Are you sure? I just ran some rounds, and Marshmallow beat Sparrow 75-25. --Dummy
Did it? Did you run version 0.9.9.3? Maybe you didn't run enough rounds. I will run a few more tests now for sure. -- PEZ
Ow... I think I selected an older version of Sparrow --Dummy
What can be just as interesting as trying to beat your ProblemBots without degrading your performance against other bots is trying to become a one of someone else's ProblemBots. Been a hobby of mine. It looks like some really good strategy might be coming out of it, too... -- Kawigi
Does SandboxDT have any problem bots? Any at all? Just curious... I don't think I have yet seen a single robot that can beat it one on one. (keyword *yet* (hopefully!)) -- Vuen
The last 1v1 problem bot for DT was David Alves' Duelist at the turn of this year - I had to create a new version to beat that. More recently ABC's Tron had the edge on DT in Melee until I put in a new melee movement for DT. It may be possible that some of the pattern matchers may have a 1v1 advantage against DT in a 10 round battles with no data files - but this is a little artificial. -- Paul Evans
PEZ, iiley and I are all working on producing another problem bot for you, Paul. We'll let you know when we think we have one. :-p -- Kawigi
While PEZ's original description says you lose against problem bots, it is more accurate to say that you do not do as well against the robot as you are expected to, based on your ratings. If you have a high rating and cannot soundly beat bots with low ratings, those bots are likely to be problem bots, and keep your rating from getting much higher. A common situation lately has been ram bots, which tend to score around average (1600) and are difficult to beat as devastatingly as other bots that keep their distance with inferior movements and targeting. -- Martin