Ultimate AI - Feedback and Discussion *UPDATE 1.6*

[Smarmy]

To the guy above the guy above me:
Too lazy to hit back.. heh.. anyway..

Princess's being snobby is usually because they've never left their birth cities, you don't notice them "grow up" and they fester in the lap of luxury becoming banes on your family. If they're out making agreements and traveling the land they tend to be more charming, seeing the common man where he lives and doing an allbeit haphazard job as diplomats. (I usually don't need or want princess's, I have diplomats for a reason, so I just load all of a generation's daughters onto a boat and send em out to sea. Get more sons that way, too.)

Inactive armies? I can atleast debunk one of the things you mentioned, that army sitting on the bridge to Venice. That's part of the AI's strategy, it's making use of their capital cities' defensive position. A bridge battle is much more favorable than a city wall battle.

Island prison factions?
I have a guess as to why that is, they feel threatened, their protocol for recruitment slams into high-gear for land armies as they are being destroyed. Ships are rather useless in that situation, and sadly in the case of Byzantium with Cyprus and/or Rhodes, as those are both castles he favors them more for unit recruitment. More than likely those are going to be the only castles he's left with in such a situation, and it's 'smarter' to recruit professional soldiers from a castle.

Now, It is also my assumption that he does not attack with those stacks, because he feels threatened from everywhere. An island doesn't have to watch out for one general direction, any enemy could land from anywhere to try and take his island, so I guess couple that with the rest of it and..

As well, he probably doesn't have any money.. but anyway.
Eh edit for wrong thread, brain is still booting up.
Not too big a problem though, Byzantium is really the only one that happens to.
I've got Egypt invading Crete, Rhodes, etc, naval invasions are working.

[GrandViZ]

There is no possibility to check when a war has started. Excommunication is hardcoded, and the way the peace agreements work in M2TW there is little I can do about it. I am working on an update, that will increase the odds for peace between factions of the same religion.

I am aware, that the AI is far from perfect in this game, but I can only work with the tools, that CA has provided - and those are very limited. And I find myself repeating the same explanations over and over again, therefore I suggest, that you first investigate, what commands the AI scripts offer, and see what is possible and what not.

[Dimitri_Harkov]

I will release an update on the campaign AI / diplomacy file, that will make it easier to achieve peace, and reduce the duration of wars.

Very good! I'm looking forward to this.

However, there is one other thing that is far worse: The garrisons are so small that late in the game the AI loses settlements because of rebellions on a regular basis. This does not happen with the vanilla AI (at least not in that extend). Is there a way to improve this?

Other than that, I have to thank you again for the (still) best AI pack around!
Keep up your good work!

[GrandViZ]

There are other factors, that have an influence on the garrison size, such as the recruitment settings in the EDB.txt or the general settings for rebellions in descr_campaign_db.xml, the economic settings for the AI and many more.

With only the AI pack you dont get these changes from my full mod, and therefore you will not get the full potential of the UAI. You would need to change these settings for your favourite mod manually, but this is a lot of work!

[Siffo]

Hi all

First i wanted to say that you have done really great work...i like the mod very much but
i wanted to ask if there is any possibility to tune down the generals' strengh to normal med2 vanilla strengh...those bodyguards + general are just gods and for my opinion i think that is not very realistic.

Greets

Siffo

[Dimitri_Harkov]

The other files that have influence on garrisons are the vanilla ones in my mod, so it could well be that the reason for the weak garrison is the AI pack. On the other hand - it might just be that the pack works much better with your modified files, for which it was sesigned in the first place. I'll have a look at them.

[tree_of_liberty]

Hey there! Your mod single-handedly maintained my interest in M2TW and inspired me to give feedback.

First, a congrats. Vanilla M2TW had such a bad AI, that it got to the point where I was playing VH/VH, and seeing if I could win domination by 50 turns. I've never even seen gunpowder units :/ With UAI, victory is no longer a given. In fact, with some factions, I lose 90% of the time (e.g. Russia, Denmark). But thats a good thing

Secondly, I want to bring up some concerns I've had and also recommend some possible changes.

