Currently my sudoku software does a very crude job of estimating puzzle difficulty. There’s an especially wide range of difficulty levels in my “super tough” puzzles - some of them are pretty easy, others are genuinely “super tough.”

My experience shows that most other computer programs do a pretty poor job at this too. The Los Angeles
Times puzzles, for example, are rated very inaccurately.

I’m launching a little informal research project which should help my software do a better job at this. The
basic idea is to find out how difficult real humans find the puzzles, and use this information to improve
the software.

If you’d like to help out by solving a few puzzles and sending me your results, then read on….

This entry was posted on Thursday, September 1st, 2005 at 4:00 pm and is filed under Jim's Projects, Puzzles. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.