ChessBase ran an article about a program that composes chess problems. I agree with the commentators who opined that the quality of the problems wasn’t very good. However, early chess-playing programs also weren’t very good, but the quality eventually increased.
theproblemist.org/
I think the British Chess Problem Society summed it up best with this quote:
“I think the best way for the player to begin to understand and enjoy problems is through using the chess skill he already possesses to solve them simply as chess puzzles. This will give him a sense of achievement and in the search for the answer he will have to examine what happens in the problem; as he does this – at least this was my experience – he will begin to appreciate the beauty of a good problem and to understand that although it is partly a puzzle (and can be enjoyed as such) it is also something deeper and more significant. While it is the struggle in a game of chess that is the central element, most players get pleasure from the ideas that occur in a game and not just from winning; so, although at first some may find problems rather bloodless, there are few who will not grow to enjoy them. In any event, I am sure that it is worth while at least making the attempt to widen one’s range of chess experience.”
C. H. O’D. Alexander
(from the Foreword to Chess Problems: Introduction to an Art by Michael Lipton, R. C. O. Matthews and John M. Rice, published by Faber & Faber in 1963).
theproblemist.org/what-are-c … troduction
I’m not saying that computer program will never spit out a good chess composition, but that quote kinda sums up why I think chess compositions is more than just pieces on a chess board. The creator has to think about what the central idea or theme of a problem he wants to create, then systematically implement that idea in a way that actually creates a good composition. I do think compositions are more than puzzles, but in a way, it’s art unto itself.
I think if I can really learn how to solve compositions, maybe I’ll be able to understand enough to make some of my own. Be a long road I think. Time and Tides.
Is it possible to create a position where all 16 white pieces are either mating or assisting, i.e. where each white piece is attacking either the black king or one of the 8 squares immediately surrounding the black king? (Double checks and double assists are permitted.)
Bill Smythe
Black Ke4
White Ke2, Ra3, Rh3, Rf5, Rd5, Rf6, Rd6, Nf1, Nd1, Bg1, Bc1, Ba7, Bh6, Bh5, Ba6, Qe5(having just captured a piece). There were eight underpromotions.
Is it possible to do it with 8 pieces and 8 pawns, no promotions?
Bill Smythe
In another thread, an example of a position where stalemate was acheived in ten moves from the starting position was given. If a computer was asked to perform a similar task, I guess unique methods could be considered “compositions” by the program. It would just take a human to set the parameters and cull the results.
Black Ke5
White Pc3, d3, e3, f3, g3, g4, c4, c5 (4 total captures needed), Be7, Bf7, Kd7, Ra6, Rh6, Qe6 (from b6 and having just captured a knight), Nd2, Ne2
Can you compose a legal sequence of moves to get there without stalemating?
Probably. The black knight bounces between g7 and e6 while everything else moves into position.