-Same issue others are having WRT neverending war. The AI doesn't seem to want to make peace. Ever. I've read your claim that # of current opponents is a factor in making the AI seek peace. However, in almost every UAI game, france gets really strong early, gets excommunicated, and then has a crusade called on Toulouse. After being at war for decades, and raped of half their provinces, France is still not willing to make peace, even when I offer them too much.

-Too big of a challenge disparity btw different factions. No matter what I do, even with disabling the settlement tax, I can't do anything as Russia. The settlements they have available for expansion are so low in population and spread out, that they don't provide the necessary income to sustain even a single stack army. Conversely, I always kick ass as Milan.

-AI predictability. I've played early game as Milan five times, and EVERY SINGLE TIME, Venice moves all of its garrison to go siege Zagreb in turns 3-5, essentially handing me their capital city.
Also, some factions always kick ass in a predictable way (France, Byzantines, Turks).

-My BIGGEST issue, imbalanced growth. Did you change the growth rates of cities and castles from vanilla? The reason I ask is that you doubled the number of turns by making each turn 1 year instead of 2. However, without changing the corresponding growth rate, that has in effect doubled what was already exponential growth. This has resulted in unrealistic growth of all settlements. For instance, I'm at turn 60 right now (1140), and I have 5 fortresses, 1 huge city, and 4 large cities (as Milan, i possess all of northern italy, france, belgium, and half of the HRE). One thing your mod taught me was to build farm upgrades first, and prioritize chivalry in most generals (ty, ). Following that strategy though, my capital city, as an example, went from 5000 population at 1180 to 17000 population a scant 60 turns later. That seems a little... extreme. Also, as someone pointed out, the AI gets double the stated growth rate. With their already unnaturally high chivalry governers, that results in their cities mushrooming out of control. I recently captured the Danish castle directly south of their capital, only to find that on turn 60 its population was 18000!!!! HOLY *****. Up from an original 2-3k at turn one. I can only guess that they had it garrisoned with a max chivalry general for the entire time. Needless to say, that seems a little silly.

I would recommend straight-up halving the growth rate of all settlements. One of the things this mod promised was realistic, slow pace of progress. So far I've seen anything but.

A formula for doubling time:
D_t = 70/(growth rate in %)

Since the AI gets double their rate to begin with, and it's not unreasonable to have 3.5% growth given a high chivalry governor, you're looking at 70/7% = 10 years doubling time. Seems a little ridiculous for coming out of the dark ages.

Anyway, I hope you don't take this criticism negatively. It's only out of love for both the game and your mod that I've put so much thought into the issues I'm seeing.

I hope my comments are constructive and will be watching for your response.

-Duke Francesco the Battlefield Tyrant of Milan

[uk_john]

tree_of_liberty - excellent post with everything explained clearly - welcome aboard! We need more posters like you! An early +Rep for you I think!

[smallsam]

hi there... cool mod. here are some bugs:

1) i noticed that after year 1200, even though my economy is generating 9850 worth of money, it is somehow not added to the pot of gold.... from 1200 onwards, my moneys kept getting more and more negative even though if i click on the financial summary, i am generating 9850

2) i am playing england and even up till 1200, i still only have peasants archers??? no longbowmen, no yeoman archers or whatever... why is this the case? i have the various advance knights though, just NO advance archers.

your help appreciated

[Resurrection]

ad 1) I don't know, but for 2) read the readme... (in UAI is implemented era system, when advanced units (high and late) are not available before certain time).

[GrandViZ]

@tree_of_liberty

Thanks for your feedback. It is indeed well thought and constructive! I will investigate every single of your points and try to improve the balance. As a short term solution, playing on hard or normal campaign difficulty will reduce the AI bonus effects, and reduce the overall growth rate as well.

