The Checker Maven

State of the Art


Shown above is what's said to be a "state of the art" kitchen. At least, that's what it was ten years ago. But things move ahead. Ten years is a long time. A lot can change.

In recent times, there have been some significant developments in the field of computer checker engines. Machine learning arrived in KingsRow and later in Cake. That was a huge accomplishment, but it wasn't quite the last chapter of the story.

When today's latest computer engines didn't solve Brian Hinkle's "Prize Problem" world class checker programmers Ed Gilbert and Martin Fierz both looked into the issue. Their conclusions were similar and reasonable. Computer engines are set up for maximum strength during practical play, not for solving highly unusual problems.

But of course Ed and Martin didn't stop there. Ed created a selectable "solver mode" for KingsRow while Martin made some changes to the search function in the Cake engine. The Prize Problem was now solvable. (These new program versions have yet to see public release at the time this article was written.)

But Brian didn't stop his work, either, and went on to create a series of more and more unusual problems which took longer and longer for computers to solve..

So what was the outcome? There may always be problems which it simply isn't practical to program computers to solve in any reasonable amount of time. But the state of the art for checker engines has definitely advanced with the intriguing new work done by both Ed and Martin.

Here's one that Brian sent to Ed and Martin. We warn you, it's not easy.

White to Play and Win


Do attempt a solution on your own before clicking on Read More to see how it's done. These problems are a great deal of fun and definitely test your thinking. Best of luck!null


9-6 2x9 18-14 9x25---A 24-20 13x22 19-24---B 5-9 20-16 9-6 10-14 6-2 16-11 1-6 32-27 6-1 27-23 2-6 23-18 6-2 18-15 2-6 11-8 6-2 8-3 2-6 3-7 6-10---C 15x6 1x3---D 14-10 3-8 10-7 8-12 7-11 White wins.

A---At The Checker Maven we run KingsRow on modest computer equipment. We use the 10-piece endgame database but not the new solver mode. KingsRow quickly sees the win at this point.

B---Making 31-27 pointless, as it won't free up the Black position at all after 24-31.

C---1-5 14-18 5-9 7-10 6-1 15-11 1-5 11-7 5-1 7-2 1-5 10-6 9-13 18-14 5-1 6-9 13x6 2x9 White wins.

D---1x17 7-10 17-13 10-14 White Wins.

Our thanks to grandmaster problemist Brian Hinkle for this crafty problem and solution.

12/04/21 - Category: Problems -Printer friendly version-
You can email the Webmaster with comments on this article.