Thanks a lot for your reply. Sorry for my poor English.
- Can you give some more details about this? What map type? How did the religions fall out?
It was on custom continents, 3 continents, large map, 14 opponents, monarch, aggressive AI, 12/12 build. The thing is I had enough space to develop and leverage Asoka's and India traits... The AI didn't attack early enough, maybe because I had luck (there were no very close enemies until some time, they were something like 10/15 squares away).
I used my advantage by building cities far away from my capital to block the enemy progression in the north and in the south, it worked well and allowed me to fill the empty space later. The AI declared a few wars between AI (2/3 vassals created in the whole game, 4/5 wars, that's not muxh for an 'aggressive' AI), but nobody really tried to stop me to win. Alexander who was second finally invaded from another continent with 4 transports, but with obsolete unit (riflemen vs my infantry), he got crushed immediately. He should have made alliances (there were none in the game, though the option was on) and attacked earlier, when the technology he had was on par with mine.
The religions were similar on my continents, confusianism for me and all the northern part of my continent (Hatshepsut, Hannibal, Mehmed). It allowed me to become quickly friendly with Egypt and Carthage, though they never accepted even a defensive pact (maybe because I was first). Mehmed was pleased because 'our close borders sparked tensions', but none of them tried anything against me.
The southern part was hinduist (Roosevelt and Bismarck). Bismarck attacked early but the stack was a bit too small (I think it's a good idea you had with this small earlier stack attacks, anyway) to be dangerous, then no attack either though they didn't like me. Roosevelt could have attacked earlier. He finally did, but too late and I crushed him too. I vassalized both, ultimately. I used the AI bonuses by giving a few money and military techs to them when Alexander invaded, and they immediately upgraded to infantry (good thing)
My success wasn't dependant on wonders, I built only the later ones, but I totally agree with your changes, early wonders must be built a bit more again by the AI.
- How did you win? How was your power relative to the AIs? It is hard for an AI that is behind in tech and behind in power to do anything to stop the tech and power leader from winning.
One thing that I think could definitely use some refinement is to make 'limited' wars more pillage and harrasment wars, rather than just the same as total wars with the AI more willing to accept peace.
I won by space race but a conquest would have been easy considering my advantage (4000 points vs 2000 for the second), just a bit boring (when I know it's won, I just finish as quickly as possible).
I was 4th or 5th in power, usually not so far from the first, except at the end when I had only 2/3 of Alex power, but maybe he added my 2 vassals' power and that's why he didn't attack more (?) I agree with you it was hard to stop me at the end, but earlier it would have been possible with an alliance. Overall the AI did a relatively good job, but it whould have been more aggressive, and earlier. I agree with you on the limited wars, when the AI sees the stack won't allow him to take cities, or that he won't win the war, maybe it should pillage instead (?) Bringing someone else (or even 2 more AI) in the war is a must too, especially when you're weaker. Had 3 AI attacked me simultaneously and if possible no too late, I would have been in real trouble because of the sheer number of units (if they're not too outdated).
- Can you give some examples? Large stacks should not be attacking with poor odds. The wandering in front of a more powerful stack is a different issue. Unfortunately, we do not have a lot of control over the pathing algorithm, we can choose where to go, but not how to get there.
Well more often than not I see one or two (or three) units from his stack attack my cities or my stacks, often they die and when they die it probably thinks it won't work and stops attacking. The question I asked myself is why did it attack in the first place. The odds were apparently relatively low since it rarely / never works. Aren't there a bug in its odds calculation ? 80 % win is a minimum to attack in my opinion, except if it can force the win by sacrifiing a great number of units (for example because he has 3 times the number of defending units). The process of bombarding then sacrifiing siege units then attacking is ok, but it doesn't apply to all the cases, sometimes it's just better to attack immediately (for example when you can force the win because of the number of units...)
One of the things which kill him is the lack of medic in his stacks (a Medic 1+2+3 you can make with a great general does wonders in a stack to allow them to heal quickly and attack again), another is the lack of defense (bad stack position, lack of tile defense and so on). Is there a way to analyze the danger around the stack (for example my own stack approaching to destroy his) ? In attack, is there a way to evaluate the whole stack attack like humans do by comparing units, promotions and odds, like humans do ? It would allow to know if it will probably work before the first unit attacks. I mean when a human attacks a city with a stack, he does it because he knows he has 80 or 90 % chance to take the city in this turn or the next... otherwise he doesn't attack. And usually he's right. Can the AI do the same complex calculation ?
Controlling the path used would allow us to make the AI stacks use hills and forest to protect while moving.
Last thing, the units chosen are sometimes too generic, I saw too many mounted units with combat 1 and 2 promotions, not enough city raiders 1/2/3 melee (macemen...). For cats and trebs, city raiders 1/2/3 are a must I think, they will still do collateral damages and bombard well anyway, so...
Overall, the AI did a good job, and I already see all your improvements. Make them more aggressive, make them use their stacks better, and make them really try to win (by attacking and checking who wins), it will be great
After all its just a beta version!