@smallsam
There is an additional growth tax for the human player, that is not displayed in your budget menu (no way to mod that). This additional tax is based on the number and level of the settlements you control.

[smallsam]

hi Grandviz... is there a way to remove the growth tax ourselves?

[uk_john]

smallsam, I am with you, because now the only thing that really affects the gameplay in anegative way is how hard it is to have an economy! It is constantly on a kniofe edge, and to add insult to injury, the AI never seems to have a problem. Whatever a factions wealth, they seem to be able to produce units until the cows come home, and I am struggling with no ability to generate money and with huge costs just to move army units around to defend key points whether at home or abroad!

[GrandViZ]

To disable the additional upkeep script replace the attached file in the folder

..\ultimateAI\data\world\maps\campaign\imperial_campaign

I have also uploaded a fix for other issues / balance requests that were reported before.
You can download it in the bugfix thread, here: http://www.twcenter.net/forums/showp...02&postcount=5

Best regards
GrandViZ

[repman]

I have only two points to complain about UAI 1.6 in DLV:
- I don't see to many naval invasions by the AI.
- If i build permanent forts on defend points on the campaign map, the AI is not populating them.

Both is happening if i use the the AI Medimod from CavalryCmdr
http://rtw.heavengames.com/cgi-bin/f...f=9,,,all&st=0
But this mod has other problems with DLV....

Maybe something to investigate ?

repman

[Darvist]

HI iv been following the AI since 0.9 and so far 1.6 is the best to come. I started playing kingdoms but had to go back to vanilla cause the AI was to easy to fool and beet.

There are only a few things that iv noticed some of witch are just me personal opinion.

-The AI is good. Perhaps to good. Civilizations seem to drop off the map quickly if there out gunned. But its better that way than not good enough.

-Charge is to powerful in my opinion for cavalry. Often the cavalry units particularly in later eras have a powerful attack combined with an almost equal charge bonus can devastate a group of units even Spearmen who have a bonus. Spearmen notably do better but not quit as effective as i would think.

-Generals Bodyguards are to powerful in the early game but not sure if that can be helped without making them to week in the later game.

-Several terns after the Mongols but before the next age since i can't remember the years right now. I usuly have several or at lest a few huge city's. Wile in this case the Turks who has 40 some territory's only had 2. The AI don't seem to be adept a handling large city's population quit right. I got a good view of Sicily who had had the locations for long time and every so often one would revolt. Its not the disease eather. Rome for example should be Huge buy now and they have the money to afford it. Player has a late game advantage being able to boost pop faster buy keeping taxes low and putting high chivalrous generals (easily accomplished but grouping a bunch of family members in a crusading army on a nearby excommunicated factions City that i called).

-Pop growth is very nice. Castles seem to get a bit large 30000+ but they don't revolt like they used to and can be manged. I noticed other faction castles grow to 30000+ or larger size much faster and assume thats dew to the difficulty growth bonuses. I'm just confused as to why that same bonuses doesn't put the city's to the huge size.


-AI is much harder to trick on the battlefield. Will never be perfect but some old cheep tricks don't work like they used to and passive units are fare fare less common. Rebels seem to be the only passive units when engaging like what earlier posts mentioned about just standing there and taking arrows. I think that has more to dew with how one computer handles the battle mechanics rather than another.

-I have a problem with walls. Archers still seem to work but crossbowmen can't seem to shoot from them except for a few who can see in between the gaps but in a row thats like 5 out of 60. Particularly Pavise crossbowmen who also had much difficulty following my instructions to move were i told them to. Mercenary crossbowmen followed movement instructions but still didn't shoot off walls very well.

-A cheep tactic in a siege is to run some cavalry out and bust the siege equipment units (Catapults, Trebuchets) killing them or at lest damaging them. The army then charges the walls with the built siege stuff (battering rams). I have seen them if the catapult units are not killed enough resume lunching them but all ya have to do the do it agen. Cavalry usuly takes some casualty's but unless they have a very large number of archers its less than half the unit. Less for generals and stronger units.

