Dominions 2. AI. Suggestions, that how to fix it. [Archive]

licker

October 28th, 2003, 07:24 PM

There's nothing wrong with several smaller armies running around causing havoc generally...

The problem arises when these smaller armies are faced with a few different obsticales. The most obvious is when they hit a choke point, at that point they need to aggregate to take it, or find something else to do. Its also a problem when they suffer casualties becasue they are not large enough to overwhelm their weaker enemies, thats a matter of attrition. Even though they win, they get weakened, while the larger concentrated force loses less. Eventually this is a big problem.

Speculating here... but if the AI is going for a rapid conquest of neighbors (indies mostly in the first 20 turns) but loses more than it gains back in terms of income and production advantage from conquering more provinces quickly, then when it runs up against the human who did a slower expansion with more concnetrated forces, it will lose, as even if it has more income and production, it does not have the time to use them to recoup its initial losses. This of course is something that is difficult to account for from an AI perspective, but it could be an element in why some players are seeing AIs with smaller unit totals when they meet. On a bigger map with fewer AIs the rapid expansion should be more viable, on a cramped map, the concentration of units to minimize battle losses should be more viable.

I've got no idea how the AI handles different situations, or if it really does try different things for different situations. Some of these problems can probably be mitigated by the start up selections (as noted in other places) but maybe some more attention to how the AI responds to different starting conditions is appropriate.

There is also nothing really wrong with an AI having some single mindedness, like I'm gonna pull all my units into one army and charge your capital. Either it makes it or it doesn't (game over for someone), but at least it gives the apperance of making a bold move. From the demo I'd imagine its impossible to tell if the AI does resort to this tactic as there isn't enough time to make enough units to make it appealing (probably).

MStavros

October 28th, 2003, 07:26 PM

licker you had good ideas there.

My main problem is, that I never ever seen a bigger AI army within these 40 turns. I seen more, small armies, and those are very ineffective, you can crush them one by one without a single problem.......

[ October 28, 2003, 18:27: Message edited by: MStavros ]

ywl

October 28th, 2003, 07:27 PM

Originally posted by MStavros:
1.) I understand it, but it seems to me, that it don't have enough units..well this is maybe it is trying to make more and more small armies? I have no idea, but I had the feeling that it never had enough troops.....

2.) Kris, I think that the AI should make at least 1 bigger army in the early game. It is too easy to kill the AI's small armies 1 by 1.

3.) I meant that the provinces of the AI are usually empty.

4.) Well examples: trying to attack your undefended side, trying to focus his little armies on one point, make at least 1 big main army etc. 1) I've seen very large AI armies in my few test games - e.g., Pan, a lot of undeads.

3) It's a good move. You don't and shouldn't play thd game by scattering your force to "protect" individual provinces. Your army will be eaten piece by piece.

Usually, you would rather concentrate and hide your mainforce, releasing them only when the fight breaks out. If you need to protect key provinces, you build fortress and have some defensive force (even cheap units are enough) behind the wall to defense against a seige and wait for rescue.

4) It's hard enough to teach human good strategic moves - not to mention AIs.

Can you summarize them in a few clear rules for the program - better yet, an algorithm? For example, before major war, split your army to expand. And once under attack, concentrate them and go all the way to your enemy's capital. But I guess some of these have been implemented already.

Kristoffer O

October 28th, 2003, 07:29 PM

Originally posted by MStavros:
1.) I understand it, but it seems to me, that it don't have enough units..well this is maybe it is trying to make more and more small armies? I have no idea, but I had the feeling that it never had enough troops.....

2.) Kris, I think that the AI should make at least 1 bigger army in the early game. It is too easy to kill the AI's small armies 1 by 1.

3.) I meant that the provinces of the AI are usually empty.

4.) Well examples: trying to attack your undefended side, trying to focus his little armies on one point, make at least 1 big main army etc. One larger army might be an idea.

I believe that they are empty because the armies are occupied elsewhere. I and JK rarely keep defending armies at peaceful borders unless we expect an attack. The AI probably shares our motivations to some extent. The problem is that when armies arrives at his borders the AIs armies are probably busy conquering other lands. As in most games attacking strongly and unexpectedly is a winner (MP or SP). Then your opponent cannot strike back in time. If the AI attacks you first I believe it would be another matter (but on the other hand I do not now this was not the case).

The AI avoids provinces that are well defended and usually goes for weaker ones.

One thing it doesn't do (unfortunately) is trap humans. Nothing is more satisfying than luring your MP enemy into a province where you can attack him from several directions.

Saber Cherry

October 28th, 2003, 07:36 PM

I know this is not really an "AI" solution, but it might be a good idea to scale the AI's starting army size with indy strength, so they won't get into long-term trouble if they make a bad strategic choice on turn 1. In other words, start them with 25% more units per level of indy strength - +25% units at 4, +50% at 5, and so on. Eventually I suppose they would need an extra starting commander to control all the extra starting troops.

-Cherry

Gandalf Parker

October 28th, 2003, 07:39 PM

I find the AIs doing better on games where the indeps are set to about 7. On 2 I can use my human flexibility to get to them too quickly and make maximum use of the situation.

I think the AI for an indep 2 game would have to operate very differently. Such as investing in a much more defensive position much earlier in the game.

LordArioch

October 28th, 2003, 07:44 PM

I've found the ai rather good so far...more or less like the dom1 but with some human exploits like patrolling removed. I don't see the normal AI beating me but it at least offers a threat...and the impossible one is probably quite tricky.

Sure beats the Age of Wonders AI http://forum.shrapnelgames.com/images/icons/tongue.gif
Or that of a number of games i could list.

licker

October 28th, 2003, 07:44 PM

How does the AI respond to the intricacies of the strategic map? Do they recognize choke points and make early attempts to gain them? Do they account for fronts and important provinces (ones that leed to many other provinces)?

Once the AI controls a strategic point do they prioritize its defense? Do they look at the list of controled provinces and have a defenseive ranking?

Just questions while I think on this issue...

MStavros

October 28th, 2003, 07:45 PM

Hm yeah I understand the problem with the empty AI provinces. You cannot defend everything.

Yeah indy str 7+ and you cannot conquer that fast.

Once again I think the biggest problem is that the AI won't make bigger armies. I never seen one at least so far. Maybe I am crushing the AIs too fast? I dont know.
More, small armies are very ineffective, this is the problem IMHO.
+ The AI cant be bothered if you kill his little armies one by one. I never seen more small armies in the same province, also I never seen an invading, bigger attacking force so far.

[ October 28, 2003, 18:46: Message edited by: MStavros ]

Saber Cherry

October 28th, 2003, 08:04 PM

Originally posted by MStavros:
[QB] Once again I think the biggest problem is that the AI won't make bigger armies. Maybe they finally understand supply rules? http://forum.shrapnelgames.com/images/icons/confused.gif

I have yet to encounter an AI, so I don't know=)

-Storm-

October 28th, 2003, 08:30 PM

Yup, I had the same experience. Well 40 turns is nothing, but still I didn't met with any "bigger" armies controlled by the AI.

Mortifer

October 28th, 2003, 09:05 PM

I didn't played too many games, but mostly Ive seen small enemy (AI) armies. I met with a mediocre army 1 time, when I attacked a home province.
Sadly the AI isnt that good as I expected.

The battle AI + spellcasting AI is excellent. I am not sure, that why, but the strategical AI is lot worse. http://forum.shrapnelgames.com/images/icons/icon9.gif

Pocus

October 28th, 2003, 10:23 PM

The AI is too contrived most of the time. On bigger maps (speaking of doms I here), an AI would emerge (Ermor often) and roll over the opposition. Then you would face it.

ywl

October 28th, 2003, 10:58 PM

I saw army big enough for early game (at least 50 to 60 units) many times. Overall, the strategic AI behaves quite similar to Dom 1.

I have several observations:

1) AI tends to have a large number of crappy units, whereas human will go for quality (knights, emerald guards) whenever practical.

2) When you're seiging the AI capital, they'll keep trying to retake it. It's a sensible thing to do but being so predictable will make it easy to exploit. It doesn't hurt to add a degree of randomness to it.

3) Even without the tax-n-patrol trick, I can still expand faster than the AI. Do you guys have similar experience? Is there any way to improve the situation? We all know that if you lost the open-game, it's very hard to catch up.

4) Are there any nasty tricks hard-coded for the AI? Any experienced human players would have a wide arsenal of them, the AI need some to be competitive.

MStavros

October 28th, 2003, 11:03 PM

Indy strength on 7 won't help at all. I had a game right now. 40 turns was not enough to kill the AI, but I almost did it.....It had like 2-3 provinces left, when the 40 turn limit ended my game.
I didnt met with a decent AI army, I only fought against some small forces...

Mortifer is right. This is very frustrating. The spellcasting AI is so cool, I wonder why is the strategic AI so weak?

MStavros

October 28th, 2003, 11:06 PM

Originally posted by ywl:
I saw army big enough for early game (at least 50 to 60 units) many times. Overall, the strategic AI behaves quite similar to Dom 1.

I have several observations:

1) AI tends to have a large number of crappy units, whereas human will go for quality (knights, emerald guards) whenever practical.

2) When you're seiging the AI capital, they'll keep trying to retake it. It's a sensible thing to do but being so predictable will make it easy to exploit. It doesn't hurt to add a degree of randomness to it.

3) Even without the tax-n-patrol trick, I can still expand faster than the AI. Do you guys have similar experience? Is there any way to improve the situation? We all know that if you lost the open-game, it's very hard to catch up.

4) Are there any nasty tricks hard-coded for the AI? Any experienced human players would have a wide arsenal of them, the AI need some to be competitive. Yeah the AI either have small forces or mediocre army with bad unit/commander combinations, IE weak units etc.
There must be a way to improve this...

Oh just a little addition. I think that the research AI is good. Usually the AI is doing good in research as I see it from the graphs.

So battle AI, spellcasting AI, research AI seems ok, but the strategic AI is too weak in many ways.

[ October 28, 2003, 22:07: Message edited by: MStavros ]

Particle

October 28th, 2003, 11:18 PM

Originally posted by MStavros:

So battle AI, spellcasting AI, research AI seems ok, but the strategic AI is too weak in many ways. I agree with this statement....

->Hopefully Illwinter will upgrade the strategic AI somehow. http://forum.shrapnelgames.com/images/icons/icon6.gif

If you ask me...if the strategic AI would be as good, as the spellcasting AI....the AI would kick the you know what of the human players. http://forum.shrapnelgames.com/images/icons/icon10.gif

[ October 28, 2003, 22:19: Message edited by: Particle ]

LordArioch

October 28th, 2003, 11:56 PM

The AI assasinated one of my researching mages...that was pretty annoying.

Gandalf Parker

October 29th, 2003, 01:33 AM

Do you know how to psuedo code or basic? Not really programming but writing out a set of instructions. What kinds of things would you add? Im not trying to be funny its just that for me this would be where I wouild go next after saying the AI needs some smart code.

If (indep less than 5) or (mapProvinces less than 140) then
spend income/2 on PatrollingUnit
endif

If (PatrollingUnit = MaxAssigned) then
go forth as large army
begin new Patrolling Unit
endif

If (neighbor not owned by me) and (neighbor not owned by indep) then
purchase elite troops
endif

[ October 29, 2003, 00:34: Message edited by: Gandalf Parker ]

LordArioch

October 29th, 2003, 02:44 AM

I'm not to be trusted near code...but I find the AI should be perhaps more agressive...I have yet to see a computer player take the initive in attacking me so I can deal with them one at a time. If they were more agressive I'd definitely be losing the resultant multi-front war. http://forum.shrapnelgames.com/images/icons/icon7.gif

Nerfix

October 29th, 2003, 07:14 AM

BTW i just saw AI having army of 110 units... I have also seen armies of 30-50 units several times. The AI also uses province defence a lot. The Machaka AI i played against used blitz forces made out of Spider Riders, Spider Knights and Black Hunters.

[ October 29, 2003, 06:16: Message edited by: Nerfix ]

Saber Cherry

October 29th, 2003, 08:20 AM

Yeah... I'm playing Machaka, with all races on "difficult". Ctis keeps trouncing me, attacking with... oh, around 80 units, on turn 10. This is with default settings (except indies are 5). I keep slaughtering them, but their armies still grow bigger every turn.

I keep finding that gold is a precious commodity in this game! Wow... especially for machaka. Oh, and sacred units are hard to come by, as well. I kind of hoped all Machaka's spiders would be sacred...

-Cherry

Pocus

October 29th, 2003, 09:49 AM

a poster got a very valid point about the AI : the AI dont go enough for quality. This is why we see so many indeps units in the AI armies.

I think the AI algorithm should be refined on what to buy. An ideal proportion of weak troops should be fixed, and recruitment in these weak troops should be stopped when the % is reached.

You can then refine the system by changing the proportion depending on what type of AI level, the type of map, or nation the AI play, but as of now, here is a raw first implementation :

troop quality (TQ) : gold + resource cost

National average of TQ : for each unit (not leader!) type in the national roster, compute TQ. Average it. It gives the average TQ of the nation (Ulm would have a higher TQ than say Ctis IMO). Lets call it NATQ

ideal TQ proportion to be reached:
TQ inf 0.5 NATQ : 20%
0.5 NATQ inf TQ inf 0.9 NATQ : 20 %
0.9 NATQ inf TQ inf 1.3 NATQ : 40%
1.3 NATQ inf TQ : 20%

have to use inf for the inferior sign, as otherwise the post decipher it as an HTML flag http://forum.shrapnelgames.com/images/icons/icon9.gif

Troop recruitment:
order of province browsing: to be determined smartly. I can post an algo, but this is not my point here, and it have to include a bunch of params.

so ...

get a sorted list of province where the recruitment will take place.
sort each type of troop available, from highest TQ to lowest. if troop Category is in a TQ which is undermanned, recruit 1. Repeat until no more gold, resource, or TQ is ok. Proceed to the next troop type.
Then proceed to the next province.

improvment :
overrides (partial or complete) should be used if the province has a particularly high interest in the 'province which will undertake recruitment list'.

this is very crude. You can refine by splitting desired proportion between main line units, missiles, cavalry, etc.

Pocus

October 29th, 2003, 10:12 AM

forgot something of importance:

Some of you wonder why the tac AI is so good, whereas the strat AI is average.

The main difference lie in a word 'cooperation'. Cooperating mechanisms in AI coding is one of the hardest thing to do well. It is even more difficult that learning behaviors.

A given leader/unit in the tac AI dont cooperate that much with the other. Sure, it evaluate that casting poison clouds everywhere if your army is not poison warded is detrimental, but a simple fitness function can tell him that. The tac Ai, for spell casting for example, browse all the spells at his disposal, and get a numerical appreciation on how the spell would be of interest.
With that in mind, with some serious tweaking and work, you can get a good AI for your mages.

The situation is completely different and much more complex for the strat AI. You cant simply tell him to start with biggest stack, seek the best objective for this army, then proceed to the next. I wont discuss further the issue, but believe me the devs made already a very good job with the dom AI.

so much for the 'why dont they do a strat AI as good as the tac AI' http://forum.shrapnelgames.com/images/icons/icon12.gif

Aristoteles

October 29th, 2003, 10:18 AM

Yeah the AI should be more aggressive if you are at war with him, also the AI is mainly using cheap - weak units.

Just a question: Can't you guys contact with Adam West from Crosscut about scripting? Maybe he could help you scripting a better strategical AI, if hes got enough time for that.
Just a suggestion..

MStavros

October 29th, 2003, 03:56 PM

Yeah, if the indy setting is 7 or more, nothing will happen. The only difference is, that you cannot expand that fast.

I think we mentioned the biggest weaknesses of the strategic AI. It is up for the devs now, to upgrade it.

Oh and Kris, if you had some games against the AI, please post your opinion.

MythicalMino

October 29th, 2003, 04:04 PM

ok, just won my first game...I was Pythium...2 AI's (Marignon and Jotun)....

Throughout the game, Jotun pretty much had the lead in army size and provinces...Marignon just never really got off the ground...I led the entire game in research, something I have NEVER done in Dom1....

It took about 20 or so turns to finally meet up with the Ai (Jotun)...Marignon soon followed....Marignon and Jotun seemed to be close, so perhaps they were fighting quite a bit, and that could be why Marignon just didn't do much.

