Beta 4C: Combat AI Exploit

I hate to post this because it will get nerfed but I suppose....

While watching the very cool battle scenes I noticed that the combat AI seems to concentrate all fire at the largest of my ships until that ship is dead.

Solution? Make one or two of your largest hulls into a dumb target. Put one weapon (sometime the AI ignores it if it doesn't have one gun at least) and then fill the rest up with defense.

Match this up with a bunch of tiny ships that have zero defense and nothing but guns. The AI will sit there and pointlessly hammer away at a ship that it can never hurt while ignoring the small gunship blowing it's fleet to bits. This has to be fixed or combat will become quite pointless and easy for anyone who exploits this.

It takes all the fun out of the combat really. Threat analysis should be the AI's first consideration for firing solutions...
3,725 views 3 replies
Reply #1 Top
Mabey it could calulate the attack to defense ratio of a ship, and determine how to do the most damage with the least losses. 'corse, any AI upgrades go for your ship too!
Reply #2 Top

It's known. We were wondering when someone would post about it.

Don't worry.

Reply #3 Top
The ratio itself isn't the only important part. HP count as well. Basically, you're looking for the path that reduces the amount of effective offensive fire you're facing the fastest.

Examples: If you've got tons of one type of defense, little of the others, and there's only one enemy ship that bypasses most of your big defense, you're probably best off taking it out first. Or at least later in the game, when the difference between the right defense and the wrong defense is 100 vs 10.

Also, if you've got 4 4/4 ships with 10 HP, are you better off shooting those, or concentrating on the one 8/8 ship with 20 HP. My gut instinct tells me to go for the smaller ships, unless you've got enough defenses that the larger ship is the only one consistently damaging you.

Theoretically, it should be possible to compute the "perfect" firing sequence, but with the number of factors involved, I'll settle for something "just" close to optimal (No, I'm not even volunteering to determine if this is NP-complete or not. Though I am curious )