I rewrote my program to try to test different Jotto AIs. I put the different AI subroutines into the Jotto module, and I rewrote the regular (dumb) AI in a cleaner way than I did way-back-when. I also rewrote the little program I made to get the stats for the output of the test program, to find the average and the standard deviation of the scores. It's cleaner, and it also uses Statistics::Descriptive, instead of my own subroutines to find average and standard deviation. I've got the test program running with the dumb AI now. It's going to be much faster than before, since my scoring algorithm is something like 5 times faster.
I have a run of 1000 games I did the last time around, so I'm generating another 1000 games with this rewritten code to make sure that the dumb AI has around the same average score as the old one. The next step is to rewrite the smart AI. Not only was the smart AI poorly written because of my poor Perl skills, the algorithm just plain didn't work: its average score was higher than the dumb AI. So, we'll see if I can think a little better about the algorithm now.
I'm not sure that I'll do it immediately, though. I want to get some more dishes washed, and it's probably time to start thinking about eating. And I want to start watching the movie before it gets too late.