Oh yeah, game settings: default indeps setting, Victory Points turned on (10 points wins the game)

Marignon, I never seen much of an army from them....Jotun, I seen several 50+ strong armies, BUT, mostly Militia. Some were Barbarian Horsemen, Barbarians, and the like...but, mostly militia. I did see several mages/priests...now i am not sure I remember.

Perhaps Jotum would have given me some trouble, but I went for an all or nothing push to the final VP's at turn 33. They were beginning to hit me in 2 different spots at that time, taking my provinces, and re-taking theirs that I took from them, but, it was too little too late. I had gotten close enough to that final 3 point Province, which was VERY poorly defended I thought for being one of the highest VP spots on the map.

I did see Jotun split a 60 man army, and start to attack in two different directions on one of my borders, but again, too little too late.

I am going to try bumping up the indeps a bit, maybe go up to 5 or 6...also perhaps play with 5 AI's this next game.

Anyways, that was my battle report....oh, I am kinda thinking that maybe the VP's kinda screws up the AI too....Jotun just didn't seem too concerned with that 3 point province they had...I boosted up the defense level up high, and built fortresses in my higher point provinces...

Oh well...will let you know about the next game when it is finished or time runs out....

Kristoffer O

October 29th, 2003, 04:07 PM

Originally posted by MStavros:
Oh and Kris, if you had some games against the AI, please post your opinion. I will be away for a couple of days, but I'll be back sunday.

The main issue is that humans use very focused strategies, compared to the computer that tends to dabble in a bit of everything.

Sadly also (just like Dom I) it looks like the computer doesn't use it's pretender for anything beyond sitting in the capital doing research, casting rituals, and making magic items. It doesn't use the predtender for combat advantage the way a human would. As far as I can tell, it also doesn't use the pretender for site searching (in person).

On the impossible level Illwinter should have hardcoded more focused behavior. Only allowing certain units, certain research, and certain spells to cast. It wouldn't be as good as a human, but could mimic some human behavior.

Hey IW- If you guys want I could come up with a few simple rules for each nation. Perhaps you could code that into a new level of AI. Feel free to post them or send them to me. I can't promise that it will be implemented, but if and when we adress the AI it can be useful.

MStavros

October 29th, 2003, 04:10 PM

Helpful post cpbeller.
I had the same experience. 50-60 units in an AI army *or less*, and mostly cheap, weak units.

LordArioch

October 29th, 2003, 04:52 PM

Are any of you people that find the AI easy playing with the suggested number of AI? It's quite possible to outexpand them if the entire map is empty space because I think they tend to fight among themselves more than against you.
Try playing the maps with 6+ AI and not 2...see if that changes things any.

Mortifer

October 29th, 2003, 04:58 PM

I had various games with 2-6 AIs. With 6 AIs, the challenge is bigger, but the AI is not better.
I think the AI's focusing on cheap units. This is very weird. I barely met with stronger enemy units, or if yes, they were small in numbers.
Right now it is way to easy to beat the AI, since strategically it is weak.
It is making maneuvers with more - small armies usually + those armies are built up from weak troops. = no real challenge for a strong human 'campaign army'.

MythicalMino

October 29th, 2003, 05:15 PM

ok, will try the recommended amount of players (AI's) in the next game i set up...which, i am getting ready to get it started in just a bit...

Will post how that one is going....

Mortifer

October 29th, 2003, 05:16 PM

Hm maybe we should send in saved games?

licker

October 29th, 2003, 05:24 PM

Simple AI rules eh?

Well I'm sure there are others who are better equiped to deal with specifics, but I'll spew some thoughts out just in the hopes to get the ball rolling.

It seems to me that perhaps the AIs are not focused enough, maybe they have too many inputs to weigh and consider, and those inputs are not yet balanced correctly, or perhaps even fataly flawed due to slight game mechanic tweeks (such as dealing with supply better, and worse).

With that in mind I'd like to see a set of AI styles that focus the AI in one (or maybe a few) direction for the entire game.

Examples of these styles can be picked up from other TBS games, AoW1 and MoM had settings for the AI to follow (expander, conquerer, defender, ...) so perhaps it is posible to give these styles to the different AI nations in a game, and let them try to focus their energy in specific directions (it would be interesting to let the player even pick the style, or leave it as random). If it is randomized each nation should probably have weighted choices as certain styles will better fit certain nations.

Ok so now we have a rough framework from which to work with. Lets try some specifics.

Firstly the style has to be picked (or incorperated) from the instant the AI designs the nation. Lets make two broad styles with several sub-styles possible within them. The two broad catagories can be Might and Magic ( http://forum.shrapnelgames.com/images/icons/tongue.gif ). A nation with a Might style would focus on creating armies based on strong recruited units (not militia and LI) a nation with a Magic style would focus on creating armies with mages and or summons, not relying as much on the stronger (more costly) recruitable units. Obviously there would be some overlap between the styles.

Sub-Styles could be things like, expander, conquerer, defender, builder (economizer?), researcher, ...

An expander would attempt to expand its boarders through indies, trying to avoid early national conflicts, a conquerer would expand by starteing early national conflicts, a defender would expand, and try to build up more internal defense before attacking new provinces, a builder would be similar to a defender I guess or you could make it an artificier, a researcher would devote more resources to mages and getting spell levels quickly. Furter distinctions within these sub-styles could be based on how many resources are spent on buffing commanders vs. summoning vs. more units.

The main idea is to give each AI a path for it to follow, rather than allowing it to dabble a little in everything (which is rarely going to make it successful). This doesn't go into strategic elements for actual game play, but others can work on those issues better than I can. One thing that the AI needs is the ability to use powerful combos. In fact the styles can be made such that they aim toward these powerful combos.

licker

October 29th, 2003, 05:30 PM

One further note I forgot to add. The AI must be capeable of switching styles (though not easilly) if the circumstances merit. Say the AI was origionally a magic researcher, but the first 5 provinces it conquered were very resource rich and site poor, there need to be some mechanism by whcih the AI can switch to the more appropriate Might and conquerer strategy. I'd say that the limit to make that decision should be for every 5 provinces (or a scaled number based on map size) that the AI gains or loses. However, at some point further restrictions should be placed on changing a style, as once enough headway has been made with a style it is probably a bad idea to change it (at least broadly).

Mortifer

October 29th, 2003, 06:00 PM

Mmmmmmmmmmm interesting idea. So you say that the devs should script various AI types, and let the AI decide that when to change it during the game?
This sounds like separating AI scripts! Sounds cool!

licker

October 29th, 2003, 06:18 PM

Originally posted by Mortifer:
Mmmmmmmmmmm interesting idea. So you say that the devs should script various AI types, and let the AI decide that when to change it during the game?
This sounds like separating AI scripts! Sounds cool! Heh, it does sound cool, but for all I know the AI already does this type of thing http://forum.shrapnelgames.com/images/icons/icon7.gif

If that's the case then it would be nice to learn what the inputs to the AIs decision making process are so that they can be tweeked to make it better respond to various situations.

I'm no programer or AIist though, I hope that those who know more about his type of thing will wiegh in with their comments and suggestions, but I do feel that the AI should have a strict frame of reference within which to work, that will keep it on track for whatever its goal is, and not lead to as many situations where it tries to do too many things.

Mortifer

October 29th, 2003, 06:23 PM

Gotcha. If you ask me this is an awesome idea. Different AI behaviors [main scripts], all connected with a 'central' script. If the situation changes, the central script will change the main script when it will be needed.
IE.
Main Script 1: Conquering indies
Main Script 2: Research
Main Script 3: Forge Items
Main Script 4: Recruiting armies
Main Script 5: Strategical moves
etc.
etc.
All this commanded by a central script.

Seriously I love this idea, I wonder what is the devs opinion about this.

[ October 29, 2003, 17:24: Message edited by: Mortifer ]

licker

October 29th, 2003, 06:32 PM

Yeah, your way of looking at it is probably a little bit better than what I described. Each style will have all the sub-style included in it, but each sub-style will have a weight attached to it. Those weights are what will change based on the inputs.

There still needs to be an overriding goal though, victory through might, through magic, through dominion (blessings), ... then the weights for recruitment, research, artificing, ... are doled out. Further each area would have sub areas depending on the top style, so recruitment would have heavy, medium, light, research would have the different schools weighted (or just scripted for a particular spell or combo), artificing also by gem income and combos...

This still doesn't adress the strategic moves made, but it does give a better (maybe?) framework from which the AI can make those moves. Without the right resources and backup no amount of brialliance on the map will help the AI become competative with humans (unless you start giving it massive cheats...)

ywl

October 29th, 2003, 07:02 PM

Some ways that I think will improve AI's skill:

1) As other people mentioned, get better units other than a large number of miserable militia or slingers. This will probably help a lot.

2) Use human pre-designed pretenders. AI use insensible pretenders. We can easily make a wide variety of pretenders, catered to the play style of AI.

3) Teach it how to use super-combatants. Among various human tricks, this one seems to be the easiest to teach AI. Put a few items on a big creature and send it out. This, combined with human-design pretenders might help AI use their pretenders more effectively.

4) Castle? I don't think AI build castles on important provinces.

5) Other tricks? For example, it's quite easy to tell human to aim for "Storm" & "Wrathful Sky" combo. How easy is it to teach AI to aim particular spells like these?

There are strategies that are simple but very effective. For example, a Jotun with a simple Wyrm and maximum scales, keep building Woodmens, Seithkona and aim for Evocation 7. Another example is to use a basic Prince of Death for Ulm, aim for "Boots of the Behomoth". Put the heavy Ulm armors and the "Boots" on the PoD and send him out.

Are there any ways to script such strategies for AIs? Or should we start designing a scripting language for Dom AIs http://forum.shrapnelgames.com/images/icons/icon7.gif ?

DominionsFan

October 29th, 2003, 09:47 PM

WOW guys, these are cool ideas! I bet that IW will like them! http://forum.shrapnelgames.com/images/icons/icon6.gif

Zerger

October 29th, 2003, 09:56 PM

Good suggestions...I wonder...how is the strategic AI works? Any devs can give us some infos?

licker

October 29th, 2003, 10:14 PM

Originally posted by Zerger:
Good suggestions...I wonder...how is the strategic AI works? Any devs can give us some infos? Yes it would be good to get some Dev feedback on this issue. Many of the suggestions made may in fact be impractical with the current implementation. I'm hopeful though that scripts do exist that are triggered by various inputs. In that case it should be easier to both modify the scripts (and add new ones) and tweek the inputs to get a better focused AI.

At the least I hope the AI is using some killer combos, if not I hope that ability can be scripted in for them http://forum.shrapnelgames.com/images/icons/icon7.gif

MStavros

October 29th, 2003, 11:02 PM

Yeah some dev feedback would be nice.

One more thing: in my games the AI never built any forts! Is this normal? [I can see this from the graphs]

-Storm-

October 30th, 2003, 01:21 AM

Originally posted by Pocus:
forgot something of importance:

Some of you wonder why the tac AI is so good, whereas the strat AI is average.

The main difference lie in a word 'cooperation'. Cooperating mechanisms in AI coding is one of the hardest thing to do well. It is even more difficult that learning behaviors.

A given leader/unit in the tac AI dont cooperate that much with the other. Sure, it evaluate that casting poison clouds everywhere if your army is not poison warded is detrimental, but a simple fitness function can tell him that. The tac Ai, for spell casting for example, browse all the spells at his disposal, and get a numerical appreciation on how the spell would be of interest.
With that in mind, with some serious tweaking and work, you can get a good AI for your mages.

The situation is completely different and much more complex for the strat AI. You cant simply tell him to start with biggest stack, seek the best objective for this army, then proceed to the next. I wont discuss further the issue, but believe me the devs made already a very good job with the dom AI.

so much for the 'why dont they do a strat AI as good as the tac AI' http://forum.shrapnelgames.com/images/icons/icon12.gif That is true.
Anyways, the devs will figure out a way to improve the strategic AI, I am sure about that.

Zerger

October 30th, 2003, 01:52 AM

Originally posted by MStavros:
Okay I decided to make this topic, since the AI is very weak.

As I see, the biggest problems are, that the AI:

1.) isnt recruiting enough units.
2.) is trying to operate with small 'blitz' armies, I never ever seen a bigger AI army in the demo so far.
3.) isnt protecting its provinces very well
4.} isnt making smart strategical moves

Tell me what you think. I think these are the major problems with the Doms 2. AI. 1 - I agree. However some AIs had nice amount of units in some of my games, but all of those were weak units. They were toast against my heavyweight army.

2. - Very true

3. - Hm I don't agree with this. This part is working well. When I leave a province with a small defense force only, the AI always trying to conquer it with a small army, and he is getting it almost every time.

4. - I agree. One of the biggest problem is, that the AI has no counter for any sort of
efficient strategy.

2 main problems with the AI:
* The AI won't recruit an effective army. Mostly it is trying to attack with tiny - mediocre armies.

** The AI has no counter for any sort of
efficient strategy.

Setting indy str to 7++. It helps a bit, that will slow down the human player, but it also slow down the AI. Also I noticed that when I set it to 7, 8 or 9, and I reach the AI, its hellish easy to conquer the AI's provinces, since the AI had killed the indy armies, and propably the forces of the AI are almost totally dead when I reach the freshly conquered provinces.

===>> This won't help at all in anything.

[ October 29, 2003, 12:54: Message edited by: Zerger ]

apoger

October 30th, 2003, 01:58 AM

The main issue is that humans use very focused strategies, compared to the computer that tends to dabble in a bit of everything.

Sadly also (just like Dom I) it looks like the computer doesn't use it's pretender for anything beyond sitting in the capital doing research, casting rituals, and making magic items. It doesn't use the predtender for combat advantage the way a human would. As far as I can tell, it also doesn't use the pretender for site searching (in person).

On the impossible level Illwinter should have hardcoded more focused behavior. Only allowing certain units, certain research, and certain spells to cast. It wouldn't be as good as a human, but could mimic some human behavior.

Hey IW- If you guys want I could come up with a few simple rules for each nation. Perhaps you could code that into a new level of AI.

[ October 29, 2003, 13:06: Message edited by: apoger ]

ywl

October 30th, 2003, 04:16 PM

A scripting language which can handle the game situation is nice on paper. But it's a tremendous task - that's why I put a http://forum.shrapnelgames.com/images/icons/icon7.gif on my comment. So, unless the developers say that they're willing to work on it, or somebody could send them a language spec and preferrable a compiler, it'll be more or less remain just an idea.

It might be fun and easy enough to give the AI different preference to exploration, conquering, etc. But that will only change the weakpoint AI in strategy to something else http://forum.shrapnelgames.com/images/icons/icon7.gif . It wouldn't really solve the major issue here.

Overall, I don't think it's even realistic to expect an AI can play such a complicated game well. But some simple fixes should be able to make the AI more challenging - and we have proposed some.

licker

October 30th, 2003, 04:20 PM

Originally posted by Gandalf Parker:
The area of unit selection is really hard to move from complaining to real suggestions. To keep a database of each race and each gaming situation would be extreme. They might as well move the AI to a seperate program so it doesnt slow down the MP play. Many of the people complaining about the AI are playing with independents set at 2 or 3. The strategies and units for such a game would be very different than if it were set in the middle, and that would be different than the high setting. And thats just ONE of the pre-game settings.

OK Ive heard that the AI should not buy cheap units but instead buy a "real army". So insteadof figuring out how to get the most units with the least upkeep (patrolling strtegy) they should buy the most expensive? That would lead to armys of hvy cavalry, which I guess isnt too bad.

Personally I think an AI best bet is to buy everything. Start at the cheapest and buy 1, move up thru the choices then start over. They would reach a point of not being able to buy another hvy cavalry or elephant or hydra and would use up the left over amount on cheap troops so they would end up with slightly more lesser troops than expensive ones. But that wouldnt be too bad.

What about leaders? How do you make sure the AI buys a mix of leaders? It has no memory so it cant say "I bought this Last time and now need to buy this". Hmmmm well actually it might be able to fake that. If turn number divides by 5 evenly then buy the most expensive leader THEN buy troops. *sigh* still not great but maybe it would be better? Certainly the AI can look at its army composition at the start of its turn (units and leaders) and build appropriately no? Say it sees a deficiency in heavy troops, or mages, it can start building those troops until its reached its equilibrium level. This is sort of the approach MoO3 used actually for fleet compostion, it was moddable and eventually the modders were able to 'force' the AI to build more intelligent fleets by changing the ratios of scouts to beams to missiles...

