Hi all!
Thank you, Kevl for narrowing my points.
Xlimi, it
It just seems contradictory to me.
On one hand you seem to want an AI, that allows you to massacre them while totally outnumbered without too much resistence and on the other hand you also want them to be smarter.
Yep. This whole X-COM game is either about dealing with outnumbering forces or much stronger enemies. The progress is always incremental with you getting technologies to beat even more serious missions, with even more strong enemies. You start naked and mendicant (except, I guess, WH40k) and get stuff. This is what major mods about. And this is, actually, a classic RPG scenario.
Almost all of my AIs behavior is controlled by scoring stuff.
Let's think what scoring goals can add more mystery, silliness and deviation, then)
As I believe, your AI clearly can win most human players.
Now it's turn for AI to let human players win, because the game is being designed to entertain.
The key here is that sometimes we can consider that enemy:
- is confused with being attacked (overall awareness of the battle goes on)
- doesn't want to die no matter what (some)
- tries to rescue injured out of player's LoS (doesn't matter they have no healing abilities yet)
- tries to capture player forces alive (interrogation purposes)
- does brutal things (meatshielding, blocking, distracting, destroying surroundings)
- produces mistakes that were described above, like pure miscalculations etc.
- performs tasks out of battle perspective (weird behavior described by kevL).
- performs guard/escort activity for high-tier members until engaged
More than that, as many of the enemies have respectful classes (ranks), they can possess different deviations types. Like, sectoid medic/engineer are likely behave more cautious than sectoid soldiers. While sectoid leader/commander is more likely to coordinate the forces attack while stay safe.
Sorry for putting all this too late
With all respect,