XML tags for the Planetfall Leaders

[Pfeffersack]

Another problem I noticed is the overall AI leaning towards Democracy. Except Yang, who always chooses PS as soon as possible, hardly no one ever leaves Democracy for PS or Fundamentalism. Not sure if it is too good or if there are not enough outdrawn wars, but the effect is that everyone becomes Lal's friend and signs a DP with him. DPs further concrete peace and make Democracy even more likely. My current game is paralysed by not less then 7(!) DPs. I think at max. 2 or 3 are healthy for a 7 player game.

Not much can be done about the fanatic leaders, the have already "Pleased" as treshold for DPs. But the other 4 are still at "Cautious" and while that may be realistic and atmospheric for leaders being more realpolitikers, it hurts the game too much (at least until we get the SMAX factions; a 12 player game is less endagered to get paralysed, especially with the alien factions).

Eventually the values for war consideration need to go up as well, but I would wait with that until some others have reported their experiences.

 

[MadmanAtW]

IIRC (and someone will surely correct me if I'm wrong), the "favored religion" is mostly ignored in BTS- if you are playing with the unrestricted religion names then they will choose their favored name for the religion, but it has no other gameplay effect. They won't switch to that religion preferentially nor aim for the founding tech based on that preference.

 

[Maniac]

I'm afraid I still haven't looked at your leaderhead XML itself. (I discovered a crash, the fixing of which got my priority)) But in the meanwhile some comments on what you wrote here:

- While the combination of Voice of Planet and Terraformed isn't allowed, it is possible to run Edenism under Hybrid (especially an AI controlled Miriam is prone to do that in games she doesn't switch to Terraformed) - is that intended?

Yeah, kinda. The Edenism religion can be interpreted in many ways. Either it's Christianity Miriam-style, or it could be the desire to turn Chiron into a paradise garden. I guess a paradise garden doesn't necessarily need to mean "completely anti-Planet, pure Terran", so I figured I'd leave the option open. When in doubt, give the player the most choice I guess.

- I noticed that the religios flavour section in the leaderhead.xml isn't used for Planetfall yet. My suggestion would be to attach each religion to one leader (Morgan and Santiago left without a preference):

What MadMan said. Plus 'Choose Religions' should never be used in Planetfall. Eg Voice of Planet being enabled by researching Terraforming is just silly.

For the record, the Spartans' natural religion would be Homo Superior. Of course they'd interpret the ideology a little different than the Peacekeepers...

- Another limit is reached when it comes to modeling the effect of different or (better said) opposed civics - there is no such entry like we have it for religions.

It's on the to-do list, but I probably won't get to it any time soon.

I tried to counter that to some extend by increasing the likelihood of civic switch demands and the attached penalties for refusing, but that of course only affects relations to the human player.

The penalties for refusing a civic/religion switch should be set to zero!!! See earlier discussion on removing anarchy for switching due to a leader demand.

- That would be probably hardest to code...a thing I miss from SMAC is that AIs sometimes offered you a "compensation for your expenses" (credits, tech and/or an alliance), if you entered a war on their behave.

I haven't delved into diplomatic AI at all - no idea if this would be possible.

Another problem I noticed is the overall AI leaning towards Democracy.

The Democratic civic is just very powerful - I also use it most of the time... (also have used Consensus before)
Ideas on how to make the civics in the line more balanced, are always welcome.

 

[Pfeffersack]

The penalties for refusing a civic/religion switch should be set to zero!!! See earlier discussion on removing anarchy for switching due to a leader demand.

Oh, sorry...I was just to busy with bringing everything in line with the current version and forgot that you were already about to change the demands.

I will change that and also the DP stuff I mentioned above and load it up as v3 then.

