|08/11/2012 16:17:50 A testing question|
A question for experienced map makers.
When creating a new map that is not symmetrical, but where you still desire a degree of balance between the players, it is necessary to test it by playing it. This playing phase can take a while because you are waiting for players to complete their turns. The player’s variable skill both over all and on a given day can greatly influence the out come and bias the test results.
To get around this is it possible to allow the AI bots like ai_ZappBrannigan to fight amongst themselves? They are identical and they respond punctually just being algorithms thus negating this bias mostly.
For example lets say I wished to test a map with three sides Blue, Red and Purple. I add an additional side Yellow and then change the colours of the respective sides. In one corner of the map out of the way of the action and immune to the others I place a single solitary Blue Trooper. The colours of the other three sides pieces and bases are changed. Blue becomes Red, Red become Purple and so on.
Then I start up the game inviting ai_ZappBrannigan as Red, Purple and Yellow. Then every half hour or so I log on and end my turn. I can watch to see how long the battle continues before one of the three sides attains overall dominance. If it is a relatively balanced map then this should only be after many rounds of play.
|08/11/2012 16:46:45 Re:A testing question|
|Don't use bots - Last I recall Zapp doesn't build all the units, doesn't handle water, and won't capture harbors or airfields. I think the most common methods are to locate a few people that play quickly and often that are willing to playtest with you or to just create several accounts and play all sides yourself.|
|10/11/2012 18:02:01 Re:A testing question|
I'm always happy to test maps. I don't have to play rated games, I don't quit games, and I move at least twice a day at different times, usually more.
I'll help. I"m also an average/slightly above average player so I can represent the majority.
|14/11/2012 00:46:12 Re:A testing question|
|Bots have been pitted against each other before and it is usually an exercise in futility. They just simply aren't smart enough and often end up in situations where they're just shuffling units back and forth. They are simply not capable of exerting pressure in the right places, at the right times to bring about a swift and decisive victory.|
|16/11/2012 08:50:11 Re:A testing question|
|Must be infernally difficult to write a bot smart enough to detect and exploit weaknesses. Just goes to show the quantum magic of the human neural network|