AI Bonuses

Thalassicus

Bytes and Nibblers
Joined
Nov 9, 2005
Messages
11,057
Location
Texas
This compares AI bonuses between the unmodded game and the G&K Enhanced Mod (GEM). King is the average difficulty for GEM players. Blue numbers indicate the AI is stronger than the unmodded game, and red numbers show the AI is weaker.



AIs spend gold smarter in GEM than vanilla. This makes the game more challenging at all difficulty levels. Click for details.

AIs start with (0.25 x difficulty x gamespeed) units in GEM.
  • Land: archers.
  • Coastal: triremes and archers.
  • 50% fewer units if not militaristic.
AI capitals have (difficulty - 2) :c5production::c5gold::c5science::c5culture: bonus yields. They get 25% less if not militaristic, and 50% more if they scored poorly on the worst AIs poll. The production bonus is limited by capital population.


Difficulty
1. Settler
2. Chieftain
3. Warlord
4. Prince
5. King
6. Emperor
7. Immortal
8. Deity
Gamespeed
067% : Quick
100% : Normal
150% : Epic
300% : Marathon
 

Attachments

  • AI Bonuses.PNG
    AI Bonuses.PNG
    53.6 KB · Views: 10,848

Txurce

Deity
Joined
Jan 4, 2002
Messages
8,280
Location
Venice, California
That's very helpful. You may want to briefly explain how to read the percentages. For the layperson higher would seem better, but it's not always the case (happiness for example).
 

nemini

Chieftain
Joined
Jan 28, 2008
Messages
41
I think the AI gets FAR too many starting units. I mean, cmon, two archers, two work boats, two workers on top of the settler and warrior the human player gets?

How are we supposed to compete with that?

Screenshot from a game with the following settings: Difficulty Prince, Gamespeed Epic.
 

Attachments

  • Civ5Screen0004.jpg
    Civ5Screen0004.jpg
    300.8 KB · Views: 1,664

Thalassicus

Bytes and Nibblers
Joined
Nov 9, 2005
Messages
11,057
Location
Texas
The AI is not very good at setting up a defensive line with siege units behind, so the strength bonus helps compensate for that.

In v151.3 I added a 25% AI cost reduction for wonders.
 

Thalassicus

Bytes and Nibblers
Joined
Nov 9, 2005
Messages
11,057
Location
Texas
Oh right, it was in v.2. :)

On a related note...

I chose a flat -25% wonder cost across all difficulty levels. This is different from vanilla where wonders are easy to get on Prince and impossible on Immortal. I don't like handicaps giving us fundamentally different gameplay. My goal is for harder difficulties to provide the same "style" of gameplay, just harder. This is also why middle difficulties have high AI experience bonuses. I want equal economic-military competition with the AI on every difficulty.
 

Zaldron

King
Joined
Jun 5, 2006
Messages
822
I'm a little curious about this 25% bonus: A few versions back (around v143-v148 I'd say) the emperor AI was beating me to all three of GL, Pyramids, and HG in about 90% of my games even given a concerted effort by me (mining->writing). Clearly something was working in those versions that enabled the AI to get the wonders without a significant production bonus. Did something else change that reduced their ability to produce wonders resulting in the need for this bonus?

EDIT: I'm worried that with this bonus a human player will *never* be able to get the early wonders.
 

gdwitt

Prince
Joined
Sep 27, 2010
Messages
359
Location
Austin, TX
I agree on the wonders. I've been getting nearly all the wonders recently on emperor.
And it's not due to a tech advantage.
Please help on the handicaps file:
What is:
AIWorldConstructPercent
AIWorldTrainPercent'
AIWorldCreatePercent

I can guess at what the non-world yyy percentsmean.

And is it true that lower numbers after these means quicker?

Except for ResearchPercent a higher number means quicker?
And for AIWorkRateModifier a higher number also means faster?
 

Seek

Deity
Supporter
Joined
Aug 7, 2010
Messages
3,410
I agree on the wonders. I've been getting nearly all the wonders recently on emperor.
And it's not due to a tech advantage.
Please help on the handicaps file:
1) What is:
AIWorldConstructPercent
AIWorldTrainPercent'
AIWorldCreatePercent

I can guess at what the non-world yyy percentsmean.

2) And is it true that lower numbers after these means quicker?

