Saturday, January 8, 2011

Artificial Ignorance (Part 1)

How do you get a chess-playing computer to play poorly?

In the case of programs like Talking LCD Chess  see "Jerome Gambit and the Perfesser (Part I)", Part II, Part III and Part IV – you can limit how deeply it searches for each move. It sees less, it misses more.

In the case of programs like Chess Challenger 7 – see "A Jerome Gambit 'Challenger' "   you can limit the amount of time it spends on choosing each move. Again, playing strength has a lot to do with how far the computer "sees".

Or, when you design a program, you can have it play a relatively decent game, but every once-in-a-while have it choose the 4th or 5th or worse move choice. That's the blunder-as-a-ticking-time-bomb model: with strong players, large mistakes are rare; but with weak players things go *boom* quite regularly.

I was thinking about this "problem" (most of the time programmers are trying to make their chess engines stronger and smarter) today while discussing the Chess Titans program (which is included in the Windows 7 operating system) with my son, Jon.

The youngest of the "Kennedy Kids", home on vacation from his work in Haiti, has been spending more time on chess lately. Of course, he wanted me to show him the details of the Jerome Gambit, and of course I spent a lot of time doing so.

He wants to return to The Haitian Project, play his boss at chess, and beat him with the Jerome Gambit...

No comments: