[Risk] Many rounds

I think that there should be many rounds to determine which script is truly the best. In testing my script, I've noticed a certain amount of randomness - a different script always seem to win. After many trials, one usually beats the rest in terms of most wins. So, for judging, there should be many iterations to help eliminate the random aspect. Has anybody managed to consistently beat the other scripts (using different random maps)?

I think you're forgetting that for random activity, more tests will only prove more randomness ;).

I agree more rounds would be theoretically better, but there are practical problems that must be considered. If most of the scripts will be slow, each game will take such and such amount of time (let's say an hour). Even with only 3 rounds of 4 games, the judging would take 12 hours for each group of 4 scripts. That's 2 days and a half with 20 submissions (which is not an unreasonable number). Since I occasionally leave my computer, it will take about 5 days to run all these games...

With more rounds, it would take even longer, and I don't want it to ;) Like I said, if the scripts will be fast enough, I'll have 24-game rounds, which will give each script 72 games. I think that's more than enough...