The Democratic civic is just very powerful - I also use it most of the time... (also have used Consensus before) Ideas on how to make the civics in the line more balanced, are always welcome

Hmm, I don't think cutting the GP bonus would be a good idea...25% e.g. would be just to little IMO to make a real difference for GP generation at all.
That leaves happiness for the biggest cities, which I think is the greater deal anyway. Reducing number of affected cities is likely out due map size issues, but cutting the boost to happiness to 1 could be an attempt to quick-fix the problem. Some other random ideas:
- add increased war weariness as penalty
- If we ever get a kind of U.N. charta, the civic could also force you to obey to it regardless if the council has abandonded it.
- Happiness penalties for certain buildings (Genworker Factory, Punishment Sphere)
- increased support/construction times for formers (represent labor unions)

 

[Maniac]

I have a couple questions after looking at the Excel file:

iLimitedWarPowerRatio : what does this do? Your values are generally higher than those in vanilla.

iDeclareWarTradeRand: what does this do? In vanilla all are 40, except one (Pacal) is 60. In your v2, 5 of the 7 factions are 60.

iDemandRebukedSneakProb & iDemandRebukedWarProb : what is the difference between these?

 

[Pfeffersack]

I have a couple questions after looking at the Excel file:

iLimitedWarPowerRatio : what does this do? Your values are generally higher than those in vanilla.

It is a percentage value and describes the power relation between the AI having that value and a potential target for a limited (=tatical) war. It functions as a treshold, because a declaration of a limited war is only possible, if they the potential target is not stronger then iLimitedWarPowerRatio I choosed higher values to make the AI more aggressive (especially the ones who do not engange in full-scale wars do often)

iDeclareWarTradeRand: what does this do? In vanilla all are 40, except one (Pacal) is 60. In your v2, 5 of the 7 factions are 60.

Yeah, I stumbled about this one as well. First of all, I don't get the exception with Pacal. The guide you linked has an obvious error for this, as it said "Lower = more likely, lower = less likely". I think it's pretty sure that it is a 1/x-chance-per-turn-value and according to a guide, it determines how often one AI tries to bribe another AI (likelihood of requests to human player to become a war ally are handled later among the "contact rands") into an ongoing war.

Now that you say it, I see that I have done here the opposite of what I intended to do - higher values make bribes less likely instead of more likely (which I think would be better). I would now even suggest to put that at 20 - I don't think it will lead to a lot more wars, because there are enough brakes and retsrictions in leaderhead rules for Planetfall, which prevent too frequent and artifcial values. Thanks for spotting this one!

iDemandRebukedSneakProb & iDemandRebukedWarProb : what is the difference between these?

Both determine the reaction of AI, when a human player refuses a demand (I firmly believe that the terminus "demand" is here strictly used for tribute, at least thats what my playing experience clearly shows). Both are percentage value for a single RNG roll after a human refused tribute. Logically, iDemandRebukedWarProb must take place, as it determines the chance of an instant war declaration on the defying player. It that roll fails (or does not take place for some leaders), there is the a second roll (iDemandRebukedSneakProb), which determines if the AI starts planning a war vs. the defyer. I generally set lower values for the iDemandRebukedWarProb , as thats more prone to damage the AI - the low chances are just in for giving certain leaders personality and to keep that moment of uncertainty for the human player...you never now how some of them react. I removed that impulsive behaviour for the rational leaders, though.




Edit: I noticed that the civfanatics modiki now provides information about the leaderhead.xml as well: http://modiki.civfanatics.com/index.php/Civ4LeaderHeadInfos

Reading through it I spotted that iNoTechTradeThreshold probably works different then the guide I used suggested - it is not mood based, but refers to actually happened trades in the past:

iNoTechTradeThreshold An AI leader will stop trading a rival any
technology, if he realizes that the rival is becoming too advanced, and
this happens when the rival has received via trade a certain number of
technologies. This value sets the number of technologies.
(5 stops trading soon; 20 stops trading late (all 5 except Mansa Musa))

I set the value to 10 for all (which seemed to be a kind of low middleground), but maybe it needs adjustment, if playtests show that this hard limit effectively prevents tech trades in the late game.