I don't see why this wouldn't work for Dom as well. You'd have a list of commander types (maybe also broken down into magic types for different mages) and a list of unit types (classes rather than specific units so indies fit in). Then a weight is assigned to each type, a crude example for just infantry could be:

LI: 25%
MI: 50%
HI: 25%

So the AI would look at its current army and see that it had only 10% LI, then it would build more LI to get the ratios set. Something like that should be made external so that modders can tweek away to their hearts content. In fact you can use lists like that to control many aspects of AI choices, you can do it for magic, artificing, recruiting, summoning, ... and have master matricies to control the embedded ones...

That's what I'd like to see anyway http://forum.shrapnelgames.com/images/icons/icon7.gif

licker

October 30th, 2003, 04:37 PM

Originally posted by ywl:
A scripting language which can handle the game situation is nice on paper. But it's a tremendous task - that's why I put a http://forum.shrapnelgames.com/images/icons/icon7.gif on my comment. So, unless the developers say that they're willing to work on it, or somebody could send them a language spec and preferrable a compiler, it'll be more or less remain just an idea.

It might be fun and easy enough to give the AI different preference to exploration, conquering, etc. But that will only change the weakpoint AI in strategy to something else http://forum.shrapnelgames.com/images/icons/icon7.gif . It wouldn't really solve the major issue here.

Overall, I don't think it's even realistic to expect an AI can play such a complicated game well. But some simple fixes should be able to make the AI more challenging - and we have proposed some. It would change the AI weakpoint to what exactly? Certainly if you think the AI is weak in many areas, improving some of them is worthwhile, it doesn't make the unimproved areas any weaker, other than on a reletive scale.

The easist thing to fix in the strategic AI is probably the unit balance within a nation. I would guess that if that were fixed (or made moddible to let the modders play around with it) other defiencies would be less obvious since the main hanicap of not fielding competative armies would be gone.

I'm sure it's a deeper issue than we touch on though, only the Devs can tell us what is realistically possible. We are just presenting different ways to look at the problem. (I hope http://forum.shrapnelgames.com/images/icons/icon7.gif )

Mortifer

October 30th, 2003, 06:13 PM

Well if you ask me, separating various AI scripts sounds very good, however it would work? I have no idea, I am not an AI scripter.
The devs will know that this is a good idea or not.

Once again, the first and biggest problems are that the AI tends to use weak or mediocre units, also bigger AI armies are very rare. (in the demo)

I have a question: the AI is casting rituals in a proper way? IE. Is it summoning creatures and enlist them in his armies? Perhaps Beta testers could answer this?
(I barely saw summoned creatures in the AI armies in the demo)

-Storm-

October 30th, 2003, 08:55 PM

Hey, maybe JK should comment these AI suggestions. He coded the AI... http://forum.shrapnelgames.com/images/icons/tongue.gif

licker

October 30th, 2003, 09:21 PM

Originally posted by -Storm-:
Hey, maybe JK should comment these AI suggestions. He coded the AI... http://forum.shrapnelgames.com/images/icons/tongue.gif Yep it would be nice for a Dev to step into this thread and give some reaction. Though depending on how much of the AI is hardcoded it may be difficult to really effect the types of changes suggested here...

MythicalMino

October 30th, 2003, 10:07 PM

ok, here is a report on my current game...

Playing the Aran map, Indeps set to 4 strength. 5 other AI's, 3 Easy, 2 Normal...I am playing as C'tis (Miasma theme)

Caelum just attacked me with about 30 troops...the make up of the troops:

Wingless (had several of these)
Blizzard Warriors (Several of these)
Spine Horn Warriors (had several of these also)
1 Wyvern
Seraphine
Caelian Infantry
1 Temple Guard
also, they were attacking with their Prophet

In another border province, they have about 20 troops, mostly made up of Spire Warriors, and they have a Wyvern there also.

On my other border, Michtlan is making their move i think, against me:

110 Troop army:

made up of Warriors, Slaves, Sun Warriors and several Fiends of Darkness

They are placed as 4th place in Army Size. 3 other AI are ahead of them. Michtlin is a Normal AI, Caelum is easy AI.

Anyways, there is my report for this game....what i have seen so far....I have come across several times 20+ armies on my borders, but they pull their troops elsewhere...possibly to fight on another border with another nation? IT would seem....cause the AI's are losing/gaining provinces quite a bit...

Gandalf Parker

October 30th, 2003, 10:17 PM

originally posted by licker
I don't see why this wouldn't work for Dom as well. You'd have a list of commander types (maybe also broken down into magic types for different mages) and a list of unit types (classes rather than specific units so indies fit in). Then a weight is assigned to each type, a crude example for just infantry could be:
LI: 25%
MI: 50%
HI: 25%
So the AI would look at its current army and see that it had only 10% LI, then it would build more LI to get the ratios set. Something like that should be made external so that modders can tweek away to their hearts content. In fact you can use lists like that to control many aspects of AI choices, you can do it for magic, artificing, recruiting, summoning, ... and have master matricies to control the embedded ones...Seems simple and workable. On the other hand, lets pick one race and try to write it all out. Ulm should be simplest. It has a cheap and expensive Version of each of its units which is nice. But I suspect that anything we come up with to "fix" this will amount to fixing it for the indep 2 games people are playing (fast player conflict) and will need fixing again for 4, 6, 8. Then for levels of magic research, and levels of resources, etc. Im thinking that this level of "thinking" might end up being equal in size to the game now (Ive run into this problem before).

Gandalf Parker

October 30th, 2003, 10:23 PM

originally posted by cbeller
Anyways, there is my report for this game....what i have seen so far....I have come across several times 20+ armies on my borders, but they pull their troops elsewhere...possibly to fight on another border with another nation? IT would seem....cause the AI's are losing/gaining provinces quite a bit...
Ive seen that. Im just guessing but it might be a simple way of providing some "smarts" to the AI. He moves the big army around inside his borders. If he kept it still Id be able to plan better. And many times Ive moved on a province of his that I thought was clear only to meet his big army head on.

It would probably work better if he had more armies, a variety. Maybe kept up with the number of provinces he had such as armies=provinces/5

licker

October 30th, 2003, 10:30 PM

Originally posted by Gandalf Parker:
<blockquote>quote:<hr />originally posted by licker
I don't see why this wouldn't work for Dom as well. You'd have a list of commander types (maybe also broken down into magic types for different mages) and a list of unit types (classes rather than specific units so indies fit in). Then a weight is assigned to each type, a crude example for just infantry could be:
LI: 25%
MI: 50%
HI: 25%
So the AI would look at its current army and see that it had only 10% LI, then it would build more LI to get the ratios set. Something like that should be made external so that modders can tweek away to their hearts content. In fact you can use lists like that to control many aspects of AI choices, you can do it for magic, artificing, recruiting, summoning, ... and have master matricies to control the embedded ones...Seems simple and workable. On the other hand, lets pick one race and try to write it all out. Ulm should be simplest. It has a cheap and expensive Version of each of its units which is nice. But I suspect that anything we come up with to "fix" this will amount to fixing it for the indep 2 games people are playing (fast player conflict) and will need fixing again for 4, 6, 8. Then for levels of magic research, and levels of resources, etc. Im thinking that this level of "thinking" might end up being equal in size to the game now (Ive run into this problem before). <hr /></blockquote>I don't know if its as large of a task as you make it out to be http://forum.shrapnelgames.com/images/icons/icon7.gif

There would be a series of matricies, and it would be a bear to optimize them, but if they are moddable... then we can count on the relentless fans tweeking their fav nations to get them to be kick *** wicked http://forum.shrapnelgames.com/images/icons/icon7.gif

As to the differences in game settings... I agree completely that these need to be factored in, but I don't think its that hard to do. You have a matrix for the game settings with 'correction' values to the national matricies. Or you may have a game setting matrix for each nation (though some nations are similar enough...).

What I think we would find if we did this is that there are indeed preferable units for nearly all situations, especailly when we are dealing with shades of grey in Ulm Infantry, in fact this would give the AI a boost as it wouldn't be tempted to use inferior units (slight though the difference is) as a human would be.

I'd still like to hear from the devs is such a system is at all possible before I devote any more time to specifics. Though if others want to take a nations unit roster and break it down go for it. Remember that the %s should differ depending on what the goal of the nation is (magic, might, dominion, ...) also modified by the available richness of provinces, income, ... its alot of tables, or inputs, but its not that hard to do (I think http://forum.shrapnelgames.com/images/icons/icon12.gif )

Gandalf Parker

October 31st, 2003, 01:12 AM

I would love to see a scripting language for AIs. Ive said since before Dom 1 that games like this should have one but its probably just my MUD background talking.

It would be great to have player-made scripts. For some that would mean that the great players could create smart scripts for their favorite race so that we solo players could play "them" without playing MP. Or at least get a feel for how we would do against their tactics (a script cant be as good as a player). Of course some scripts would suck but for me (Fanatic of Random) that would be fine also. We could have scripts for playing styles of barbaric horde, home defender, slow careful expander, diplomatic, backstabber, aggressive expander, comes in waves, elite units only, etc etc etc. Then I could write a script to randomly select from the pile for each game (I love it). Player-made scripts will always allow far more variety than one set of programmers straining to think in styles different than they wouild play themselves. OR at least (as it has in some other games) it lets them say "we gave the players the means and they didnt do anything with it".

The problem is that the variables and hooks needed for such scripting to work usually has to be written into the game from the beginning.

[ October 30, 2003, 12:14: Message edited by: Gandalf Parker ]

Gandalf Parker

October 31st, 2003, 01:30 AM

The area of unit selection is really hard to move from complaining to real suggestions. To keep a database of each race and each gaming situation would be extreme. They might as well move the AI to a seperate program so it doesnt slow down the MP play. Many of the people complaining about the AI are playing with independents set at 2 or 3. The strategies and units for such a game would be very different than if it were set in the middle, and that would be different than the high setting. And thats just ONE of the pre-game settings.

OK Ive heard that the AI should not buy cheap units but instead buy a "real army". So insteadof figuring out how to get the most units with the least upkeep (patrolling strtegy) they should buy the most expensive? That would lead to armys of hvy cavalry, which I guess isnt too bad.

Personally I think an AI best bet is to buy everything. Start at the cheapest and buy 1, move up thru the choices then start over. They would reach a point of not being able to buy another hvy cavalry or elephant or hydra and would use up the left over amount on cheap troops so they would end up with slightly more lesser troops than expensive ones. But that wouldnt be too bad.

What about leaders? How do you make sure the AI buys a mix of leaders? It has no memory so it cant say "I bought this Last time and now need to buy this". Hmmmm well actually it might be able to fake that. If turn number divides by 5 evenly then buy the most expensive leader THEN buy troops. *sigh* still not great but maybe it would be better?

Aristoteles

October 31st, 2003, 01:53 AM

I like the idea of this kind of scripting.

Another problem: The AI is sending in his pretenders to fight in the Arena. I think it should send in commanders instead of pretenders.

Forts and castles: I sorta agree with this, in the demo the AI didnt built a single one. Maybe 1 or 2 in like 6 games.

Strategic AI: I am sure that it can be improved but it will need precise scripting, also propably lot of time.

DominionsFan

October 31st, 2003, 03:47 PM

Yeap. It seems, that the Dominions 2 AI love to use these worthless units. http://forum.shrapnelgames.com/images/icons/icon9.gif
Ok they are not worthless, but without good units and monsters, those armies are very very very weak, compared to the elite human armies. http://forum.shrapnelgames.com/images/icons/icon9.gif

licker

October 31st, 2003, 04:03 PM

I think its hard to tell from the demo how effective the AI is with units. The game settings do make alot of difference. I still think the discussion and suggestions in this thread are valuable, and I wish a dev would deign to let us know how things are currently set up.

I had some time to fool around Last night and I was playing with Indies at 6 or 7 (can't remeber exactly) with all nations on normal or difficult on the bigger map. Anyway, by turn 10 I had expanded or scouted to other nations borders (not surprising as the map was crowded), by turn 15 I had 4 nations declare war on me, though only one was actually adjacent to me. I routinely saw armies of 100+ units moving around though, and I managed to catch Marginon's big force with mine (playing Pythem) I slaughtered them, mostly because I had sent in an assassin who picked off a couple of mages (well inquisitors... you know...) and my Arch Therug with 3 slaves doled out some nastyness...

Anyway, a few turns later I regrouped and reenforced and set seige to their capital. The next turn I had to repell a relief army with 8 summer lions and assorted pikeneers and x-bows. I beat them back, but it wasn't clean... So the AI does seem capeable of using summons, and it will make bigger armies.

Part of the problem some of you are encountering is probably due to not having enough opponents on a too big map. With a cramped map the AI is forced to consolodate more (though it could still probably use some tweeks to the units it recruits). In the next 20 turns of this game it will be interesting to see what the other nations who declared war on me do, 2 more of them now have provinces boarding mine, and if I can't finish of Marg quickly I may have to recall that beseiging army to deal with other threats. All in all this game has been entertaining, at least for the short duration it's Lasted.

Again though, there is alway room for improvement of the AI, I hope that the discussion in this thread is not being ignored http://forum.shrapnelgames.com/images/icons/icon7.gif

-Storm-

October 31st, 2003, 04:22 PM

I am sure that we are NOT ignored, but IW tends to don't reply usually. http://forum.shrapnelgames.com/images/icons/icon7.gif It is all cool, it is good enough, if they are READING our suggestions at least.

I have a very good idea. Well hopefully it is a very good idea.

So we all know about the debug mode, and that we can pass the control to the AI.
What about adding a special debug mode.

Special debug mode:
((
1. make a god
2. set up a game
3. let the AI control our nation
))

-->>
We will be able to see the actions of the AI. That would be awesome. Than, it would be lot easier to post suggestions about the AI. {{Now, after you've gave the control to the AI, the game will quit.}}

What do you think? [I hope that this is possible.]

[ October 31, 2003, 15:26: Message edited by: -Storm- ]

Mortifer

October 31st, 2003, 04:43 PM

Man, that is an excellent idea! If we could see the "moves" made by the AI, after each and every turn....that would help hella lot!
Seriously I love this idea. http://forum.shrapnelgames.com/images/icons/icon6.gif

licker

October 31st, 2003, 04:50 PM

It is a good suggestion, as would be some form of autoplay, where the Comp would run for a set number of turns and then allow the user to view the contents of every province.

That way you could set different nations to different levels and get a better feel for how well each nation is tuned rather than just watching one nation at a time.

MStavros

October 31st, 2003, 05:30 PM

Indeed. Good suggestion.

October 31st, 2003, 06:31 PM

And the suggestion about special debug mode is also very useful for trying out/balancing mods.

In any case, yesterday I was playing iron faith Ulm with 5 normal AI opponents, indies on 5, and I've encountered several big armies. On turn 39 Mictlan showed up on my border with an army of 280 men. I know they are not of the highest quality, but still, it's decent size force. Btw, I'm not suggesting that the strategical AI need not be improved (the AI can never be improved enough).

What I found, and I don't know whether other people have noticed as well, is that the AI is constantly moving its armies back and forth in the border provinces. So, if you want to take the province, order an attack when there is an army present - it will leave, and you'll take the province undefended. If you want to wipe out the army instead, attack when the province is empty. And this is almost exclusively the case as far as I was able to see - armies never stay in place on borders to defend them or move unpredictably, they keep on pacing and exploiting is easy because of this. I don't know what's the reason for this behaviour(maybe they are running out of supplies in the border province or something), but it's very predictable, and a possible target for improvement.

MStavros

October 31st, 2003, 07:27 PM

Originally posted by HJ:

What I found, and I don't know whether other people have noticed as well, is that the AI is constantly moving its armies back and forth in the border provinces. So, if you want to take the province, order an attack when there is an army present - it will leave, and you'll take the province undefended. If you want to wipe out the army instead, attack when the province is empty. And this is almost exclusively the case as far as I was able to see - armies never stay in place on borders to defend them or move unpredictably, they keep on pacing and exploiting is easy because of this. I don't know what's the reason for this behaviour(maybe they are running out of supplies in the border province or something), but it's very predictable, and a possible target for improvement. I had noticed this too. Another AI weakness.