3) Except for ResearchPercent a higher number means quicker?
And for AIWorkRateModifier a higher number also means faster?

1) The first is the wonder cost modifier and the others are unknown I believe.
2) and 3) Both yes.
 

Thalassicus

Bytes and Nibblers
Joined
Nov 9, 2005
Messages
11,057
Location
Texas
The reason for the AIs poor performance is Palaces no longer give 5:c5production:, and gold can't be used to buy wonders. AIs generally prioritize food in the capital, so it has lower production for AIs than old versions.

In v151 people were saying the AI never gets wonders. If a 25% bonus means the human can never get wonders now, we can try a middle ground like 10%.
 

Thalassicus

Bytes and Nibblers
Joined
Nov 9, 2005
Messages
11,057
Location
Texas
Either AICreatePercent or AIWorldCreatePercent modifies the cost of projects (Apollo, Manhattan, etc). It's not clear which is which. AIWorldTrainPercent is unknown.
 

rfxmills

Warlord
Joined
Jul 4, 2011
Messages
172
Location
London, UK
Could you bring up the production for AI palaces from 2:c5production: to 4:c5production:? And maybe a low modifier, like 10% you suggested, to help them get situational wonders in other cities.

I know one of the reasoning for reducing the production bonus was to promote decentralization, but I think a small :c5production: increase for the AI in their :c5capital: will be good for producing settlers (I'm guessing they won't switch to production focus for this either) early on, which helps them get an empire to be decentralized with :)

Also a bit more centralization for the AI helps it cope with losing cities better and stay in the game, losing the :c5capital: also loses the palace, so it won't OP anyone else.
 

markusbeutel

NiGHTS
Joined
Sep 26, 2010
Messages
1,976
Location
Vancouver BC Canada
The reason for the AIs poor performance is Palaces no longer give 5:c5production:, and gold can't be used to buy wonders. AIs generally prioritize food in the capital, so it has lower production for AIs than old versions.

In v151 people were saying the AI never gets wonders. If a 25% bonus means the human can never get wonders now, we can try a middle ground like 10%.

Have you tried boosting WonderCompetitiveness for the AI? Setting this at 7+ for the AI pretty much makes them competitive without having to give specific bonuses. I've set it at 9 for all AI, as situational starting terrain is generally random enough to produce a varied ebb and flow as to which AI build what wonders (and how quickly) from game to game.
 

Thalassicus

Bytes and Nibblers
Joined
Nov 9, 2005
Messages
11,057
Location
Texas
@rfxmills
AIs get a capital production bonus equal to the average human difficulty setting.

@markusbeutel
The "competitiveness" values in a leader's personality affect diplomatic modifiers (You built a wonder we wanted!). I'm mostly certain those do not affect what AIs build. The leader_flavors table determines what they build. Changing these would have detrimental side-effects. The AI can only build one thing at a time, so if it focuses more on wonders, it's weaker in other areas of the game. For example, a high wonder flavor can block the production of settlers.
 

rfxmills

Warlord
Joined
Jul 4, 2011
Messages
172
Location
London, UK
@rfxmills
AIs get a capital production bonus equal to the average human difficulty setting.

Ok, could it get average human difficulty setting +2? If the AI gets a wonder production modifier to compensate for not going into production focus, I think they should get a settler production modifier as well, only around 10%.
 

Thalassicus

Bytes and Nibblers
Joined
Nov 9, 2005
Messages
11,057
Location
Texas

rfxmills

Warlord
Joined
Jul 4, 2011
Messages
172
Location
London, UK
That would make the early game harder, instead of the late game, which would counteract the goal detailed in the early vs late difficulty balancing thread.

Is that just the +2:c5production: on the :c5capital:? Or the modifiers as well?

I find that default focus really hates production tiles, I always have to manually assign citizens to get some decent hammers going. Could you give +1:c5production: per 3 or 4:c5citizen: for all AI cities? Which would fit in with the early/late scaling and is decentralised.
 

Gamewizard

Emperor
Joined
Jan 24, 2011
Messages
1,234
I agree that the city default focus is a problem. For example, it will prioritize a 2 food 1 gold coast tile over a 3 production 2 gold mined riverside Gold tile, as long as you have a happy empire. This is clearly giving the AIs problems, particularly early in the game when the empire is really happy. They will never be competitive for wonders when they are working all the coast tiles first.
 
Top Bottom