The Modiki also states that values for iDemandRebukedWarProb are actually ignored in game...I'm not sure about this. I definetly remember some occasions, in which defying tribute lead to instant war (though this seems to happen more rarely than the actaul BTS values for the tag suggest - maybe other factors are still involved?)

 

[Pfeffersack]

I digged through some other threads, which were linked in responses to the guide you brought up - some things are just a lot more complex then expected. The entire war-declaration-procedure uses a lot of other varaibles which are outside the leaderhead. xml - e.g. it seems to play a huge role, if AIs have financial troubles (defined as not breaking even with max. 40% Gold from the slider) - this seems to prevant war declarations most of the time, but OTOH some leaders might be tempted to go to war because of this to fill their coffers by pillaging (sadly I didn't get yet what makes the difference between causing on or the other effect). BTW, are they AIs aware of the new stockpile process? It is nearly always a good idea to have it running in at least one city.
Also they might go to war just because they have gone overboard with unit bulding. That article also brought some light in to when the war/peace percentages attached to the relation levels come into play.

Even the tech trade limitations aren't that fix. The seem to get ignored if relations are
very good or in case of the counter for past tech trades, there is a decay function at work.

I also understand now the tags for different religion effects better - there naming seems to be correct, it is just that there is a hidden factor (you get an extra -1, if the AI has the Holy city of its religion) and that the main penalty is attched immediately and there is just one time another -1 after 5 turns.

--------------------

I have edited the changes towards the DP tresholds and the removal of penalties for refusing civic/religion demands:

 

[Maniac]

I would now even suggest to put that at 20 - I don't think it will lead to a lot more wars, because there are enough brakes and retsrictions in leaderhead rules for Planetfall, which prevent too frequent and artifcial values.

Have you figured out why the AI seemed to declare less wars in Planetfall?

The Modiki also states that values for iDemandRebukedWarProb are actually ignored in game...I'm not sure about this. I definetly remember some occasions, in which defying tribute lead to instant war (though this seems to happen more rarely than the actaul BTS values for the tag suggest - maybe other factors are still involved?)

For the record, I can't remember EVER getting an instant war declaration because I refused some demand.

I have edited the changes towards the DP tresholds and the removal of penalties for refusing civic/religion demands:

Thanks!

 

[MadmanAtW]

For the record, I can't remember EVER getting an instant war declaration because I refused some demand.

It happened to me in a recent game, and I'm pretty sure it isn't something that was added by BetterAI. I think it is a possible but rare circumstance in BTS.

 

[Pfeffersack]

Have you figured out why the AI seemed to declare less wars in Planetfall?

Not yet. I did some further editing and testing after v3 mainly by increasing the war rands again. The problem is that all I cause with that is AIs going into war preperation state forever ("we have enough on our hands now"), but actually they never declare. Thats even worse, because then you (and other AIs as well) cannot even bribe them into wars.

Some theories what could be reasons (after lecture of this thread: http://forums.civfanatics.com/showthread.php?t=286180&page=2 ):

- Planetfall AIs seem to be often in great financial troubles. They build a lot units, which they pay support for and they neither seem to use the maintenance process nor do they consider a switch to Police State (ok, Emperor gives them discounts and free units...but it is not uncommon for me - if have the espionage means - that I see the AI running the Gold slider somewhere between 50 and 90% and not even break even with it). Financial problems of the AI can prevent war declarations:

A) 4. AI is in financial trouble when (Inflated Costs + foreign trade deficit) > 60% of (beakers per turn + taxes income + foreign trade surplus), threshold increases by additional +8% when aiming for cultural victory, +12% when at war or preparing one, +10% when researching Future Techs

1. An AI in financial trouble only opposes war if it is currently not using AI_STRATEGY_DAGGER and it isn't an aggressive raider who likes wars as a means to get money. Your understanding of the condition is correct -- the developers IMHO intended that an AI in financial trouble will still consider to start one of the 3 types of war (MaxWar, LimitedWar, DogpileWar) if its respective iWarRand is smaller than 100 (iMaxWarRand < 100 --> consider a MaxWar-Raid; iLimitedWarRand < 100 --> consider a LimitedWar-Raid; iDogpileWarRand < 100 --> consider a DogpileWar-Raid). However, as described in post#84, they made a slight mistake so that now the iMaxWarRand affects both the war check for MaxWars and LimitedWars. Since the number of leaders with iLimitedWarRand < 100 (20 leaders) is way bigger than the number of leaders with iMaxWarRand < 100 (6 = only Alex, Boudica, GKhan, Monty, Ragnar, Shaka) this bug decreases the likelihood of limited wars started by financially troubled civs.