Mortifer

October 31st, 2003, 08:44 PM

280 men? Hm that is a nice little army indeed!

This is so weird. Why the heck the AI won't use that much heavy/elite units??

October 31st, 2003, 09:16 PM

Originally posted by Mortifer:
280 men? Hm that is a nice little army indeed!

This is so weird. Why the heck the AI won't use that much heavy/elite units?? I was saying that Mictlan as a nation doesn't get very strong units. The report said that it was composed of eagle and jaguar warriors in addition to other units, so it wasn't just rabble. I'm just clarifiying, I'm not saying there is no room for improvement.

Mortifer

October 31st, 2003, 09:42 PM

Originally posted by HJ:
<blockquote>quote:<hr />Originally posted by Mortifer:
280 men? Hm that is a nice little army indeed!

This is so weird. Why the heck the AI won't use that much heavy/elite units?? I was saying that Mictlan as a nation doesn't get very strong units. The report said that it was composed of eagle and jaguar warriors in addition to other units, so it wasn't just rabble. I'm just clarifiying, I'm not saying there is no room for improvement. <hr /></blockquote>So you had a battle with that AI army?

October 31st, 2003, 09:59 PM

Originally posted by Mortifer:
So you had a battle with that AI army? Not with that one, it was the Last turn, but I had skirmishes with Mictlan before, and their bulk is somewhat low tech. I encountered eagle warriors, and they're slightly better. Jaguar ones I haven't seen in the field, but they sound like elites... http://forum.shrapnelgames.com/images/icons/icon7.gif The biggest strength for Mictlan I've seen so far are the massed volleys of javelins- those can be really devastating.

Mortifer

October 31st, 2003, 10:06 PM

Originally posted by HJ:
<blockquote>quote:<hr />Originally posted by Mortifer:
So you had a battle with that AI army? Not with that one, it was the Last turn, but I had skirmishes with Mictlan before, and their bulk is somewhat low tech. I encountered eagle warriors, and they're slightly better. Jaguar ones I haven't seen in the field, but they sound like elites... http://forum.shrapnelgames.com/images/icons/icon7.gif The biggest strength for Mictlan I've seen so far are the massed volleys of javelins- those can be really devastating. <hr /></blockquote>Owww. Oh well, the full game is coming soon. http://forum.shrapnelgames.com/images/icons/icon6.gif

DominionsFan

October 31st, 2003, 11:45 PM

Originally posted by -Storm-:
I am sure that we are NOT ignored, but IW tends to don't reply usually. http://forum.shrapnelgames.com/images/icons/icon7.gif It is all cool, it is good enough, if they are READING our suggestions at least.

I have a very good idea. Well hopefully it is a very good idea.

So we all know about the debug mode, and that we can pass the control to the AI.
What about adding a special debug mode.

Special debug mode:
((
1. make a god
2. set up a game
3. let the AI control our nation
))

-->>
We will be able to see the actions of the AI. That would be awesome. Than, it would be lot easier to post suggestions about the AI. {{Now, after you've gave the control to the AI, the game will quit.}}

What do you think? [I hope that this is possible.] Hell yeah! We need something like this, to test the AI! Devs? Hello?

Mortifer

November 1st, 2003, 01:14 AM

I had an interesting game today with the demo:

Indies were set to 9! I couldny expand too fast, and when I conquered a few provinces, I met with the first AI oppoment: Man. The AI [man] had a nice army right near my province. It had 120+ troops, but those units were weak/mediocre, + the army consisted very few "heavyweight" units. Oh and I didnt met with summoned AI units that much at all.

I think Kris posted something about the AI - supply thing. Maybe that is causing the trouble? IE. The AI won't make/summon decent units?

[ October 31, 2003, 12:17: Message edited by: Mortifer ]

LordArioch

November 1st, 2003, 01:21 AM

Mictlan doesn't really have heavy infantry. I saw the AI throw big big Groups of troops at me containing warriors (a bit too many without armor) and jaguar warriors, eagle warriors, and sun warriors. Unfortunately I don't think he blessed the eagle warriors which is sad. I don't want to spoil them for you but they're a really cool unit idea. http://forum.shrapnelgames.com/images/icons/icon7.gif For mictlan elite troops are just sacred, only sun are really better armored than standard, and i think they have only 11 prot.

DominionsFan

November 1st, 2003, 08:46 AM

Hey, how can you bless your troops? I am new you know. http://forum.shrapnelgames.com/images/icons/tongue.gif

MStavros

November 1st, 2003, 11:06 AM

I met with an AI army in my Last game. It had more than 300 units, mostly weaklings of course.
The AI will make bigger armies, however they are weak compared to the human armies.

The AI should be tweaked various ways.

1. The AI is preferring light troops. -> not good. Light troops are very weak compared to the heavy troops. This is why it is so easy to crush the AI.
2. The AI should summon more creatures.
3. The AI should use more assassins.
4. The AI should guard his mages / commanders better.
5. The AI should use better battlefield tactics.

[ November 01, 2003, 10:07: Message edited by: MStavros ]

Wendigo

November 1st, 2003, 11:06 AM

Just finished a demo game vs 4 impossible AIs, indeps lv6, relaxed playing: Ermor, Mictlan, Tien Chi & Machaca, killed Mictlan & Machaca before the time ran out.

I agree that the AI builds far too many light troops & militia types, yet this is something that should be easily patched with a resources boost & some hardcoded build orders that require a % of HIs & elites & prioritize national troops over indeps.

I faced some sizeable forces, Tien Chi had around 500 troops at its capitol by the end of the game, I faced also some good summons by the AI: Fiends of darkness by Mictlan & Mech men + Crushers from Machaca. So while there's certainly room for improvement regarding the strategical AI, it is not as bad as the prophets of doom anounce. It's definitely correct for a game of this type.

The tactical AI is even better than in Dom I, it now Groups its troops by type & has them perform with pretty correct scripts. Mictlan Fiends ignored my troops at front and flew over them to hit my mages in the rear.

The tactical spell AI is still great: Machaca managed to kill my Pretender with a good choice of spells when I stormed its capitol (which is something that many human players have failed at, btw).

November 1st, 2003, 11:19 AM

I've seen a good move by the Ai earlier today. I was Abysia, besieging Vanheim capitol. Instead of constantly pouring a few troops every now and then on my besieging army, it seems that the AI waited for the reenforcements to arrive, and then hit me with everything plus sallied forth from the castle. It was a cool fight where my moloch killed his ghost king in one-on-one fight, only to be smitten by a puny dwarf the next turn casting shards from close range - they all hit the spot. And I didn't even manage to catch that dwarf later... Grrr....

Mortifer

November 1st, 2003, 12:02 PM

DominionsFan

November 1st, 2003, 03:23 PM

Originally posted by MStavros:
I met with an AI army in my Last game. It had more than 300 units, mostly weaklings of course.
The AI will make bigger armies, however they are weak compared to the human armies.

The AI should be tweaked various ways.

1. The AI is preferring light troops. -> not good. Light troops are very weak compared to the heavy troops. This is why it is so easy to crush the AI.
2. The AI should summon more creatures.
3. The AI should use more assassins.
4. The AI should guard his mages / commanders better.
5. The AI should use better battlefield tactics. Yup, nice listing. If these things would be tweaked! The AI would roxor! http://forum.shrapnelgames.com/images/icons/icon10.gif

Mortifer

November 1st, 2003, 04:09 PM

Yet another annoying thing:

I just had a game, and I met with Ulm at turn 12. They had some small, wandering armies.
One of those armies was like this.

-Pretender God [Earth Mother]
-Ulm Cpt
-2 heavy infantry.

I was like wtf! http://forum.shrapnelgames.com/images/icons/shock.gif http://forum.shrapnelgames.com/images/icons/icon8.gif

I mean what the hell is the point in this AI "maneuver"? If we can call this a maneuver at all!
Of course I've killed his pretender.....

[ November 01, 2003, 15:09: Message edited by: Mortifer ]

Alneyan

November 1st, 2003, 05:54 PM

Originally posted by MStavros:
I try to help:

1. The AI won't be smarter @ higher diff. levels, but it will have more design point to spend. I think thats all. Correct me if there is anything more.

2. I am not a beta-tester, but you can find lot of suggestions, that how to upgrade the AI.

3. That would be a good idea, however I don't think that any of us can script a really good AI.
I guess Dominions AI is lot more complicated than the Space Empires IV. AI. [I am not sure..]
However..who knows. Thanks for the answers! For the second point, it was not quite a suggestion, but rather a question to see if the AI was merely having troubles in the early game (which is common in other games), or if it was... "generalized". Being attacked by the God alone in the late game would be a bit harder to explain than at turn 12.

I cannot say for sure about the AIs, it depends on the fields I would believe. (You don't have to handle colonization, construction is much easier in Dominions, but you have to program commanders, spell casting and so on)

Aristoteles

November 1st, 2003, 06:30 PM

I really would like to see some comments from the developers. What is their opinion about the ideas & suggestions?
http://forum.shrapnelgames.com/images/icons/blush.gif

[ November 01, 2003, 17:30: Message edited by: Aristoteles ]

LordArioch

November 1st, 2003, 06:52 PM

Originally posted by MStavros:

1. The AI is preferring light troops. -> not good. Light troops are very weak compared to the heavy troops. This is why it is so easy to crush the AI.
2. The AI should summon more creatures.
3. The AI should use more assassins.
4. The AI should guard his mages / commanders better.
5. The AI should use better battlefield tactics. In my normal AI experience (mostly against Mictlan)

1. The AI does seem to overuse light units. IMO, the best way to fix this would be to better balance light troops. AI did a decent job of throwing in jaguar warriors/sun warriors. And Mictlan doesnt really have heavy troops.
2. I've seen the AI cast more blood summons than I have. Mictlan never seems to move without having a few fiends of darkness and assorted demons about. I don't find most summons more than a decent way to spend gems anyway, the low level ones usually don't turn the tide of battles (anymore)
3. The AI assasinated my pretender and a researching mage and made about 4+ other attempts. If anything he should use assasains less since they usually aren't well equipped enough. (pretender died due to bug)
4. This has always been true, but at least removal of attack commanders makes it harder to abuse. Be nice if any commander got 1-2 troops always set to guard. I started doing this to my mages when he started assasinating them.
5. His battlefield tatics aren't ingenious but I have yet to see anything stupid from the AI here. Usually just a general troops go forward archers shoot kind of thing. Not wonderful but general purpose and usually pretty effective.

Gandalf Parker

November 1st, 2003, 07:29 PM

I have a suggestion to "fix" the AI. I think the choices of AI should link directly to the next page where the game selections are made.

If easy AI is selected then the next page should default settings such as independents to 2 or 3 so that the tactics used by the AI match the game.

If a hard setting of AI is selected then a default setting such as indeps to 6 or 7 would make it less likely that people would force the AI to play stupidly.

Of course all of the game settings apply, and too bad that map size cant be taken into account also. The other fixes are good also but this might be quicker and sooner plus take care of many of the complaints.

Aristoteles

November 1st, 2003, 08:50 PM

We have lot of problems with the strategic AI.

My list about the strat. AI problems:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

(((((Feel free to add more, please continue with number 9.)))))))

Zerger

November 1st, 2003, 10:38 PM

Gandalf Parker

November 1st, 2003, 10:50 PM

Originally posted by Aristoteles:
We have lot of problems with the strategic AI.Maybe you do. Im still trying to figure that out.
I dont see the problems you are having but then Im not forcing the AI to play its hard-game strategies on easy game settings so that I can rush them while they are still building up. Some of these "bad tactics" you see are perfectly good tactics but being used in the wrong game.

I know that the hard AI works well in hard games. Im kindof curious now if the easy AI might not do better than the hard AI when played in games of light settings

[ November 01, 2003, 21:52: Message edited by: Gandalf Parker ]

st.patrik

November 1st, 2003, 10:54 PM

JK posted Last week that he was going to be away until Sunday. So no, he probably hasn't been reading this thread the Last few days http://forum.shrapnelgames.com/images/icons/icon7.gif

I'm sure he'll see it when he gets back - it seems the devs are usually pretty good about keeping up with stuff that's being talked about.

Zerger

November 1st, 2003, 11:04 PM

Originally posted by st.patrik:
JK posted Last week that he was going to be away until Sunday. So no, he probably hasn't been reading this thread the Last few days http://forum.shrapnelgames.com/images/icons/icon7.gif

I'm sure he'll see it when he gets back - it seems the devs are usually pretty good about keeping up with stuff that's being talked about. Actually that was Kristoffer if I am correct. JK is Online quite a lot -> just check out that who is Online, his name is there usually.

PvK

November 1st, 2003, 11:06 PM

Actually, I wonder how many of the people who are having an easy time with the AI, are already expert human players.

Personally, I am getting nailed by the AI!

PvK

Zerger

November 1st, 2003, 11:15 PM

Originally posted by PvK:
Actually, I wonder how many of the people who are having an easy time with the AI, are already expert human players.

Personally, I am getting nailed by the AI!

PvK Odd, I am not a vet, but I can easily beat the dominions 2. impossible AIs.
The AI is very weak in many ways.

[ November 01, 2003, 22:15: Message edited by: Zerger ]

-Storm-

November 1st, 2003, 11:26 PM

Originally posted by Aristoteles:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.
Period. Fix these, and we will be happy. http://forum.shrapnelgames.com/images/icons/icon12.gif

Just a little addition for 1. -> The AI should recruit heavy/elite units, and make complex/varied armies.

[ November 01, 2003, 22:28: Message edited by: -Storm- ]

Gandalf Parker

November 1st, 2003, 11:40 PM

Originally posted by PvK:
Actually, I wonder how many of the people who are having an easy time with the AI, are already expert human players.

Personally, I am getting nailed by the AI!

PvK Most of them are playing with indeps at 2 or 3 which makes it really easy to rush the AI. If the AI is set to impossible level it actually becomes even easier than that since they have "lied" to the AI and told it to choose its actions planning for a long conflict instead of a short one.

If you are playing with the AI set at a level about equal to the game settings then you are getting a more realistic result. You can write-off the rest of them (and some which may even be multiples of one player) since an AI which takes into account every setting posibility is unlikely to ever get created in this game.

[ November 01, 2003, 22:43: Message edited by: Gandalf Parker ]

November 2nd, 2003, 12:21 AM

Originally posted by PvK:
Actually, I wonder how many of the people who are having an easy time with the AI, are already expert human players.

Personally, I am getting nailed by the AI!

PvK It might also have to do with the style of play. I also mostly roleplay, and take my time to savor the moments (until it abruptly ends at turn 40, of course). I mostly make contact with the AI around turn 20, and even then it's usually just the skirmishing for the rest of the demo instead of the charge towards the capitol. People who rush the AI the way they play in MP are always going to beat it senseless - but that's what that style of play is geared toward anyway: beating the opponent senseless. I've seen this happen with other games as well, such as Total War, where people complaining about the AI immediately go for the kill, and then wonder why did the kill happen.

I'm not saying there is no room for improvement, and I would be more than happy to see the improvements, but there are other things to take into consideration as well in these discussions.

[ November 01, 2003, 23:23: Message edited by: HJ ]

Alneyan

November 2nd, 2003, 01:07 AM

I will have a few questions about the AI, as you can see, I have not played much yet. http://forum.shrapnelgames.com/images/icons/icon12.gif

1) Do we know what are the differences between the different levels? Is it only improved intelligence, or is there a resource/gold/research/magic gem (and so on) bonus the player does not have? Dominions AI may not rely as much on bonuses than the vast majority of AI.

2) For the beta testers mainly: is the AI more efficient in the middle and late game than in the early game? Perhaps the AI does not handle well early conflicts, or has trouble with early expansion (compared to an experienced player who tries to expand as fast as possible), or some other problem. Most AIs tend to have a lot of problems in the early game, and some can even be wiped out without much hassle in the beginning of the game. (And then we would have to wait for the full game to see what is the situation after, say, 100 turns)

3) Is there any chance players could alter/make their own AI files? That would depend obviously on how the AI behaviours are handled. There are games in which the original AI has been improved by players. (Space Empires IV for example)

MStavros

November 2nd, 2003, 01:21 AM

I try to help:

1. The AI won't be smarter @ higher diff. levels, but it will have more design point to spend. I think thats all. Correct me if there is anything more.

