A fair number of the original games end up in draws even when the players are some of the best Checkerists ever. I know some Grandmasters, like Tinsley would just settle for a draw when he already had enough wins for a tournament. Still a lot of these endgames likely couldn’t be won over the table. I wonder if these should be in a class by their own and not in the regular puzzle category. 4935 is a good example.
Yeah these are unfortunate, these endgames are often winnable only by a long series of moves beyond human comprehension, but the puzzle ends because you promote a king. I don’t really know what to do about them, or how to detect them automatically. If they’re not fun, maybe they should just be disabled. Or like you said, if they are fun they could be moved into a “studies” section, but even then they should be many moves longer which I don’t have a reliable way of doing.
I do have some improvements to the problem generator ready to go, it should at least force puzzles with only a king advantage to continue until there are multiple winning moves. It will have the same problem sometimes but hopefully less frequently. And I can now run it on existing puzzles to update or replace them. But it’s also stricter about the puzzle requirements and will disable some existing puzzles. Assuming of course there are no bugs.
Because it’s going to disable something like 10% of existing puzzles I haven’t gotten brave enough to run it. But maybe I should, maybe it’s good riddance. I’m currently thinking I should I let it finish scanning for new puzzles first and then have it try to clean up old ones.
What do you think, should I run this “verification run”?
Maybe do the new puzzle scan before the verification run.
It would be interesting to move the ones which were draws in the original game to a separate area. Then we could manually sort for ones drawn by top players that weren’t obvious mistakes. These would be really tough endgames.
Just looking at which games were draws would be a start, but a lot of the draws are because of missed tactics or the tournament situation and those I would want to keep in the main set. I don’t know how to tell those apart. Maybe I could look for puzzles where the players played all the puzzle moves and then later ended up in a position where the evaluation is drawn. That’s certainly a good indicator that there was a lot of play left in the position. But even that wouldn’t solve the problem because many of these endgames were in fact won. But being winnable by Tinsley doesn’t make the endgame clear. Food for thought.