Given war rands <100, AIs can behave the opposite way (declaring war for "raiding" and acquiring money), but I don't seem to get that working now.


- OTOH, it is beneficial for warmongering if they spend a lot of money on unit support (al long as they don't go broke of course!):

I think what this means is that when a high fraction of the AI's total expenses is due to its military units, but the AI is not in financial trouble at the same time, then the likelihood for preparing or starting a TOTAL_WAR will increase significantly (even more with Aggressive AI on).

Example:
iUnitSpendingPercent = 19%
iHighUnitSpendingPercent = 6%:
iWarRandThreshold = 4 (with AggAI 9)

RNG can roll [0,1,2,3,4] for a total war! That's 5 times the probability of the usually required RNG = 0 (which is still valid for limited and dogpile wars)!

Considering the large amounts of units the AI starts with at deity and the high discounts it gets for training more, this might explain parts of the often occurring DoWs. There should also be a peak after Courthouses (or earlier Ziggurats!) since this will quickly lower the maintenance costs for cities. Maybe selling resources to the neighbors or getting them to adopt expensive civics might help a little....if I understand all of this correctly.

IMO, that's even more of an argument to teach the AI using the maintenance process - as that would keep their unit spending up - the unit maintennace still has to be payed, there is just another source of income to cover it!


- There might a problem with the units the AIs build or with the strategies they attach to them - in BTS e.g., Saladin has a personality which severly hinders him in regard to offensive wars (Planetfall leaders have no favourite AI strategies set yet, but maybe setting something offensive her would help):

Sissy Sal:
Saladin has discarded his war plans (out of WHEOOH again) because he has assessed his offensive power not high enough after the (fixed!) turns of preparation.
If I understand it correctly the game mechanics for this assessment are rather funny as they only let Sal consider his offensive task force (= number of units with AI ATTACK, ATTACK_CITY and PILLAGE) in relation to the number and size of his cities. In contrast to his decision to start preparing for a war, this judgement is completely independent of your power and the strength of the units he has, so upgrading his Camels and Jumbos didn't help either. As part of his personality he also favors to build units with (the wrong) UNITAI_COUNTER (XML-value UnitAIWeightModifier) during times of peace. Together with his protective trait this turns him more into a strong defender than a real threatening offensive aggressor (compare to Monty!).

I went back to the 990AD save and deleted all your units, canceled open borders and bonus trade deals with Sal, traded with his worst enemy Mehmed, switched to Pacifism and Christianity so I got him to -9 FURIOUS and became his new worst enemy--he still switched out of WHEOOH in 1130AD !!!
(...)
It works the other way around too. This time I buttered up to Sal and switched to Theo|Islam, gifted him Gunpowder which got him to PLEASED so that I could sign a DP after he researched MT. Then I WB-ed him 20 Warriors (UNITAI_ATTACK) and this happened:

- I also checked the power ratings of the units (= contribution to "Soldiers" in the demo screen and also relating into the power grpah), since the power evaluations play a great role in the war decision routine as well. I found (with one exception, its Helion vs. Cyborg - the former is rated lower by 2 despite having better values in everything) resonable values here, but since techs and buildings provide power as well, there might be room to optimize (e.g. giving units with offensive funtions more power)


- Also, the AIs do not start wars, if they feel that their units aren't advanced enough and the follow a strategy for getting better units:

The AI's current strategy and financial situation is pretty important for the likelihood of DoWs. Especially the fact that the AI will not even start thinking about any wars when it is following the GetBetterUnitsStrategy might be responsible for the much earlier DoWs on Deity compared to the lower difficulties. The AI needs to be able to build at least 2 offensive units (UNITAIs ATTACK_CITY, COUNTER, RESERVE and PILLAGE) in its capital. Units of the types Axeman, Spearman, Chariot qualify here (basically everything better than Warriors and Archers), so once it manages to hook up the necessary strategic resources it will drop that strategy and start rolling the dice. With the 2 starting settlers and discounts for training more plus the fast teching I think Deities will reach that point 1000 yrs earlier.