2. I am not a beta-tester, but you can find lot of suggestions, that how to upgrade the AI.

3. That would be a good idea, however I don't think that any of us can script a really good AI.
I guess Dominions AI is lot more complicated than the Space Empires IV. AI. [I am not sure..]
However..who knows.

MStavros

November 2nd, 2003, 09:18 AM

Originally posted by -Storm-:
<blockquote>quote:<hr />Originally posted by Aristoteles:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.
Period. Fix these, and we will be happy. http://forum.shrapnelgames.com/images/icons/icon12.gif

Just a little addition for 1. -> The AI should recruit heavy/elite units, and make complex/varied armies. <hr /></blockquote>Yes, yes, yes.
This is a more accurate list, than mine. http://forum.shrapnelgames.com/images/icons/icon7.gif

DominionsFan

November 2nd, 2003, 12:43 PM

Originally posted by MStavros:
<blockquote>quote:<hr />Originally posted by -Storm-:
<blockquote>quote:<hr />Originally posted by Aristoteles:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.
Period. Fix these, and we will be happy. http://forum.shrapnelgames.com/images/icons/icon12.gif

Just a little addition for 1. -> The AI should recruit heavy/elite units, and make complex/varied armies. <hr /></blockquote>Yes, yes, yes.
This is a more accurate list, than mine. http://forum.shrapnelgames.com/images/icons/icon7.gif <hr /></blockquote>Hey, I agree with these! JK please do something about the mentioned errors! http://forum.shrapnelgames.com/images/icons/icon12.gif

[ November 02, 2003, 11:44: Message edited by: DominionsFAN ]

Nerfix

November 2nd, 2003, 12:56 PM

Originally posted by Aristoteles:
We have lot of problems with the strategic AI.

My list about the strat. AI problems:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

(((((Feel free to add more, please continue with number 9.))))))) 1.Is true, but some nations like Mictlan just don't have heavy troops.
2.As paradoxal this is, i have seen indies protecting their commanders, but AI doesn't seem to protect it's commanders...
3.The forts are pretty expensive though... I haven't built a single fort in the demo... But AI doesn't seem to build them.
4.Yup.
5.Hey, i wander alone with my pretender too... Are you sure the pretender wasn't searching for magic sites? Or can the AI do that?
6.I noticed that too.
7.From multiple directions? Can you give any other examples?
8.The doesn't seem to be equiping commanders at all.

Nerfix

November 2nd, 2003, 02:36 PM

Ok, 4 things i saw today:
The AI can show some glimmer of tactical inteligence:
Caelum apparently used "Attack Rearmost" command to attack my commanders...

The AI can use assasins!

The AI equips it's commanders: The aforementioned assasin had Enchanted Sword with him.

The AI either holds it's armies in a province or then moves several armies in and out of the same province:
Caelum had an army of about 80 units in one of his temple provinces for 4-5 turns.

EDIT:I was using 6 Normal AI's, Aran and default settings(indy str 3).

[ November 02, 2003, 13:52: Message edited by: Nerfix ]

Mortifer

November 2nd, 2003, 03:54 PM

Originally posted by Nerfix:
<blockquote>quote:<hr />Originally posted by Aristoteles:
We have lot of problems with the strategic AI.

My list about the strat. AI problems:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

(((((Feel free to add more, please continue with number 9.))))))) 1.Is true, but some nations like Mictlan just don't have heavy troops.
2.As paradoxal this is, i have seen indies protecting their commanders, but AI doesn't seem to protect it's commanders...
3.The forts are pretty expensive though... I haven't built a single fort in the demo... But AI doesn't seem to build them.
4.Yup.
5.Hey, i wander alone with my pretender too... Are you sure the pretender wasn't searching for magic sites? Or can the AI do that?
6.I noticed that too.
7.From multiple directions? Can you give any other examples?
8.The doesn't seem to be equiping commanders at all. <hr /></blockquote>I agree with Aristoteles. These are the {major?) weaknesses of the AI. As I've said in the prev. topic, if these things will be fixed/updated, we won't have problems with the AI.

Nerfix

November 2nd, 2003, 04:02 PM

I'm not sure about this, but anyway:

9)The AI builds too much temples

The AI could be in perpetual broke, explaining the cheap units and lack of forts, because a temple costs 200 gold and there could be better uses for the money.

Not that AI shouldn't build temples, it just seems to build too much of them.

[ November 02, 2003, 15:02: Message edited by: Nerfix ]

Gandalf Parker

November 2nd, 2003, 04:31 PM

Originally posted by Nerfix:
I'm not sure about this, but anyway:

9)The AI builds too much temples
The AI could be in perpetual broke, explaining the cheap units and lack of forts, because a temple costs 200 gold and there could be better uses for the money.
Not that AI shouldn't build temples, it just seems to build too much of them. Depends on the AI. Some AI's would be incredibly stupid not to build temples since its their big strong point. Also unless its changed in Dom2 not all AI's have to spend 200 for a temple. Some spent only 100 per temple.

Nerfix

November 2nd, 2003, 04:46 PM

Originally posted by Gandalf Parker:
<blockquote>quote:<hr />Originally posted by Nerfix:
I'm not sure about this, but anyway:

9)The AI builds too much temples
The AI could be in perpetual broke, explaining the cheap units and lack of forts, because a temple costs 200 gold and there could be better uses for the money.
Not that AI shouldn't build temples, it just seems to build too much of them. Depends on the AI. Some AI's would be incredibly stupid not to build temples since its their big strong point. Also unless its changed in Dom2 not all AI's have to spend 200 for a temple. Some spent only 100 per temple. <hr /></blockquote>You're rigth, but i have only seen Carrion Pangaea and Man AI was quickly terminated by other AI's in my latest game.
OTOH i haven't seen Mictlan building temples, but since Mictlan's temples don't spread Dominion they don't realy need them.

Particle

November 2nd, 2003, 06:09 PM

Originally posted by Aristoteles:
We have lot of problems with the strategic AI.

My list about the strat. AI problems:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.Yup, I tend to agree with you on all points.
Maybe the 8. was a bit too early to mention, since the 40 turn limit demo is very short.
Well maybe not, I can fully equip a commander/pretender until turn 40... http://forum.shrapnelgames.com/images/smilies/rolleyes.gif

[ November 02, 2003, 17:10: Message edited by: Particle ]

MythicalMino

November 2nd, 2003, 08:30 PM

my question concerning the strategic AI is this: If it is indeed as weak as ppl are claiming it is, then what happened between Dominions 1 and dominions 2? Granted the AI in Dom1 wasn't the "end all, be all" of AI, but it was decent enough, hell, it can still give me a very big run for my money....so, what happened between the 2 games?

Something had to have happened during the final stages of development, where the beta testers couldn't have caught it. If they were saying that the ai was great, but now, completely new guys are stomping it, then something had to happen towards the end, right?

It either has to be something to do with a late-stage development code change or data change somewhere....or, the ppl that are saying that they can wipe the virtual floor with the AI are either lying about it (NOTE: I AM NOT SAYING ANYONE IS LYING, THIS IS JUST A POSSIBLE OPTION!!!!!!!!)....or, they are setting the game up strange or something along those lines....

Now, if it is indeed something went wrong in the late stages of development, then possibly IW will be able to find it easy enough....and fix it of course....

Alneyan

November 2nd, 2003, 08:45 PM

My own opinion is that the current games played with the demo are limited in time, so people tend to rush the AI with full strenght, supercombattants, you name it. But the AI isn't supposed to handle such situations well, hence easier victories. Multiplayer tactics are usually not liked by the AI. http://forum.shrapnelgames.com/images/icons/icon12.gif

However, there are other concerns that cannot be explained this way: like the problem with unguarded Pretenders for example. But it seems like this problem is more along the lines of "Pretender forgot to seek reinforcements". I am still waiting to see what will happen without the turn limit, if late game AI armies are featuring stronger units or not. If it is the latter, I will start to be worried.

I would like a confirmation: is Dominions AI cheating? I would tend to believe it is not the case. If so, Dominions AI is really not that bad, and is one of the few AIs that does not rely on extra resources to provide a challenge. And I will be then in the side of "The AI isn't that bad, really". Finally, if you do want additional challenge, you may very well give the AI units/abilities/items and so on like someone mentioned in another thread dedicated to the AI. (If you are reading me, I am sorry, I forgot your name. You shall be called henceforth the Anonymous Poster, if you don't mind that is. *Silly me*)

Edit: it was Nagot Gick Fel in this thread (http://www.shrapnelgames.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=74;t=000251;p=1) , about halfway down in the page. (At the moment of the writing that is)

[ November 02, 2003, 19:47: Message edited by: Alneyan ]

Zerger

November 2nd, 2003, 09:17 PM

Originally posted by cpbeller:
my question concerning the strategic AI is this: If it is indeed as weak as ppl are claiming it is, then what happened between Dominions 1 and dominions 2? Granted the AI in Dom1 wasn't the "end all, be all" of AI, but it was decent enough, hell, it can still give me a very big run for my money....so, what happened between the 2 games?

Something had to have happened during the final stages of development, where the beta testers couldn't have caught it. If they were saying that the ai was great, but now, completely new guys are stomping it, then something had to happen towards the end, right?

It either has to be something to do with a late-stage development code change or data change somewhere....or, the ppl that are saying that they can wipe the virtual floor with the AI are either lying about it (NOTE: I AM NOT SAYING ANYONE IS LYING, THIS IS JUST A POSSIBLE OPTION!!!!!!!!)....or, they are setting the game up strange or something along those lines....

Now, if it is indeed something went wrong in the late stages of development, then possibly IW will be able to find it easy enough....and fix it of course.... cpbeller you are one of the most ignorant persons on these Boards. Frankly, I don't wanna hurt your feelings, but I am sure that you've seen this poll - http://www.shrapnelgames.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=74;t=000167

Maybe you've seen these parts of it:

What should IW improve in Dominions II.?
AI 65% (39)

What do you think about the Dominions II. AI?
Average 28% (17)
Below average 20% (12)
Poor 13% (8)

To answer your question: Yes propably these 40 Users were lying about the AI.

...
http://forum.shrapnelgames.com/images/smilies/rolleyes.gif http://forum.shrapnelgames.com/images/icons/shock.gif
...

What happened with the AI? I remember that I've seen some infos on the first Doms2. Shrapnel site, that the AI will be very good.
You were right in one thing, something propably happened, since the AI is far from good.
There is a huge thread about this, so I won't argue about it.

Fyron

November 2nd, 2003, 10:20 PM

Guys, there is no reason to get hostile over this. It is just a game. Go take a walk and cool off or something.

MythicalMino

November 2nd, 2003, 10:23 PM

ok,

1. I am not IGNORANT...I may be a lot of things...but ignorance is not one of them....

2. The purpose of my post was to say it was one of the 3 things....
1. late stage development gone wrong
2. game setups getting messed up
3. ppl are lying about their games....

Now, it would have to be an incredible conspiracy, if I was to actually think that 35-45 ppl were lying about the same thing....

If you re-read my post...you will see that I spent more time asking about the possibility of the late stage development going wrong, than the other ones....and if I was going to accuse someone of lying, I would just go ahead and accuse them, i don't need subtle shadow attacks on ppl to get my point across....

Now, as far as the ai goes....I have noticed that the AI isn't great either...at least not in the demo....i have seen some large armies, but again, most are weaker units....only a few have had tougher troops...

If i offended anyone because you think I said you were a lier....dear God, forgive me....that was not the point of the post....I can read the poll results...40 ppl saying that the ai needs work, well, that should raise a red flag.

at any rate....I can't seem to post anything without getting busted for something from ppl....

have fun

-Storm-

November 2nd, 2003, 10:56 PM

Originally posted by cpbeller:
40 ppl saying that the ai needs work, well, that should raise a red flag.
Period. http://forum.shrapnelgames.com/images/icons/icon6.gif

PvK

November 2nd, 2003, 11:29 PM

Originally posted by cpbeller:
my question concerning the strategic AI is this: If it is indeed as weak as ppl are claiming it is, then what happened between Dominions 1 and dominions 2? ...How about: Between Doms 1 and Doms 2, some people got really good at Dominions?

How about a poll of players who are new to Dominions, about how much of a pushover the AI is for them?

Then, how about a list of games where expert human players have a hard time against the AI?

http://forum.shrapnelgames.com/images/icons/icon12.gif

PvK

Kristoffer O

November 2nd, 2003, 11:39 PM

Originally posted by Alneyan:

I would like a confirmation: is Dominions AI cheating? I would tend to believe it is not the case. If so, Dominions AI is really not that bad, and is one of the few AIs that does not rely on extra resources to provide a challenge. And I will be then in the side of "The AI isn't that bad, really". The AI is not cheating on normal level. At higher levels it gets more design points and earns more money. Thats about it.

I got an idea regarding AI dependancy on LI when I was away. Vacations are 'foyson' for the mind. http://forum.shrapnelgames.com/images/icons/icon7.gif

Regarding fort construction. I'm not sure how the AI works, but where would you build a fortress? Consider fort type, nation, geography, income, resources etc.

What is the most important matter? How should they be weighted? How much shall current wars affect the spending of time and money. What army should build the fortress?

I wouldn't mind a numerical evaluation of this such as:
Castle cost / 2 < Income + res + gem income x 25 + neighbors x 5
Add a couple of other conditions.

Just to make you think. http://forum.shrapnelgames.com/images/icons/icon7.gif Eventually it might result in something good.

-Storm-

November 2nd, 2003, 11:49 PM

Sammual

November 2nd, 2003, 11:51 PM

Originally posted by Nerfix:
I'm not sure about this, but anyway:

9)The AI builds too much temples

The AI could be in perpetual broke, explaining the cheap units and lack of forts, because a temple costs 200 gold and there could be better uses for the money.

Not that AI shouldn't build temples, it just seems to build too much of them. The AI temples have been giving me problems. I have been playing low domain pretenders and my sacred troops / Prophet are worth spit in the neg domain fights.

Sammual

Chris Byler

November 3rd, 2003, 12:49 AM

Originally posted by Kristoffer O:
<blockquote>quote:<hr />Originally posted by Alneyan:

I would like a confirmation: is Dominions AI cheating? I would tend to believe it is not the case. If so, Dominions AI is really not that bad, and is one of the few AIs that does not rely on extra resources to provide a challenge. And I will be then in the side of "The AI isn't that bad, really". The AI is not cheating on normal level. At higher levels it gets more design points and earns more money. Thats about it.

I got an idea regarding AI dependancy on LI when I was away. Vacations are 'foyson' for the mind. http://forum.shrapnelgames.com/images/icons/icon7.gif

Regarding fort construction. I'm not sure how the AI works, but where would you build a fortress? Consider fort type, nation, geography, income, resources etc.

What is the most important matter? How should they be weighted? How much shall current wars affect the spending of time and money. What army should build the fortress?

I wouldn't mind a numerical evaluation of this such as:
Castle cost / 2 < Income + res + gem income x 25 + neighbors x 5
Add a couple of other conditions.

Just to make you think. http://forum.shrapnelgames.com/images/icons/icon7.gif Eventually it might result in something good. <hr /></blockquote>Some responses to these questions:

1. In your list of factors you don't list population type and sites in the candidate province. I often fort to enable building more of an indep troop (especially amazon sacred cavalries, which are all useful; but also e.g. heavy cavalry for Caelum or C'tis, longbows/crossbows for any nation with weak missile troops) and/or protecting a location with high gem income or useful mages (either pop based (amazons again) or site based).

A good first rule is that if you can't spend all your gold buying useful units in forts you already have, you need a new one; if your units are frequently starving because of insufficient supplies far from your forts, you need a new one in that area. (Which reminds me - supply production can be a factor in fort location too.)

Rarely build a fort next to another fort (it is occasionally worth it for something that can be only found in that province, for example I will always fort a province that contains Mount Chaining in Dom I - in Dom II it isn't as powerful but is still pretty good if you are a blood using nation).

2. You don't need an army to build a fort. I usually don't use one. There is some risk in building a fort where you don't have an army, but I find it is often better to have the army doing something else - either taking more indeps (if they are available) or distracting the enemy so he doesn't have time to Call of the Winds every place I might be building a fort. Get some province defense though - it's annoying to have your fort sieged by Call of the Winds (and of course even more annoying to have the construction interrupted).