-Income is good but for late game with huge city's popping up it gets kind high. At around 25 territory's most city's are building something continuesly and but 30 32 or so i get 20000 to 40000 every tern i dont know what to do with.

-The three largest factions were the turks, me, and england at 30+ followed buy only Sicily and Mongols at like 10 or lower. The turks used to be huge 42+ at the peek almost wining the game early on if i didn't call a crusade on them to slow them down. Plundered city's along the way back to giving me a huge income boost. The turks were on a rapid decline buy me and England so i was going to see if i could bribe a city (the one on the fare bottom right) but i couldn't (probably to close to family member or there leader) so i made a ridicules demand to become a vassal state. Well to my shock for 72000 for 2 terns they would. This normal? Mind you the diplomat was 1 point from max skill. Combined since vassal kingdoms count toward your owned regions i would win the game. I settled for buying it for 100,000 to help them soften up England after i tested it.

This is from my Spain game witch i had played an earlier game on an older AI that i never finished so i decided to try another one.

Thank you sincerely for this wonderful AI for i would not be playing MTW still if it wasn't for it. And I'm looking foreword to Trying BC when its completed. Looks awesome. I hope something is useful. I'm keeping my fingers crossed for a kingdoms AI in the future.

[Tenom]

I'de love to try out your mod but the download link is broken .

[GrandViZ]

I'de love to try out your mod but the download link is broken .

Try the alternative download link. It works.

@repman
I think the AI in UAI 1.6 doesn't populate forts, because I don't use the defend_fortified option, due to some issues with it in regard to passive AI behavior. I will investigate, if it is possible to improve this aspect and report back.

The same for naval invasions. Although there is no explicit naval invasion command, the AI uses naval invasions only under certain circumstances, and those might occur less in UAI.

@Darvist
Thank you very much for your feedback. As allways I will investigate your issues and if necessary fix them with the next bugfix. And in regard to BC - the BC team is realizing a wonderful project - worth waiting for!

@all
Make sure that you apply the latest bugfixes (if you haven't already).

Best regards
GrandViZ

[repman]

@GrandViz
Populating or creating forts by the AI is a zweischneidiges Sword. If the forts are too much populated the settlements are not garrisoned. If they stay empty because the defend point calculation has changed then they are useless.
Fort-Creation by the AI is controlled by:
- defend_fortified
- and the last lines in config_ai_battle.xml in the part <battle-analyser>: changing values there resulted and different spawning rates of AI forts.

In Medimod 0.9 the AI was creating forts and populating them, in XAI2.19 the AI is creating forts but is abandoning them quickly.

the america problem:
in all AI mods (and vanilla) the AI can't handle the american invasion. No support for the troops over there...inactivity. Probably because of the distance the AI calculations for America are of very low probability. Maybe it would be better to create a special america_invasion AI label to which the faction can switch in case of an invasion attempt.
Something like:
faction_ai_label america_invasion
<decision_entry>
<!--
Victory target
Raw Value = 1500
-->
<min_entry target_faction="aztecs"/>
<max_entry target_faction="aztecs"/>
<faction_attitude invade_priority="1500" continue="true"/>
</decision_entry>

repman

[GrandViZ]

As I said before, the defend_fortified option hurts the AI more than it helps, and therefore IMO it is better to not use it at all. As a consequence of course the AI will not use those forts.

Well, there is no need for a new AI label, which you would have to activate and deactivate by scripts. It is easier to include a faction specific section into, e.g the catholic AI profile, and precisely define the circumstances when a faction should go for the Aztecs.

However it might also be, that due to the pathfinding system (on the campaign map pathfinding search depth is bound to movement range) the AI is unable to calculate a path to the Aztecs at all, but that remains to be tested.