I'm not sure if that could be a problem for Planetfall.

 

[Maniac]

- Planetfall AIs seem to be often in great financial troubles. They build a lot units, which they pay support for

Do you think they build more units than is good for them, or they don't build enough energy-producing terrain improvements in general?

IMO, that's even more of an argument to teach the AI using the maintenance process - as that would keep their unit spending up - the unit maintennace still has to be payed, there is just another source of income to cover it!

As far as I can tell, the existing process AI should suffice. Not sure what I could do more. There's already a code which determines the value of a process, and the AI is already encouraged to use processes if they're in financial troubles. Do you know if there's an easy way to see in-game if the AI considers itself in financial trouble?

I think in future games of mine I'll create at the start three permanent alliances with two, two and three members. Will make diplomacy more boring, but that way I can always observe what one of the AIs is doing during a game.

with one exception, its Helion vs. Cyborg - the former is rated lower by 2 despite having better values in everything

Thanks! I fixed it.

 

[Pfeffersack]

Do you think they build more units than is good for them, or they don't build enough energy-producing terrain improvements in general?

Too many units for my taste if they stay in peace. I usually defend my inner bases with one unit. For bases bordering areas with native life or neighbours I can't be sure that peace is kept, I might go up to two or three. I pay for that strategy from time to time, when native life manages to take a base or a sneak attack of the AI succeds initially.
The AIs garrison are usually not below 4 combat units, while bigger cities and the capital often seem to accumulate even more units. The AI needs to build some more units then the human player to make up for its strategic and tactic deficits, but I would be tempted to start a war with so many.
Of course that doesn't make it easier to decide what to do at the moment. Less units would make them probably even less willing to go to war...

For the terrain improvements...I can't see a clear pattern. In the same game there are factions which are e.g. in love with windmills, while others don't bother with them despite being broke. The game I played yesterday is a good example of this: Santiago windmilled the entire land around her capital, while you cannot find a single one in Lal's land. I have attached a save (I played this with an experimental v4 of my leaderhead.xml, which I have attached as well, but I don't think that this makes a huge difference regarding terrain improvements)

As far as I can tell, the existing process AI should suffice. Not sure what I could do more. There's already a code which determines the value of a process, and the AI is already encouraged to use processes if they're in financial troubles.

Thats really strange, as I never witnessed an AI using the process after you have doubled its gain (before that change, I remember at least a few occasions)

Do you know if there's an easy way to see in-game if the AI considers itself in financial trouble?
I think in future games of mine I'll create at the start three permanent alliances with two, two and three members. Will make diplomacy more boring, but that way I can always observe what one of the AIs is doing during a game.

Using debug mode might help. I found a guide how to use it from jdog5000 in the BetterAI forum:

http://forums.civfanatics.com/showthread.php?t=279903

BetterAI adds some functionality to it, so if merge code of it into Planetfall, doing that for the debug stuff might help as well. I think I will give the debug mode a try as well, maybe it helps me with tracking down the strange AI behaviour regarding endless war preperations.

 

[Maniac]

The AIs garrison are usually not below 4 combat units, while bigger cities and the capital often seem to accumulate even more units.

It's possible to modify in the SDK how many units the AI considers an adequate garrison. However AFAIK those numbers currently aren't any different from those in vanilla Civ4.