3. Fort selection value: (Adjustment * (resources + supplies)) + (gems x 10) + value of rare units recruitable here (this value should depend on the rarity and power of the unit; longbows and crossbows count a little, amazon cavalries and mages count more, something like grey knights would count a lot; mages with paths different from your nation are more valuable than mages that duplicate your nation's paths) + misc site value (enter-to-summons, cost reducers).

I don't generally take income into account directly (although the first two terms value high population provinces more). "Adjustment" depends on the admin and gold cost of the castle; higher admin and cost should raise the value of the adjustment. Adjustment = 0 if there is an adjacent fort (this will reduce fort-next-to-fort except when gem income, rare unit value, or site value are high, which are the exceptions to the no-adjacent-fort rule).

4. Once you have built a fort, build a temple. Temples are vulnerable to enemy attack, but in a fort they are far less vulnerable. Then they spread your dominion, let you recruit priests and preach better. Mictlan might be an exception to this rule because their temples don't radiate - but blood sacrifices are more effective when performed in temples (IIRC), so they might want to build temples anyway so that they can sacrifice there. It's better to preach or sacrifice inside a fort because it protects the priest (high level priests especially are pretty expensive).

Build a lab if you are likely to want to recruit any mages (if you are fortifying because of local troops or sites you probably want a lab for that anyway; some nations like Marignon always want to recruit mages) or summon/construct troops there to supplement your recruited troops (C'tis usually wants to do this; but Desert Tombs C'tis can do without the lab, if you don't mind shipping in your unpriests).

Richard

November 3rd, 2003, 12:50 AM

From what I have seen the beta testers have still been complaining about the AI. So this is not an issue of a late change that has crippled things, those testers STILL are getting their butts beat.

What I do see here is mostly veteran dom I players who know how to beat the game through practical experience from the first game. I also think some folks have judged it from limited playing time. Some people reported "THE AI IS THE WORST EVER!!!" within hours of the demo being released. Now tell me how you determine that from one game.

I am sure there are ways we can make the game better, but to say such things with such little to back it up is silly. If you have complaints, feel free to post them here, but it is much more productive to give specific examples and email files to Illwinter.

Now to the other stuff. Some people need to grow up on both sides. This is a game, not the WWF. Stop calling names or I will have to deal with that issue, and I don't want to have to do that.

If you want to get results, do this the right way. We have no problem working with people if they can PROVE there is an issue and work with the developer to fix it. Just posting constantly that you think the AI sucks and your are 3l33t won't get you anywhere.

Gandalf Parker

November 3rd, 2003, 01:36 AM

Originally posted by Chris Byler:
Some responses to these questions:
[/QB]Nice work. Especially the location chart and formula. Thats the type of suggestions needed.

Just to point out to people what does and doesnt work, I was thinking of something simple like having the AI take all provinces connected to its castle, then move in a straight line away from home for 4 provinces, build a new castle and take all the provinces around it, rinse and repeat. Part of the problem I had is that it required the AI to keep a count. Your formula might work better because it involves the AI looking at the fresh map each time to make its judgements.

Maybe mine can be used by the Easy AI? http://forum.shrapnelgames.com/images/icons/icon7.gif

MStavros

November 3rd, 2003, 01:44 AM

Nerfix - I've met with problem #5. in my Last game. I have NO idea that what the heck the AI was doin' with his pretender, but he moved it from border province - border province all the time, back and forth. Another odd thing... http://forum.shrapnelgames.com/images/icons/shock.gif

[ November 02, 2003, 12:45: Message edited by: MStavros ]

MStavros

November 3rd, 2003, 10:14 AM

Originally posted by Richard:
From what I have seen the beta testers have still been complaining about the AI. So this is not an issue of a late change that has crippled things, those testers STILL are getting their butts beat.
The BETA testers are totally new or the AI in the full game will be lot better maybe?
Ive played many games with 1-6 impossible AIs, and I had no problem to kill at least 1 in the demo, easily. [40 turns limit]
Richard, check out this list by Aristoteles:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

I think this is a very good list about the major AI problems. Just ignore the trollish Posts, and read the good ones only.

Mortifer

November 3rd, 2003, 12:31 PM

Originally posted by MStavros:
<blockquote>quote:<hr />Originally posted by Richard:
From what I have seen the beta testers have still been complaining about the AI. So this is not an issue of a late change that has crippled things, those testers STILL are getting their butts beat.
The BETA testers are totally new or the AI in the full game will be lot better maybe?
Ive played many games with 1-6 impossible AIs, and I had no problem to kill at least 1 in the demo, easily. [40 turns limit]
Richard, check out this list by Aristoteles:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

I think this is a very good list about the major AI problems. Just ignore the trollish Posts, and read the good ones only. <hr /></blockquote>Yes, however I really don't see too many trollish Posts in these days. In fact we have very good suggestions in various threads. A fine example is this list about the AI weaknesses by Aristoteles. I am totally sure that he did played with the demo a lot, since he made these statements. I agree with all of them, and I've played hella lot with the demo. http://forum.shrapnelgames.com/images/icons/icon7.gif

Also there are very good polls, like the master list or my #2. poll. Needless to say, that polls are very important.

Aristoteles

November 3rd, 2003, 02:32 PM

Very good guys, just post your ideas and experiences about the AI. http://forum.shrapnelgames.com/images/icons/icon6.gif
This way the devs won't be pissed. Saying that the AI sucks won't help, we must tell them about the weak parts of it, like I did in my list.

[ November 03, 2003, 13:33: Message edited by: Aristoteles ]

Zerger

November 3rd, 2003, 04:23 PM

Hm we all know that the AI won't build forts, or just very rarely. No wonder that the AI will recruit weak troops and militias than.

The AI on impossible level is expanding very fast, even with indep. str. +9.
His home province with the fort is very far usually.
The AI is recruiting heavy troops in his home province, but it won't make 1-2 stronger armies from them. It is making lot of wandering armies with a few heavy units in all of them.

I think the these things should be tweaked first:
- Enforce the AI with a script to build more forts in strategic locations. ((strat. loc.: province which is surrounded by other provinces of the AI, near mountain ranges, do not build a fort next to an other fort, etc..))
- Enforce the AI with a script to recruit more heavy troops/sacred troops&priests/summon more monsters.

I think these are the most important things to fix first of all.
Ther are many more, the list of Aristoteles is well written about them.

Gandalf Parker

November 3rd, 2003, 05:44 PM

OK maybe this will address some of the issues about AI. Or at least clarify some. If nothing else its a VERY different game for the demo.

http://www.techno-mage.com/~dominion/Poke_Eye.map
http://www.techno-mage.com/~dominion/Poke_Eye.tga

OK the actual image is the same eye one. But the map file is very changed. The game is now called "a Poke in the Eye". Be sure to read the prophet announcements since they can be very interesting. And be sure to scout carefully even on indep settings of 2.

Download those 2 files to the maps directory under your dominions2demo game.

If you are interested in the programming code, just ask.
If you think its crap, show me the one you did and I would be interested.

[ November 03, 2003, 17:18: Message edited by: Gandalf Parker ]

ywl

November 3rd, 2003, 06:15 PM

Originally posted by Richard:
From what I have seen the beta testers have still been complaining about the AI. So this is not an issue of a late change that has crippled things, those testers STILL are getting their butts beat.

What I do see here is mostly veteran dom I players who know how to beat the game through practical experience from the first game. I also think some folks have judged it from limited playing time. Some people reported "THE AI IS THE WORST EVER!!!" within hours of the demo being released. Now tell me how you determine that from one game.

I am sure there are ways we can make the game better, but to say such things with such little to back it up is silly. If you have complaints, feel free to post them here, but it is much more productive to give specific examples and email files to Illwinter.

Now to the other stuff. Some people need to grow up on both sides. This is a game, not the WWF. Stop calling names or I will have to deal with that issue, and I don't want to have to do that.

If you want to get results, do this the right way. We have no problem working with people if they can PROVE there is an issue and work with the developer to fix it. Just posting constantly that you think the AI sucks and your are 3l33t won't get you anywhere. It's a fair post.

I've played a few more games and in cases saw my major armies destroyed by the AIs. In one example, my favorite Pythium elite Emerald guards got totally wiped out by Jotun giants - I managed to got out alive only because Jotun got backstabbed by Man http://forum.shrapnelgames.com/images/icons/icon7.gif .

I've played Dom 1 for more than a year and of course, I would usually beat *all* the AIs if I put effort. But I guess, I could probably only do it in 3 out of 4 games.

But I also agree that the list of suggestions are quite constructive and useful for the program.

DominionsFan

November 3rd, 2003, 06:20 PM

Originally posted by Gandalf Parker:
OK maybe this will address some of the issues about AI. Or at least clarify some. If nothing else its a VERY different game for the demo.

http://www.techno-mage.com/~dominion/Poke_Eye.map
http://www.techno-mage.com/~dominion/Poke_Eye.tga

OK the actual image is the same eye one. But the map file is very changed. The game is now called "a Poke in the Eye". Be sure to read the prophet announcements since they can be very interesting. And be sure to scout carefully even on indep settings of 2.

Download those 2 files to the maps directory under your dominions2demo game.

If you are interested in the programming code, just ask.
If you think its crap, show me the one you did and I would be interested. Whatcha done with the map file? Post some details plz.

Alneyan

November 3rd, 2003, 06:56 PM

EDIT: Sorry, silly double post. http://forum.shrapnelgames.com/images/icons/icon9.gif

[ November 03, 2003, 18:03: Message edited by: Alneyan ]

Alneyan

November 3rd, 2003, 07:01 PM

Originally posted by Kristoffer O:
The AI is not cheating on normal level. At higher levels it gets more design points and earns more money. Thats about it. Thanks for the answer Kristoffer. I will say good job with the AI then, AI that does not rely in heavy cheating are becoming rarer and rarer. Mainly because it is much easier to give an AI a lot of additional resources than scripting an effective AI I would tend to believe, and AI isn't exactly considered as a priority by most publishers. But I digress.

Obviously, there is always room for improvement, but that's another topic, well covered in this thread. All I wanted to say was merely an encouragement for using an AI that isn't keen on using many advantages. http://forum.shrapnelgames.com/images/icons/icon12.gif

DominionsFan

November 3rd, 2003, 08:46 PM

Yeah, it is cool that the AI won't cheat, however it is not very effective right now. http://forum.shrapnelgames.com/images/icons/icon9.gif

I had a game today with 5 impossible AIs, independent str was 6, and I've killed 2 AIs before turn 40. http://forum.shrapnelgames.com/images/icons/icon9.gif

Kris, copy and paste that list by Aristoteles, and try to fix those problematic points step by step. Oh and don't foget to reply here, that how is it going! http://forum.shrapnelgames.com/images/icons/icon12.gif

DominionsFan

November 3rd, 2003, 08:46 PM

Damn, sorry for the double post.

[ November 03, 2003, 19:46: Message edited by: DominionsFAN ]

licker

November 3rd, 2003, 08:52 PM

I think the more I fool around with the Demo the less concerned I am about the AI. I still think that making the AI better should be a high priority, but the doom and gloom aspect of it hasn't been apparent to me, at least in the types of games I've been playing (cramped maps, 6 or 7 indie strength). I've seen big armies with a solid HI core. I've seen seasonal spirits in the armies, I've seen maurading pretenders (though not always well defended...)

There is alot of good information and suggestions in this thread though, I hope that the discussion has helped the devs to look at some of the issues being brought up, and given them some thoughts on how to approach those issues.

I would still like to know if there's any way to externalize AI parameters so that modders can try their hand at various things...

DominionsFan

November 3rd, 2003, 09:29 PM

Someone had mentioned that a special debug mode would be cool to inspect the AI. [Checking out the AI's actions in his turns.]
I like this idea too. However we have many good suggestion already. http://forum.shrapnelgames.com/images/icons/icon7.gif

Particle

November 4th, 2003, 01:06 AM

It is nice to see you here Kris. http://forum.shrapnelgames.com/images/icons/icon7.gif
It is good to know that the devs are watching us from the skies. Errr... http://forum.shrapnelgames.com/images/icons/icon10.gif

RadiantFleet

November 4th, 2003, 01:17 AM

Some other factors to consider when enhancing the AI. These are observations based on my preferred dom 1 playing style, one nation human, all of the rest impossible level AI, large map, and indep set to 9. On these settings, I usually win about 90% of the time.

-AI should take different decision trees depending on type of dominion/pretender that a nation has. Consider the following two configurations for Abysia:

Maxed dominion, growth, and order with a fortified city, a zero point pretender, max heat scale, max drain scale, no extra magic for pretender.

With this configuration, slow steady growth works really well. Build a temple in every territory, use defense heavily (minimum of 21, 41 for key territories, with later upgrades to 60). Armies are only for taking territory, not for defense. Your dominion is your main weapon, you take and hold territory, the spread of your dominion does the rest.

Maxed dominion, max heat scale, max drain scale, Castle or some other decent defense city, lich pretender (ie immortal) with a four in every sphere minimum.

With this pretender, magic summoning, immortality, and your dominion is the key. Build as many temples as possible. Buying units is unimportant, summon creatures that create new creatures (ie summer lions are nice, but fire elemental leaders are better, they make more fire elementals). Minimal defense in provinces (ie 21) to protect against sniping attacks, but armies are the serious defenders (because the only cost is leaders). Leaders that can create units should _never_ be sent into combat, only sit in castles and create more units.

These two configurations illustrate how different good strategies can be, depending on the characteristics of a nation and the pretender selected.

What I would really like is to have some way of specifying what pretender a computer player can use, or at least designing the ones in the "pool" of pretenders that are chosen from. Also, exposing the various decision formulas that the AI uses in files that can be modified would be great. It would help out illwinter by getting a bunch of people working on improved AI rules while letting illwinter focus on their great game engine, etc.

Originally posted by Chris Byler:
<blockquote>quote:<hr />Originally posted by Kristoffer O:
<blockquote>quote:<hr />Originally posted by Alneyan:

I would like a confirmation: is Dominions AI cheating? I would tend to believe it is not the case. If so, Dominions AI is really not that bad, and is one of the few AIs that does not rely on extra resources to provide a challenge. And I will be then in the side of "The AI isn't that bad, really". The AI is not cheating on normal level. At higher levels it gets more design points and earns more money. Thats about it.

I got an idea regarding AI dependancy on LI when I was away. Vacations are 'foyson' for the mind. http://forum.shrapnelgames.com/images/icons/icon7.gif

Regarding fort construction. I'm not sure how the AI works, but where would you build a fortress? Consider fort type, nation, geography, income, resources etc.

What is the most important matter? How should they be weighted? How much shall current wars affect the spending of time and money. What army should build the fortress?

I wouldn't mind a numerical evaluation of this such as:
Castle cost / 2 < Income + res + gem income x 25 + neighbors x 5
Add a couple of other conditions.

Just to make you think. http://forum.shrapnelgames.com/images/icons/icon7.gif Eventually it might result in something good. <hr /></blockquote>Some responses to these questions:

1. In your list of factors you don't list population type and sites in the candidate province. I often fort to enable building more of an indep troop (especially amazon sacred cavalries, which are all useful; but also e.g. heavy cavalry for Caelum or C'tis, longbows/crossbows for any nation with weak missile troops) and/or protecting a location with high gem income or useful mages (either pop based (amazons again) or site based).

A good first rule is that if you can't spend all your gold buying useful units in forts you already have, you need a new one; if your units are frequently starving because of insufficient supplies far from your forts, you need a new one in that area. (Which reminds me - supply production can be a factor in fort location too.)

Rarely build a fort next to another fort (it is occasionally worth it for something that can be only found in that province, for example I will always fort a province that contains Mount Chaining in Dom I - in Dom II it isn't as powerful but is still pretty good if you are a blood using nation).

2. You don't need an army to build a fort. I usually don't use one. There is some risk in building a fort where you don't have an army, but I find it is often better to have the army doing something else - either taking more indeps (if they are available) or distracting the enemy so he doesn't have time to Call of the Winds every place I might be building a fort. Get some province defense though - it's annoying to have your fort sieged by Call of the Winds (and of course even more annoying to have the construction interrupted).

3. Fort selection value: (Adjustment * (resources + supplies)) + (gems x 10) + value of rare units recruitable here (this value should depend on the rarity and power of the unit; longbows and crossbows count a little, amazon cavalries and mages count more, something like grey knights would count a lot; mages with paths different from your nation are more valuable than mages that duplicate your nation's paths) + misc site value (enter-to-summons, cost reducers).

