|02/05/2009 15:57:44 Bot Developer Challenge|
Which bot is the hardest to beat on the four Zapp maps?
|02/05/2009 16:26:13 Re:Bot Developer Challenge|
|I think a easier way to define this is to have a tourney between the bots.|
|02/05/2009 16:32:00 Re:Bot Developer Challenge|
I second that motion.
This message was edited 1 time. Last update was at 02/05/2009 16:32:38
|03/05/2009 21:08:20 Re:Bot Developer Challenge|
So shall we say that bots have until June 7th to get ready?
The games should go fairly quickly.
Each bot plays each other bot on each of the 3 trial account maps twice. Once as each color. I'd prefer 2 or three times as each color for averaging.
Bots are encouraged to have programming specific to the maps.
This isn't going to be a winner takes all kind of thing because some bots will be better at certain things than others.
We need to be able to track total credits (as a measure of bot effectiveness).
All bots and all players generally have exploitable weaknesses and some have more than others. The bot results will indicate
Categories of capability:
1) If we look at the bots' results and a category is there that deserves to be a winner of something, then we'll make a category.
2) Quickest victories measured by round. Averaging this out is probably a good overall measure of effectiveness.
I'll finish this later
|04/05/2009 04:47:16 Re:Bot Developer Challenge|
I'm not sure tracking total credits will work as a measure of effectiveness. At what point in the game would they be compared?
I think quickest victories measured by round sounds like a good measure.
|04/05/2009 12:41:55 Re:Bot Developer Challenge|
I recall two bots being played against each other in the past (I think Zapp and PuzzleBot) and the game went on for days (bots being offline notwithstanding). The pendulum swung back and forth as each bot built inappropriate units and failed to capitalize on obvious mistakes.
I'm not saying a tournament isn't a fun/good idea, but I'm also not sure it would be as swift and decisive as you might imagine.