Tweet
More AI Flaws & Suggestions
1st, I think Soren did a great job with the AI for the most part. Below are some... problems I found & have read here with the AI tho. Please add any additional ones I don't mention.
1.Most recent one I've discovered is the AI still HIGHLY values dead-end techs even after the wonder has been built for them or if the unit is obsolete. The AI should have 0 value (or near zero) for a dead-end tech for which the wonder has already been built or the unit is obsolete. The only exception should be is if it is a required tech to advance to the next age. Furthermore, once say for example Bach's Cathedral is built by 1 player, the AI players researching MusicTheory should change their research to a new tech since MusicTheory is now a dead-end tech with no benefits. This would make for a smarter & better AI.
2.Once you have an AI's Territory Map the AI treats it's territory map like crap in future negotiations. I can often buy their territory maps centuries later for 1 gold! In case you don't know, strategically getting updated territory maps are great for planning attacks against them, seeing where their current roads lead (Rights of Passage value) & to see how developed they are. I was hoping the AI Germans or some AI Civs would be more secretative, but ALL AI Rulers give this away for pennies... I mean a penny. Sometimes they might charge 2-9 gold, but that is rare. And WORSE it sometimes increases the relationship you have with them!
3.If you sue for peace with an enemy and ask for a city, that should work, but if you you break the treaty a few turns later and then try for peace again, they should remember that you crossed them last time and not accept any deal but plain peace treaty." This can be exploited by take city-make peace-take city-make peace-etc. etc.
4.The AI should value it's world map (it's exploration work) much more & guard these secrets. Most should not hand this out so easily even if someone's world map is larger than someone else's world map. Specifically because large areas of claimed land is less valuable then unsettled & unexplored land. The AI doesn't know the difference it's often in a trading frenzy with maps. Revealing where empty ocean is also not smart. Making someone else waste their time to explore empty ocean is strategically superior.
5.The AI knows by instinct which city has the weakest defense, so all I did is keep moving my troops around from city to city every turn... the ENTIRE AI ARMADA would just sail up and down my coast switching targets to the least defended city EVERY turn. Because I kept moving my troops, they wouldn't make up their mind and never landed till I signed a peace treaty.
6.The AI overvalues cities... example - a size 3 city in tundra/pine trees & 1 deer/game, by no means a great city or any potential. It was no where near the Zulus or Babylonians & buried deep within the Aztec empire soon to be absorbed by culture. I asked what the Zulus would trade me for this crappy city no where near them with only a courthouse, their answer: 2 techs, 2 gold/turn, communication with another civ, & Incense!!! Babylonians offered a similar deal with 15 gold/turn included... the city earns 1 gold/turn & heavy distance corruption will keep it that way. Yet, I'm glad the AI doesn't undervalue cities... that would be worse.
7.The MillionDollar Bug - ability to get 9999999999 gold/turn from a civ.
8. Finally, just my opinion, but the AI should be a little more aggressive when the Unique Unit first enters and if it dominants the scene... even if it's just 20 turns or so. Having a 4-2-1 unit & NEVER using it is a waste. Likewise, the Greek's Hoplite is superior defense, why not start a fight if everything else is equal & you have a better defense... that advantage will not be there later. The AI should seize these opportunities. The French Musketeer rarely dominants at 3-4-1, but that would be ideal given the French AI peaceful nature.
1.Most recent one I've discovered is the AI still HIGHLY values dead-end techs even after the wonder has been built for them or if the unit is obsolete. The AI should have 0 value (or near zero) for a dead-end tech for which the wonder has already been built or the unit is obsolete. The only exception should be is if it is a required tech to advance to the next age. Furthermore, once say for example Bach's Cathedral is built by 1 player, the AI players researching MusicTheory should change their research to a new tech since MusicTheory is now a dead-end tech with no benefits. This would make for a smarter & better AI.
2.Once you have an AI's Territory Map the AI treats it's territory map like crap in future negotiations. I can often buy their territory maps centuries later for 1 gold! In case you don't know, strategically getting updated territory maps are great for planning attacks against them, seeing where their current roads lead (Rights of Passage value) & to see how developed they are. I was hoping the AI Germans or some AI Civs would be more secretative, but ALL AI Rulers give this away for pennies... I mean a penny. Sometimes they might charge 2-9 gold, but that is rare. And WORSE it sometimes increases the relationship you have with them!
3.If you sue for peace with an enemy and ask for a city, that should work, but if you you break the treaty a few turns later and then try for peace again, they should remember that you crossed them last time and not accept any deal but plain peace treaty." This can be exploited by take city-make peace-take city-make peace-etc. etc.
4.The AI should value it's world map (it's exploration work) much more & guard these secrets. Most should not hand this out so easily even if someone's world map is larger than someone else's world map. Specifically because large areas of claimed land is less valuable then unsettled & unexplored land. The AI doesn't know the difference it's often in a trading frenzy with maps. Revealing where empty ocean is also not smart. Making someone else waste their time to explore empty ocean is strategically superior.
5.The AI knows by instinct which city has the weakest defense, so all I did is keep moving my troops around from city to city every turn... the ENTIRE AI ARMADA would just sail up and down my coast switching targets to the least defended city EVERY turn. Because I kept moving my troops, they wouldn't make up their mind and never landed till I signed a peace treaty.
6.The AI overvalues cities... example - a size 3 city in tundra/pine trees & 1 deer/game, by no means a great city or any potential. It was no where near the Zulus or Babylonians & buried deep within the Aztec empire soon to be absorbed by culture. I asked what the Zulus would trade me for this crappy city no where near them with only a courthouse, their answer: 2 techs, 2 gold/turn, communication with another civ, & Incense!!! Babylonians offered a similar deal with 15 gold/turn included... the city earns 1 gold/turn & heavy distance corruption will keep it that way. Yet, I'm glad the AI doesn't undervalue cities... that would be worse.
7.The MillionDollar Bug - ability to get 9999999999 gold/turn from a civ.
8. Finally, just my opinion, but the AI should be a little more aggressive when the Unique Unit first enters and if it dominants the scene... even if it's just 20 turns or so. Having a 4-2-1 unit & NEVER using it is a waste. Likewise, the Greek's Hoplite is superior defense, why not start a fight if everything else is equal & you have a better defense... that advantage will not be there later. The AI should seize these opportunities. The French Musketeer rarely dominants at 3-4-1, but that would be ideal given the French AI peaceful nature.
Comment