AI Bonuses

@Pep
15 :c5science: base
03 :c5science: from bonus
01 :c5science: from population
19 :c5science: non-militaristic science per turn

15 :c5science: base
06 :c5science: from bonus
01 :c5science: from population
22 :c5science: non-militaristic science per turn​

22/19 = 16% higher turn 0 science rate for a struggling militaristic AI vs a doing-well peaceful AI. This effect lessens over time as science rate rises. The ratio is similar for the other yields. I don't think it would make sense to nerf the least successful AIs, but I'm open to suggestions if you have an idea for a new buff to replace the old one.

@rfxmills
I added the AI leader-specific bonuses about a week after Txurce posted the poll in April.
 
There is no doubt in my games that the "worst" AI are no longer the worst, because they can now sometimes get a head of steam at game start, and make that last long enough to be competitive until the later game.
 
@Thal

You are right regarding to science. A 16% advantage in science is really no big deal. What worries me is the production bonuses. At turn 1, Emperor difficulty, considering a capital on flat terrain:
  • "Worst AIs": 9:c5production: (2:c5production: terrain, 1:c5production: Palace, 6:c5production: Handicap bonus).
  • Normal AIs (non militaristic): 6:c5production: (2:c5production: terrain, 1:c5production: Palace, 3:c5production: Handicap bonus).
  • Human player: 3:c5production: (2:c5production: terrain, 1:c5production: Palace).

So "worst AIs" = 50% more production than Normal AIs (non militaristic). Currently Romans are considered a "worst AI" and in a recent game I saw how they almost exterminated the Incas (they survived because they had one city on an island).

On the other hand, comparing the Human and AI production, "worst AIs" tripple the production of Human players while Normal AIs (non militaristic) double it. I think this is a bit overpowered. Recently, I've been rushed early in the game with warriors and archers and also by Romans. I lost that game quicky due to the overwhelming number of AI units, even with Civ 5 poor tactical AI.

IMHO, I would keep the science and gold bonuses and lower the production ones. Also, as I'm always the player with the greatest number of policies on my games, I would increase AIs culture bonuses for AIs that don't favor culture.
 
The worst AI benefit most from production because they have a chance to roll over someone - including a human player caught off-guard. Giving them only a boost in science, etc, isn't going to do much for them on a relative scale. They need the quick expansion that production (leading to units) can give them, so the science has enough to work with for the next 200 turns, or longer.
 
Without a big initial production bonus the AI can't get early world wonders before the human. Someone pointed out elsewhere AIs use only the default city focus, and don't change to production focus when building wonders. The bonus is also helpful for AIs to have any early success at warfare.
 
The production bonus directly affects AI wonder competitiveness. Reducing the bonus (even slightly) makes it easier for the human to get the early world wonders. When the bonus was bugged at 2 instead of the intended 5 for a few weeks, there was lots of feedback about how easy it was to get early world wonders... even with a 50% wonder cost reduction (bigger than the current 25% reduction).

It would be better to change the gold, experience, or starting unit bonuses if you're encountering a problem with military rushes. Those changes won't affect wonders.
 
I think it's fun to compete for wonders with AI, but it should work in both sides. Now I can't never win the AI when building Temple of Artemis (the earliest wonder). It's not a big deal, through. I think the best solution could be lowering the initial AI production bonus and then increasing it with city population. For example, starting with +1 :c5production: and adding +1 :c5production: per :c5citizen: until population reaches 6 (for "worst AIs"). A 6 :c5citizen: human city with citizen focus on production can compete, at the expense of growing, with an 6 :c5citizen: AI city with +6 :c5production:. I don't know if it's possible with the current modding tools though.
 
My very next game, playing Babylon, I was attacked by the Mongols on t31 with 6 warriors and 2 archers. I had two scouts and my starting bowman, but he was far to the west. I rushed the Walls, and took Oligarchy as my next policy. With the help of some barbs - who also killed one of my scouts - I parked a scout in my capital and fought off the Mongols while my bowman slowly made his way back and picked off the last of the invading force. What worked here was distracting them... but the real point is that it's possible to hold off one of these supercharged "worst" AI's, as long as we have some resources available to do so.
 
@Pep
A pop-based bonus is a very good idea. I'll do that.

@Txurce
If only Mongolia had directed that attack towards citystates, they'd probably do rather well... and it would be less annoying for human players. I wish I could get those AIs to go after citystates. :badcomp:
 
I'm still playing with 154.1... and they are going after CS with a vengeance in my current game. Mongolia may have been repelled by me, but they just kept expanding - and attacking CS. So did Spain, and both have has all the CS ally against them. (France is close behind.) On t230, France now has 21 cities and Russia 18. Mongolia is down to 9 and slipping, but they took one of my cities and had 12 much earlier in the game. What they did is keep on building units; my mistake was to stop when I had a window to force a favorable peace.

So I'd say it's working pretty well!

I'll comment on 154.1 in general in the other thread.
 
That's due to me forcing them to declare permanent war on their nearest citystate neighbor, and the problem is that appears to bug out the citystate "help me in war" quest. I might have to return things back to the earlier passive state.
 
In my latest 154.1 game Songhi was at war with a CS on turn 1. I got a we need help against the tyrant request from them at the end of turn 1. I thought that was crazy. They never stopped. Declared war on one Civ or CS after another. Everyone ganged up on him and snuffed them out.

So that is hard coded? Huh...
 
Thal, you may want to sticky this thread, given how often people will want to refer to it.

Assuming the AI science boost helps as intended, maybe the biggest obvious AI issue right now is negative gpt in the late game. In v154.1, playing well into the 300s (!), France had a lot of space, a large but not crazy army, and a -1200 gpt. Interestingly the successful AI are more likely to get into bigger gpt trouble. However, this is only one game, and the overall situation may have been better than pre-154, so don't reach a conclusion based solely on my input.

V156.2 has a lot of potential improvements, by the way. It's definitely a "balance" version.
 
You were in contact with them on turn 1?

The alternative is Songhai never attacks citystates.

OK maybe it was turn 2. I moved my settler to the right one or two spaces and when I founded the city there was a CS revealed that immediately asked me for help against Songhai.

No I don't want to change it so Songhai never attacks City States. just maybe not have the ability to do it on turn 1.
 
Thal

I was wondering how to change the capital yield per turn for the AI: is that in the LUA or XML? If in the LUA which line do I make the changes from?

Thanks
 
thanks thal

edit: okay I have no idea how to change that lol. I am about as ignorant in code as someone can be. I tried to play around with the settings but I am very confused. AI_Building is the Palace for AIs right? Is there a way I can make the capital yields for humans and AIs 2 on production, gold, culture, and science?

Thanks again
 
Top Bottom