I don't generally take income into account directly (although the first two terms value high population provinces more). "Adjustment" depends on the admin and gold cost of the castle; higher admin and cost should raise the value of the adjustment. Adjustment = 0 if there is an adjacent fort (this will reduce fort-next-to-fort except when gem income, rare unit value, or site value are high, which are the exceptions to the no-adjacent-fort rule).

4. Once you have built a fort, build a temple. Temples are vulnerable to enemy attack, but in a fort they are far less vulnerable. Then they spread your dominion, let you recruit priests and preach better. Mictlan might be an exception to this rule because their temples don't radiate - but blood sacrifices are more effective when performed in temples (IIRC), so they might want to build temples anyway so that they can sacrifice there. It's better to preach or sacrifice inside a fort because it protects the priest (high level priests especially are pretty expensive).

Build a lab if you are likely to want to recruit any mages (if you are fortifying because of local troops or sites you probably want a lab for that anyway; some nations like Marignon always want to recruit mages) or summon/construct troops there to supplement your recruited troops (C'tis usually wants to do this; but Desert Tombs C'tis can do without the lab, if you don't mind shipping in your unpriests). <hr /></blockquote>

Mortifer

November 4th, 2003, 09:17 AM

I think it is not as important now. We have a nice list about the AI weaknesses, so that feature isn't needed. http://forum.shrapnelgames.com/images/icons/icon7.gif

Aristoteles

November 4th, 2003, 11:24 AM

Okay, here is the list once more, that what should be upgraded in the strategic AI:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

Feel free to add more, however I cannot think anything else right now.
Nerfix has posted that the AI is building too many temples, I don't know that this is good or bad.

Serpico

November 4th, 2003, 11:57 AM

Hi, I am far from the experienced status, but I've seen these problematic points,-what you've mentioned on your list-, in my games with the demo.
I hope that this game developer -Illwinter- will improve these parts of the AI.
I love the singleplayer at least as much as the multiplayer.

Gandalf Parker

November 4th, 2003, 04:50 PM

OK I think we still have time for this. If I modify the Aran .map file (eye is overused already) to provide the AI some of these items do you think it would clairify things abit?

1. The AI is massing weak troops

How about if each AI started the game with another army of hvy and elite units? Or even 2 more armys.

2. The AI wont protect his pretender/commanders/priests/mages

If I actually BUILT the pretenders for each AI then I could assign bodyguards to it. I can also bodyguard priests and mages but only if the .map file gives them.

3. The AI wont build forts

I could start the AI's with a couple of additional provinces and forts/temples/labs

4. The AI's battlefield tactics could be better

Hmmm I dont see where I could help this much. I could toss in a bunch of ally commands so that the AI's dont waste time and resources fighting each other.

5. The AI is making very odd things. IE. wandering around with his pretender alone etc.

Hmmm bodyguards is all I can think of.

6. The AI won't stay in a province, it is always moving his armies.

I think this is a good tactic for an AI. If you cant make it capable of good decisions based on whats coming then moving around alot helps.

7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.

Not sure what can be done here unless a new map image (tga file) is created with more bottlenecks.

8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

That would tend to be very late game but if I create the pretenders I CAN also give them magic items.

The problem with so much pre-placement on such a map is that it wont have alot of replay value. The AIs would always start in the same provinces.

Also, should I only do this for the AI's that cannot be selected for human play? Or should I do it for all the races. If its all, then the human player will also have his pretender pre-selected and extra troops at start.

[ November 04, 2003, 15:51: Message edited by: Gandalf Parker ]

licker

November 4th, 2003, 05:21 PM

While I appreciate your work Gandalf, I'm not sure that simply making map senarios is really what the folks are after here (though I expect you know that).

The fixes you propose (and you say this at the end of your post) are merely artificial boosts to try and make a specific map more difficult. I'm all for that, I love user created senarios that make life more difficult, but in the end, its not an improvement to the AI (unless there is the possibility for some *massive* scripting).

So I just ask that while your work is not done in vain, the focus of this discussion remains on how to improve the in game AI, not on how to make a senerio that is challenging.

Respecfully submitted http://forum.shrapnelgames.com/images/icons/icon7.gif

Gandalf Parker

November 4th, 2003, 05:44 PM

Originally posted by licker:
While I appreciate your work Gandalf, I'm not sure that simply making map senarios is really what the folks are after here (though I expect you know that).

Respecfully submitted http://forum.shrapnelgames.com/images/icons/icon7.gif Respectfully accepted. I have started a new thread for discussing a .map fix for these things.

Personally Im still not convinced that these are "problems" with the whole game. If they are based on settings then fixing some of these would cause problems when played with other settings (just a possibility). And if anything can be fixed with map mods then Id prefer that over a game rewrite since it would allow more variance in what can be accomplished with maps. Fixing these items might remove this level of AI play when some map makers might wish it.
Respectfully submitted http://forum.shrapnelgames.com/images/icons/icon7.gif Really

licker

November 4th, 2003, 06:09 PM

Originally posted by Gandalf Parker:
<blockquote>quote:<hr />Originally posted by licker:
While I appreciate your work Gandalf, I'm not sure that simply making map senarios is really what the folks are after here (though I expect you know that).

Respecfully submitted http://forum.shrapnelgames.com/images/icons/icon7.gif Respectfully accepted. I have started a new thread for discussing a .map fix for these things.

Personally Im still not convinced that these are "problems" with the whole game. If they are based on settings then fixing some of these would cause problems when played with other settings (just a possibility). And if anything can be fixed with map mods then Id prefer that over a game rewrite since it would allow more variance in what can be accomplished with maps. Fixing these items might remove this level of AI play when some map makers might wish it.
Respectfully submitted http://forum.shrapnelgames.com/images/icons/icon7.gif Really <hr /></blockquote>I agree that there are some (balance mostly) issues that may be negatively impacting the AI right now. However, I don't think that only modding is going to be the answer to 'fixing' the AI, unless the modders are able to somehow change the priorities of the AI in game. It is fine to make maps that should be more challenging, but it still doesn't strike at some of the underlieing (supposed) weaknesses. Starting the AI with more HI is fine, but if they don' continue to build more HI the advantage is soon lost (well depending on how many HI you start them with). Taking LI away from them completely (assuming you can do that by modding the units files) is another work around, but its also not really satisfactory in making the AI 'smarter'.

Alot of the discussion in this thread has tried to uncover the mechanisms by which the AI is making its army build selections, already it has been uncovered that perhaps the AI isn't building enough forts to allow for better troop recruitment, maybe the AI doesn't understand how to save money to build what it needs to, I don't know, but issues like that can only be 'patched' by scripting maps, they need to be fixed by some tweeking of the inputs to the AI, that's what I'm hoping is possible, as I don't want a complete rewrite, just the ability to tweek existing inputs.

MStavros

November 4th, 2003, 06:10 PM

Gandalf, you are avery helpful person indeed. http://forum.shrapnelgames.com/images/icons/icon7.gif
However this won't help too much. It will offer more challenge, but the AI itself won't be changed, thus the problems won't be solved this way.
Still I love your map, especially with indy str 9. http://forum.shrapnelgames.com/images/icons/icon10.gif

-Storm-

November 4th, 2003, 08:29 PM

The AI won't be better with these scripts. All they do is that you will have a harder start.
Still it can help until the AI is being fixed.

November 4th, 2003, 10:30 PM

Originally posted by licker:
I agree that there are some (balance mostly) issues that may be negatively impacting the AI right now. However, I don't think that only modding is going to be the answer to 'fixing' the AI, unless the modders are able to somehow change the priorities of the AI in game. It is fine to make maps that should be more challenging, but it still doesn't strike at some of the underlieing (supposed) weaknesses. Starting the AI with more HI is fine, but if they don' continue to build more HI the advantage is soon lost (well depending on how many HI you start them with). Taking LI away from them completely (assuming you can do that by modding the units files) is another work around, but its also not really satisfactory in making the AI 'smarter'.

Alot of the discussion in this thread has tried to uncover the mechanisms by which the AI is making its army build selections, already it has been uncovered that perhaps the AI isn't building enough forts to allow for better troop recruitment, maybe the AI doesn't understand how to save money to build what it needs to, I don't know, but issues like that can only be 'patched' by scripting maps, they need to be fixed by some tweeking of the inputs to the AI, that's what I'm hoping is possible, as I don't want a complete rewrite, just the ability to tweek existing inputs. I agree with most of what you said. Pre-scripting the AI conditions is pretty much the same as giving the AI the ability to cheat. So, for example that on impossible difficulty the AI gets free units, etc. This is what most of the games do to increrase difficulty, from RTS onwards, but this is not really what most people were saying that they want in this thread. Beside, one of the big pluses in Doms II is that you can start the same map many times over and have a vastly different game due to other factors being fairly randomized, such as starting locations, richness, sites, etc. Hardcoded scenarios are not really a solution in terms of replay possibilities (one of the greatest strengths of this game, IMHO).

The thing where I don't agree is that externalizing the AI input would solve the problem. I would much prefer it to be done in the vanilla game than to have mods to do it. Maybe it's just me, as I never found mods to be attractive: there are always things that I don't agree with in them, and I don't feel compelled to accept that in the same way I think about the vanilla game, so I never play them. That's not to say that mods wouldn't help input-wise, as people could have more exact suggestions to give to the devs, but I would like nothing better than to see it improved in the actual vanilla (patched) game.

licker

November 4th, 2003, 10:51 PM

Originally posted by HJ:

The thing where I don't agree is that externalizing the AI input would solve the problem. I would much prefer it to be done in the vanilla game than to have mods to do it. Maybe it's just me, as I never found mods to be attractive: there are always things that I don't agree with in them, and I don't feel compelled to accept that in the same way I think about the vanilla game, so I never play them. That's not to say that mods wouldn't help input-wise, as people could have more exact suggestions to give to the devs, but I would like nothing better than to see it improved in the actual vanilla (patched) game. Well my case for letting the mods at it wasn't to ignore the vanilla game, but to allow the devs to let the modders do alot of the work for them http://forum.shrapnelgames.com/images/icons/icon7.gif

I am all for a 'Gold Standard' in the Vanilla game, I just think you'll reach a better one faster if you have several people able to change parameters than only a few, even if those few are the only ones who really matter http://forum.shrapnelgames.com/images/icons/icon7.gif (sucking up to the devs can't hurt right? http://forum.shrapnelgames.com/images/icons/icon12.gif )

Its easier for the devs if people can tell them that by changing the ratio of HI to LI, or the amount of resources earmarked for mages by some number has this effect, rather than the Devs having to make a change, see what everyone thinks, then go back and tweek it again. Besides don't write off mods too quickly, while there is a need for greater consensus for the MP crowd, the SPers are free to dabble and use whatever strikes their fancy, most serious mods have enough details to let a new user know ahead of time whether its worth their down load or not http://forum.shrapnelgames.com/images/icons/icon7.gif

November 4th, 2003, 10:59 PM

Originally posted by licker:
Well my case for letting the mods at it wasn't to ignore the vanilla game, but to allow the devs to let the modders do alot of the work for them http://forum.shrapnelgames.com/images/icons/icon7.gif

I am all for a 'Gold Standard' in the Vanilla game, I just think you'll reach a better one faster if you have several people able to change parameters than only a few, even if those few are the only ones who really matter http://forum.shrapnelgames.com/images/icons/icon7.gif (sucking up to the devs can't hurt right? http://forum.shrapnelgames.com/images/icons/icon12.gif )

Its easier for the devs if people can tell them that by changing the ratio of HI to LI, or the amount of resources earmarked for mages by some number has this effect, rather than the Devs having to make a change, see what everyone thinks, then go back and tweek it again. Besides don't write off mods too quickly, while there is a need for greater consensus for the MP crowd, the SPers are free to dabble and use whatever strikes their fancy, most serious mods have enough details to let a new user know ahead of time whether its worth their down load or not http://forum.shrapnelgames.com/images/icons/icon7.gif In that case I agree completely with your previous post. As I already said, input-wise role is indeed very desirable, it would then be something more like an open beta for the AI. If nothing else, it would speed up the process of improving the AI for the patches, and make it easier for the devs by giving them time to implement other things while people are dabbling with the AI, as they would just test the suggestions to see whether they'll implement them or not.

DominionsFan

November 5th, 2003, 09:35 AM

Hm yeah good idea. I know some games, where you can script AIs.

Btw, why no answer from the devs about the strategic AI problems? http://forum.shrapnelgames.com/images/icons/confused.gif http://forum.shrapnelgames.com/images/icons/icon9.gif

Patrik

November 5th, 2003, 10:39 AM

What are you talking about DominionsFAN? KO posted in this thread less than 3 days ago.

Originally posted by Kristoffer O:

I got an idea regarding AI dependancy on LI when I was away. Vacations are 'foyson' for the mind. http://forum.shrapnelgames.com/images/icons/icon7.gif

Also, IW has asked for save files showing flagrant AI mistakes. What kind of 'answers' are you asking for anyway? I certianly hope it's not these

Originally posted by DominionsFAN:
Kris, copy and paste that list by Aristoteles, and try to fix those problematic points step by step. Oh and don't foget to reply here, that how is it going! http://forum.shrapnelgames.com/images/icons/icon12.gif Apart from this post bordering on offensive (It's not up to you to decide Illwinter's priorities), this is from 2 days ago!! What do you think will have happened since then??!

NTJedi

November 5th, 2003, 12:05 PM

The most important way to improve the AI is to give the AI different personalities since it cannot improve its game strategy as humans can.

Otherwise no matter how sophisticated an AI opponent is created... gamers will soon find weaknesses or exploits which turns the AI pathetic. IF random personalities were given to each AI opponent at the start of each game then human players will be uncertain as to what weaknesses are currently exist for each opponent.

The only other solution would be to provide a MOD which would allow gamers/map makers to improve the AI using script commands.

[ November 05, 2003, 14:00: Message edited by: NTJedi ]

Karacan

November 5th, 2003, 02:09 PM

Originally posted by Alneyan:
Please name one game where the AI is able to do such a thing, or come even close to that. Warlods III + Darklord's Rising. http://forum.shrapnelgames.com/images/icons/icon7.gif

I'd like to see different AI personalities, too, but for now I'd prefer a stable and challenging default one, myself.

MythicalMino

November 5th, 2003, 02:20 PM

but they could change their "personalities" mid-game though?

if so, impressive....

von_Schmidt

November 5th, 2003, 02:24 PM

Originally posted by Alneyan:
<blockquote>quote:<hr />Originally posted by Zerger:
Different personalities? Maybe...but it would be effective, if the AI itself could change the various personalities in-game, depending on the situation. Please name one game where the AI is able to do such a thing, or come even close to that. . <hr /></blockquote>In Galactic Civilisations, the opponents have different AI personalities...

Or is this more a question if the AI can switch strategies midgame, depending on the changing situation?

von Schmidt

von Schmidt

Alneyan

November 5th, 2003, 02:34 PM

Originally posted by cpbeller:
but they could change their "personalities" mid-game though?

if so, impressive.... Let me add: "Depending on the situation" after your question. That is what was asked for Dominions. (That is the same for you Von Schmidt) Games with "personalities" exist, the problem being to script for a single person a working AI able to anticipate and outsmart the player (that is, able to change its tactics and priority, or personality, during the game, depending on what is going on), and even without using plain cheating. (Say, omniscient AIs)

I never liked Warlords myself, but this AI is said to be good. Then you are right, it comes close to what was asked, with a less complex game. http://forum.shrapnelgames.com/images/icons/icon12.gif

Karacan

November 5th, 2003, 02:37 PM

Originally posted by cpbeller:
but they could change their "personalities" mid-game though? I am pretty sure if an incident where an "Expansionist" became a "Razer" type of personality, burning down the cities down and pressing onward, rather than trying to hold them, which was the behaviour I expected from an "Expansionist". (The names are made up now, it's been a long time since I played Warlords, but you'll get the essence.)

Anyway, that's all cool, but let's talk about different personalities and changing them when there's a good one to build upon.