Thats really strange, as I never witnessed an AI using the process after you have doubled its gain (before that change, I remember at least a few occasions)

I have seen the AI use the Research process recently. The only reason I can think of they don't use Stockpile Energy is because their maintenance costs aren't high enough. But that wouldn't match with the theory that the AI spends too much on unit maintenance.

Did you know btw it's possible to see some information on the AIs using Shift-F2 (when chipotle cheat mode is on I assume)? Among other stuff the unit maintenance it's paying.

 

[Pfeffersack]

I took sometime to examine the save above again, this time with using the possibilities of the debug mode.

I think I have to withdraw my statement on that unit maintennace is killing the AI financially - the screenshots shows that they pay a lot more far city maintenance. So they likely don't go broke because of this (might be a double egded sword nonetheless, as low unit maintenance costs makes them more unlikely to go to war).

So if they go broke, it must have different reasons. In that game, Lal is the best example. He has to run 90% Gold and hardly breaks even. Examining the figures, I don't think his problem is city maintenance as well. The absolue value is comparable to
mine; I have 10, he has 9 cities. So what is killing him? Clearly badly improved terrain - no windmills at all.
You are also right on the process, the AI can use it. The save is likely rather at the beginning of Lal's financial crisis and the AI has just yet not decided to use processing - in a save 20 turns later he runned it in one of his biggest cities.

He still runned 90% Gold, but at a positive cash flow of +21. The maintenance expenses shifted somewhat more to units. The big problem of no enegry generating terrain improvements stayed however.

What might be a problem for the AI is how the strategize their units, though. The table shows that the overwhelming majority of their units is assigned to defense duties. This might be a culprit for the no war problems (though I failed to proof that
so far...I worldbuildered the Gaians tones of units - this time correctly with offensive strategies-, but still couldn't get them into declaring war)

I hope that DanF5771 will take a closer look on my saves, he said that it might be a pathfinding issue:

So maybe it is a pathfinding problem. E.g. the default BTS-AI is not able to figure out that it has to use boats to get the land units to a target in the case that the target is located on the same landmass, but the path is blocked by impassable terrain (only mountain peaks and inland lakes in BTS). During my quick test of Planetfall I popped a Chopper from a Unity Pod and was a bit irritated about its available movement options until I read "Unit may not reveal undiscovered terrain except inside a player's territory" (btw. how can I have fog=undiscovered terrain within my cultural borders?). Furthermore I noticed Armors can't enter Fungus and that there is no direct way from a flat land tile up to a ridge tile. While these are probably pretty cool features I can imagine that they are rather "challenging" for a pathfinding algorithm. I'm not 100% sure, but I guess the individual attack units are grouped together and the AI searches a path for the whole selection group to reach the target. And if it can't find one because some of the units are restricted in their movement options in such ways, the whole group of units just keeps sitting in the city... This could be readily tested by some quick WB-terraforming.

I also worldbuildered Deidre a transport fleet and escort, but that doesn't help as well.

I think I will try gifting units to the other AIs as well, maybe I had just bad luck with Deidre (just seems to the most resonable explanation...if only the unit gifting would help at all...)

 

[Maniac]

Clearly badly improved terrain - no windmills at all.

Actually not only doesn't he have windmills - he hardle has terrain improvements AT ALL.

I haven't looked at your save yet, because I had made some changes to my own Planetfall v8 folder making it incompatible with your save. Meaning I have to reinstall Planetfall to look at your save. I definitely will look at the save though.

What might be a problem for the AI is how the strategize their units, though. The table shows that the overwhelming majority of their units is assigned to defense duties.

We'd need some comparison data with unmodded Civ4 though. How much of their units is the AI using for defensive duties there? Perhaps what we see is in fact normal.

Regarding pathfinding problems. I guess it's a possibility. But I play with Scattered Landing Pods turned OFF. Meaning the AI often starts on one landmass (at least in the next version with a modified starting position generator...), and a path from one civ to another should be assured. Still war declarations are rare.

 

[Maniac]

I have looked at your save now.

The Peacekeepers are the only faction with financial problems. The Spartans are running 50-50, but the other four are running 90% Science! So my first thought would be the AI is having no problems with financial management.

