BK: >>The results of tghe last battle would then become the official ones.<<
Sounds like FC to me. FC is a diffent problem, where the results are simply too different from TC, not necessarily in favor of the player.
Kris:
>>it would however be very difficult to learn the AI when (and with what units) it should ignore free strikes and when it should not.<<
Not really, it is a matter of comparing the hits and DEF of the victim to the AT and DAM stats of the attackers. It should be no problem at all to code that if the victim has enough hits left, it may ignore free strikes if the total expected damage to reach the new target is less than 2 or so. Statistics are no problem for a PC.
>>Taking 3 free strikes from infantry just to get one strike at an archer would for example be quite stupid.<<
So, set the threshold to 2 strikes. No problem there. The unit will still attack a more suitable target in this case.
>>Also, if the AI is putting too much afford into making strong units it will (at least on small and medium maps) just get rushed by a human player that is only producing level 1 units. Upgrading too much early is exactly whey the AI is so easy to beat in the demo. (Any good player can beat the demo in less then 10 turns).<<
Not really. The AI doesn't have to play fair, and shouldn't infact. A barracks, a war hall, and a thingy. A little AI production bonus and the new handy hurry feature, combined with a stiff gold bonus, the AI can mass-produce lvl3s in no time.
I never suggested the AI should be so good it could defeat the player without cheating (heavily). Actually, with Hurry Production, the new AI now has a serious advantage over the old AI in AoW1.
>>Another problem is that even if the learn do use your strategy well enough, human players will just do the same, produce only ranged attackers and strong melee units and the weak melee units will become useless.<<
Exceptions, exceptions... as soon as the player can produce as much as the AI, the AI has already lost, so let's not worry about it. The player will not be able to produce enough strong melee units before the end phase of the game.
Mind you: you are implying here that the only use of lvl1 melee units is to exploit the AI. If that is so, then so be it: the building tier would be flawed.
The AI should not be tailored to suit the available building tiers; that is doing it the wrong way around! The AI should be created able to work with the options available! Also consider this with multiplayer games in mind, where such AI tinkering would not help balancing the units at all.
>>That won't work! Any smart human player will be sure to have plenty of healing available, and will just heal his wounded strong units and will use his own ranged units to kill of the AI's ranged units. The player can also use his own strong melee units to attack the AI's "heavy hitters" and heal the damage they tale from doing that too. That will cause the the AI's "heavy hitters" to use up their retaliation strikes and they can now freely be attacked by the players weaker melee. They also will have no movement points left and can't kill any units next turn.
The AI will lose his ranged units pretty fast without having them killing any enemies, and even the AI's "heavy hitters" will at best kill only a few ranged units.<<
Always exceptions... mind you: you can heal only once per turn for each priest you take with you.
Also, as ranged units (with only limited melee capabilities), priests will serve as targets aswell for the bashers.
Finally, with enough archers and such, enemy lvl3s may take more damage that can be healed in on go, and may even die in one turn before they can be healed. I did mention the AI should mass produce units, didn't I?
Again, you seem to have this idea of a fair AI with some actual finesse. I take the more brutal approach where the AI can take maximum advantage of it's unfair cheating, and cheat badly enough to actually become a challenge for expert players.
So every priest the human player produces should be at the cost of a badly needed crossbowman and every heavy hitter at the cost of many.
So the player might have a powerful stack with a healer, or it's weak stack with many healers, but not actually own stacks that are, on average, in raw strength equal to the AI _and_ also combined with all the support to heal units repeatedly.
A player with an army like that has already won; and there's no AI that can do anything to stop it, no worries.
>>If AI's ranged units target the players ranged units the player will lose his ranged units just as fast the AI is losing its ranged units (or faster when the AI is defending) The AI's melee units should attack the "best target" it can reach in one turn, if it has an opportunity to kill an ranged unit it should of course do so, but it is better to attack an nearby swordsman with 3 strikes and actually kill it, than it is to attack an far way archer with one strike and (probably) not kill it.
By attacking weak units first the AI will, even when it lose the battle, at least kill a good number of the player's units.<<
It is all a matter of dealing out damage as efficiently as possible, while taking as little as possible damage yourself. Firing at enemy archers at the back row, reduces the expected damage value, while shooting at frontline troops, gives you a smaller range penalty. When considering that avoiding retaliatory stikes is desireable, the enemy melee units are the logical choice.
Steel Monkey