Jump to content

Recommended Posts

Hey developers of SC2. As you all know Ive been testing my global game. I find the AI makes some fairly poor attacks at times. Attacking units when to me it looks obvious they should not. like taking a single tank attacking an entrenched unit in a city when he is alone and surrounded by 3 enemy units. Or attacking an army with a 0 soft attack garrison.

The AI never seems to consider the consequences of a counter attack.

A basic rule I follow in wargames is if you can afford the worst result, then attack. If you can't afford it, dont attack.

It might be prudent that when AI attacks they check...

Ok I got a friendly unit on the left of me and an empty hex on the right

I check 2 to my left and right I got friendlies to hold the line if I die. Ok the enemy has a unit in front and a unit to my left, none to my right. The math is right, so I attack. It is unlikely I will get killed by a counter attack.

Just suggestions.

Link to comment
Share on other sites

Big Al,

Good points and you are pretty much correct in your assessment. This is something that I want to improve upon but one of the hurdles, considering now the size of the maps and the number of units, is speed. Essentially the more detailed the AI planning the longer it takes to play out a turn so this is always a concern... but if I can get it to improve with the right algorithms that are cost efficient then this is something we might just see down the road.

Thanks for your thoughts,

Hubert

Link to comment
Share on other sites

My wishlist for the AI would be for it to not make reckless attacks, specifically lunging through enemy lines and placing its units out of supply. In a few cases this is OK but generally it is suicidal and sacrificial, especially when high value tanks are used. It causes the AI to waste units unnecessarily, which makes it more of a challenge to handicap the AI with extra MPPs or extra units to compensate.

IMHO the AI should more carefully consider supply. Don't push beyond current supply range! I wouldn't mind seeing the AI behave a bit more conservatively, just don't neuter it so it becomes lame. That's the risk, and Hubert continues to look for the right balance. Nothing's perfect.

Link to comment
Share on other sites

×
×
  • Create New...