The reason why the Peacekeepers are having problems is because they got caught in the Spiral of Wormy Thought: there are several spawning spots in their territory, and they're not cleaning them up. Furthermore I noticed two formers (which is 40% of their workforce) just sitting in a base doing nothing. The reason for that is AI code: workers withdraw to cities when there is danger nearby, and just sit there until the danger's gone. They never consider moving to another city where there isn't any danger... And since fungal towers are immobile, it can take quite a while before the danger moves away itself. I'll have to modify that worker code, so they consider moving to other cities.

Anyway, I think it's safe to say financial troubles are no reason for no war declarations.

 

[Maniac]

The version of leaderheads included Plf version 9 is your version 3.
Is 4e deemed good to include in the next patch?

 

[Pfeffersack]

The version of leaderheads included Plf version 9 is your version 3.
Is 4e deemed good to include in the next patch?

v3 is the right one. Everything else was experimental with partly crazy and equalized values and didn't improve the situation so far. So as long as we don't now the reason for the few-AI-wars-problem, I think we should stick with v3 for the moment. I also hope for some feedback of the other regular users of Planetfall how the like the new AI personalities overall (even if that is only possible to an extend without war declarations)

BTW, I runned another game with 8d recently, this time examing if less total wars (all leaders to 1/900 for that game) means more limited/dogpile wars and if the instant war declaration after refused tributes works at all. Less total wars reduced the number of AIs in early eternal WHEOOHN mode, but did not lead to more delcarations. If instant war declarations after refused tribute work at all, then not at the percentage of that value (I set 100 for all leaders in that test) and the two or three times I get extorted and refused, nothing happened at all.
I experienced again the phenomen of AIs going overboard with defenders - Deidre managed to get around 40 units in 71 years, of which 31 had a defense AI strategy.
Morgan shows a similar ratio. AIs with less overall units have rather a 50/50 distribution, but it seems that any building up goes towards defense. I'm currently trying to get some comparison values for standard BTS.

 

[Maniac]

Reading the SDK function responsible for war declarations is now somewhere near the top of my to-do list.

Kitsune has already remarked that the AI seems more aggressive, for the record.

What does WHEOOHN stand for by the way??

I can mess with the function that determines how many garrisons the AI considers necessary. Some comparison data with BtS would definitely be appreciated though, both on how many city defense AIs they have, and how much of their maintenance costs are unit costs.

Given that losing a base destroys less buildings, it is a sound strategy to focus less on base garrison and more on other unit AIs. I have no idea right now though if reducing the desired number of base defenses will lead to the AI producing other units, or simply less units. Do you have any idea?

 

[Pfeffersack]

What does WHEOOHN stand for by the way??

Short for "We have enough on our hands now", the AIs explanation when it refues to go war on a players behave (think it applies in inter AI diplomacy as well, though) because of an ongoing war or war preperations.

I can mess with the function that determines how many garrisons the AI considers necessary. Some comparison data with BtS would definitely be appreciated though, both on how many city defense AIs they have, and how much of their maintenance costs are unit costs.

Given that losing a base destroys less buildings, it is a sound strategy to focus less on base garrison and more on other unit AIs. I have no idea right now though if reducing the desired number of base defenses will lead to the AI producing other units, or simply less units. Do you have any idea?

I will hopefully have soon some data from a BTS game.

I have no indepth understanding of the AI process of producing units and then assigning them to a certain strategy, but I imagine it the way, that mainly the value iBuildUnitProb determines the amount produced at all, likely further modified by factors as being at war and similar things. Then the AI likely decides what to do with the unit and assigns a strategy. Given that garrison needs lead them to assign the defensive strategy, a reduce in that need would hopefully free units for offensive duties.

Of course that explanation fits only, if it is possible to assign any combat unit to both offensive and defensive strategies and the AI can decide over this dynamically in the game. I still remember the days of Civ3, were it was preset in the editor, if the AI could use a certain unit for defense, offense or both...