MStavros

November 5th, 2003, 02:50 PM

Originally posted by Alneyan:
<blockquote>quote:<hr />Originally posted by cpbeller:
but they could change their "personalities" mid-game though?

if so, impressive.... Let me add: "Depending on the situation" after your question. That is what was asked for Dominions. (That is the same for you Von Schmidt) Games with "personalities" exist, the problem being to script for a single person a working AI able to anticipate and outsmart the player (that is, able to change its tactics and priority, or personality, during the game, depending on what is going on), and even without using plain cheating. (Say, omniscient AIs)

I never liked Warlords myself, but this AI is said to be good. Then you are right, it comes close to what was asked, with a less complex game. http://forum.shrapnelgames.com/images/icons/icon12.gif <hr /></blockquote>Yes Dominions is more complex, but script these AIs are not impossible, but it will require time.

I don't think that we need different personalities. An upgraded AI, yes.

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

Upgrading/fixing/enchancing these won't require different personalities. A core AI [like the current Doms 2 AI] could do it, if the scripts will be tweaked.

[ November 05, 2003, 13:51: Message edited by: MStavros ]

NTJedi

November 5th, 2003, 03:13 PM

MStavros ....

A 'single' upgraded AI for all computer opponents will not work in the long-term because gamers will eventually find flaws/exploits/weaknesses. These will eventually be posted and the developers can't patch the game forever! If human players don't know what to expect from AI opponents because of different personalities the game will be more challenging.

The only other option would be to allow gamers/map_makers a mod tool for adjusting/improving AI strategies.

_____________________________________
edit: adding story
_____________________________________

EXAMPLE of Dominions_II with a 'single' AI personality:

Experienced Gamer vs AI:
So here's today's story if anyone is still playing the game as single players. I brought my 400 troops to his Last fortress while small armies sat on the borders since we all know the AI throws 90% of its men into the fortress when the pretender god is threatened. Next turn walked in and killed him... time to move on.

~~~~~~~~~~~~~~~~~~~
~~~~~~~~~~~~~~~~~~~

EXAMPLE of Dominions_II using a randomly selected AI personality out of 10 others:

Experienced Gamer vs AI:

Hello everyone again... today the AI really caught me by surprise since I believed him to have the Arch-Mage personality... fudge was I wrong! As I blindly marched my 400 troops to finish off his pretender god there was 3 strong assassin armies that came out of hiding and stole two territories right next to my fortress. With both our pretender gods surrounded it looks like both of them will be dying. Yikes!

[ November 05, 2003, 14:34: Message edited by: NTJedi ]

Particle

November 5th, 2003, 03:52 PM

Jedi...what about: do not use exploits. You will cheat yourself, not just the AI.
The suggestion is good, but it would take huge amount of time to script IMHO. Upgrading the AI itself will be a hard task as well, but of course a must have.

licker

November 5th, 2003, 04:30 PM

I don't think 'personalities' are that hard to do, nor to switch during the game. In fact earlier in this thread I explained one way to approch it with the externalized AI parameters (MoO3 did this, and it actually worked, though MoO3 had other problems...)

The scripting of the personalities isn't even that difficult if you don't try to get too fancy with it. The personalities are basicly templates that provide the AI with %s of resources to spend on commanders, army and magic. Within those two broad catagories there are several subcatagories. For army its balance of light troops, medium troops, heavy troops, and summons of those classes (you can break out archers and support as well), for magic its what schools to research (for particualr spells) based on your existing ability to use the schools, and how much do devote to research v. summoning (tied to army) v. other spell (direct damage, search, overland...), for commanders its what ratio of Mages to priests to assasins/scouts to infantry/cavalry leaders...

There are alot of matricies to be made, and there are alot of values to add in, and there are even alot of interdependancies between the matricies, but once you've mapped that out, the actually application is easy. Then its up to the devs and players to tweek and mod to find what values work best for what races.

Kris has mentioned (elsewhere) that he thinks its an interesting idea, but he's not sure how much of it can be easilly done given the existing Dom2 code. I know they are taking this seriously, but it is a very tricky thing to do, especially since the Dom2 code wasn't first written with this in mind. Anyway, keep up the suggestions, and try to be more specific then general, there's been enough generalities over the Last few weeks to kill a horse http://forum.shrapnelgames.com/images/icons/icon12.gif

Gandalf Parker

November 5th, 2003, 06:09 PM

I think this might be the best direction to go. Multiple personalities written by the programmers is probably a faster action than externalized scripting for Users. Possibly even faster than improving the AI thinking of the game in general.

One big advantage is that the "wrong decisions" of the AI player can be minimalized by not allowing it to go too far in a particular direction. And the challenge can be improved by having the AI player change its tactics even if the tactics it switches to is not a particularly good one.

In this situation even a bad tactic has an advantage of being a surprise. Such as, Arcos might be given a preference for using the "build a mage then spend the rest on elite troops" style. The "buy lots of cheap units and rush outward until they are all dead" strategy would be bad for them. However, pursuing the mage-research thing wouldnt be good to do TOO much of and having a 5-turn run of barbarian horde tactics can be a good thing for them to do occasionally. On the other hand Ermor might be better with a preference for horde and an ocassional mage/research shift.

So the disadvantages of any strategy can be lessened by shfting to another for short periods without having to try and come up with perfect smart decision-making for the AI. Probably not what you wanted this to do. I know it can be done better. Im just pointing out a possible advantage.

[ November 05, 2003, 17:11: Message edited by: Gandalf Parker ]

licker

November 5th, 2003, 06:17 PM

I agree with you Gandalf, though I'm not sure that randomly swapping strategies is the best way to go. I think that there can be a rubric by which the AI can decide if a certain strategy is more needed at a certain time. An example of this would be in the early game, where expansion is key, basically you'd want all AIs to be expanders (though their means of expantion could differ) until they reach boarders of other nations. At that point you have a decision point, do I aggressively attack my neighbor(s), do I hunker down and focus on research or summoning, or building an elite force, do I ignore my neighbor and continue picking up indies, ...

The decision made *should* be able to be influanced by looking at the current compostion of the AIs nation. If the nation is filled with poor provinces more wieght to an early offensive, if the nation has rich provinces, more weight to building up a superior force, if the nation has high gem income, or lots of indie mages them more weight to going down a magic path.

This is another area where externalizing the inputs to make these decisions would be useful so that modders can tweek them to get more aggressive AIs for various nations. Different nation need different weights for these decisions to start with (and the use of SCs should be in there too somewhere...).

I think these discussions are getting more useful all the time http://forum.shrapnelgames.com/images/icons/icon7.gif Lets keep working on specifics, the Devs have asked for that time and time again http://forum.shrapnelgames.com/images/icons/icon7.gif

Mortifer

November 5th, 2003, 06:41 PM

Humzzz I like the idea as well, the question is: this can be added to Doms II.? The AI will be really better or just more diverse?
IMHO the key is, that the AI must be lot more effective. It will be more effective this way?
Frankly I have no idea.

Gandalf Parker

November 5th, 2003, 09:07 PM

Originally posted by Mortifer:
Humzzz I like the idea as well, the question is: this can be added to Doms II.? The AI will be really better or just more diverse?
IMHO the key is, that the AI must be lot more effective. It will be more effective this way?
Frankly I have no idea. I think it would definetly be more effective. Many of the things on the list of AI suggestions seem to be that the AI does too much or not enough of something. With a system like this it can be gauranteed to change that. Maybe not to the perfect level but at least not completely not happening. With so many different races and different game paramters Im not sure it would ever be possible to achieve a smart AI.

Rule #1 of AI programming. If you cant beat them, cheat.

Rule #2 of AI programming. If you cant be smart, be random.

Rule #3 of AI programming. If you cant please the Users with #1 and #2 then declare the AI to be for tutorial purposes only toward the MP part of the game, and give up.

(HEY! calm down. Im only half serious)

[ November 05, 2003, 20:08: Message edited by: Gandalf Parker ]

Aristoteles

November 5th, 2003, 09:20 PM

I still say that tweaking the main AI would be the best, but I am not sure..I mean these AI personalities sounds good and all, but I think this won't be added. Well not now..maybe not at all. I guess that would take countless time to script, and the success is not sure.
However...if that would boost the brain of the AI.. http://forum.shrapnelgames.com/images/icons/icon10.gif

[ November 05, 2003, 20:21: Message edited by: Aristoteles ]

licker

November 5th, 2003, 09:43 PM

Originally posted by Aristoteles:
I still say that tweaking the main AI would be the best, but I am not sure..I mean these AI personalities sounds good and all, but I think this won't be added. Well not now..maybe not at all. I guess that would take countless time to script, and the success is not sure.
However...if that would boost the brain of the AI.. http://forum.shrapnelgames.com/images/icons/icon10.gif Well the personalities may not make it becasue its too hard to redo the code to fit them, but if certain things can be easilly tweeked by the devs, or externalized then it would be a fairly simple process I think. I'm not sure you'd want to call the personalities scripts either, that doesn't seem the most accurate way to look at them. Though you could do it that way, just that I don't think it would be very elegant. http://forum.shrapnelgames.com/images/icons/icon7.gif

Anyway, when you say tweeking the main AI what do you mean exactly? and how would you do it? I think that the personalities or externalized paramaters are tweeks to the main AI. Though I also feel that each nation (and even each theme for a nation) should have it's own AI. Of course if you do this via a series of matricies its not that hard, though it is potentially alot of matricies...

Gandalf Parker

November 5th, 2003, 10:36 PM

Anyway, when you say tweeking the main AI what do you mean exactly? and how would you do it? I think that the personalities or externalized paramaters are tweeks to the main AI. Though I also feel that each nation (and even each theme for a nation) should have it's own AI. Of course if you do this via a series of matricies its not that hard, though it is potentially alot of matricies... Thats part of the problem I see. Just for troops it would be a huge matrix. For each race there are different units, plus all the independents, plus different situations such as small game, large game, low resource, high magic, independent strength, which opponent.

Ive helped program an AI for a MUD. Using a programmable client to create a player on the MUD we programmed it into what people thought was a pretty amazing robot "player". The bot program was as large and used as much CPU as the MUD program itself. Luckily by running from a completely different computer as a player that wasnt a problem. Im afraid that trying to create so many different smart AIs (I mean the improve AI code, not the personalities project) would make the game 20 times bigger.

licker

November 5th, 2003, 10:50 PM

It is alot of matricies theoretically. Someone smarter than I am can probably figure out a way to crunch them down. But they are just text files, not that big. For any given game, only a few are needed anyway, unless you are playing with all 17 nations, but even then only 16 are needed.

Sure you need to have made the others, but as far as which are used during a given game, its not that many.

Anyway, it is probably possible to make only a few matricies for army composition, or magic research... and then have a few operators for different nations, or different personalities that modify them. Once the game has started the modified (temporary) matrix is the only one used anyway. The .mob file from MoO3 was pretty big if I remember correctly (the file that stored all the matricies), but once you 'extracted' it you had a series of text files in a series of folders that you could easily open and edit in excel or even word pad. I mean it adds more size to the game, but I don't think it requires more processing power. And I don't think it would add more than 20MB in a worst case anyway.

I guess I don't quite see how this is comparable to a bot.

licker

November 5th, 2003, 11:01 PM

Bah, I didn't want to do this but I'll try... http://forum.shrapnelgames.com/images/icons/icon7.gif

Do you remember linear algebra? http://forum.shrapnelgames.com/images/icons/tongue.gif (I do and wish I didn't but that's beside the point)

So you have a matrix of units with weights and it looks like this:

Unit Weight
Militia 10%
LI 10%
MI 40%
HI 25%
X-bow 15%

Then you chose a rapid expander type personality and the matrix gets multiplied by:

(.5
1.5
2
0
1)
And then it gets normalized to 100%...

You further have a SC pretender so that multiplies the matrix by:
(.5
.5
3
1
1)
And it gets normalized again...

Anyway, you have a large matrix for unit type, then you have these operators (if that's the right term) for different conditions, like personality, like theme, like nation that multiply the values in the big matrix.

Actually if you did it this way you could have one huge list of all units, multiplied by national operators (to zero out the unallowed units), multiplied by theme and personality operators, multiplied by exisiting indie units (to put back in desired indie units when you get them), multiplied by...

Do you see it now? Its not that difficult to set up, it is difficult to balance http://forum.shrapnelgames.com/images/icons/icon7.gif The size of these files is tiny, though there may be 100 of them. The tricky part then comes in how you set up your algorythem for which operators are applied to your specified matricies. But again, that's not difficult to concieve how you set up those algorythems, its just difficult to balance... that's where the want to externalize all this comes from, let the players who want to fiddle fiddle, eventually people will arrive at settings that work 'best' and the devs can chose to use them for the vanilla game or not.

NTJedi

November 5th, 2003, 11:35 PM

Aristoteles and Particle...

Actually multiple AI personalities shouldn't take that long to program since for the most part just variables would have to be adjusted such as percentages regarding research, ritual spells, crafting items, etc... Also percentages regarding the use of gold for purchasing of assassins, spellcasters, priests, buildings, mercenaries, province defense, or something else...
Another would be percentages regarding how aggressive, defensive, productive, or (other actions I may miss)...

Once the formula for each is in place the developers should be able to create at least 10 very different personalities.

Another very important feature about this is if one of them has a flaw it won't destroy the game as computer opponents are randomly assigned the personalities before each game.
Whereas if only one AI personality exists and a major flaw/weakness appears gamers have to suffer until the next patch. Let me say again that a single AI can easily be mastered by gamers... multiple unknown personalities will leave gamers guessing as how the AI opponent will behave.

by Particle
Jedi...what about: do not use exploits.
Actually I never do... however during multiplayer games you can't always trust opponents across the internet. That is why I wrote what I did.

[ November 05, 2003, 22:41: Message edited by: NTJedi ]

Zerger

November 6th, 2003, 01:16 AM

Different personalities? Maybe...but it would be effective, if the AI itself could change the various personalities in-game, depending on the situation.

Alneyan

November 6th, 2003, 01:31 AM

Originally posted by Zerger:
Different personalities? Maybe...but it would be effective, if the AI itself could change the various personalities in-game, depending on the situation. Please name one game where the AI is able to do such a thing, or come even close to that. And especially a game as rich and complex than Dominions. (I believe it would be easier for a RTS) Or say how you would do that. You would need an AI able to spy on the human player and then it would have to process that knowledge to find the weaknesses in the human army/provinces/pretender/add as needed, for afterwards change its strategy to take that data into account. And you would have to prevent the player from cheating the AI that way as well.
Lastly, how a single AI scripter is supposed to implement such an AI for Dominions? (Assuming you do have someone who only works on the AI)

I would like different personalities for the AI, as this idea would be much more doable. Maybe it could even be a field in which players could make their tweaks as well.

[ November 05, 2003, 12:33: Message edited by: Alneyan ]

MythicalMino

November 6th, 2003, 01:47 AM

we have had it in a couple of movies before....Matrix.....oh, and Terminator....

would be the perfect blend of a sci-fi AI with a fantasy tbs strategy game....

MStavros

November 6th, 2003, 09:43 AM

Aha. I love this personalities idea. Hopefully the devs will find it good as well.
Randomly assigning it before the game for the AIs?
I have a question. Is there a way to give multiple personalities to the AI than? I mean for one nation? Because THAT would be the best.
Different situations IE. war, research etc. -> different AI personality in use.

-Storm-

November 6th, 2003, 12:45 PM

Originally posted by NTJedi:
Aristoteles and Particle...

Actually multiple AI personalities shouldn't take that long to program since for the most part just variables would have to be adjusted such as percentages regarding research, ritual spells, crafting items, etc... Also percentages regarding the use of gold for purchasing of assassins, spellcasters, priests, buildings, mercenaries, province defense, or something else...
Another would be percentages regarding how aggressive, defensive, productive, or (other actions I may miss)...

Once the formula for each is in place the developers should be able to create at least 10 very different personalities.

Another very important feature about this is if one of them has a flaw it won't destroy the game as computer opponents are randomly assigned the personalities before each game.
Whereas if only one AI personality exists and a major flaw/weakness appears gamers have to suffer until the next patch. Let me say again that a single AI can easily be mastered by gamers... multiple unknown personalities will leave gamers guessing as how the AI opponent will behave.

<blockquote>quote:<hr /> by Particle
Jedi...what about: do not use exploits.
Actually I never do... however during multiplayer games you can't always trust opponents across the internet. That is why I wrote what I did. <hr /></blockquote>Ok, but what AI personalities are you talking about?
Can you make a list of examples?

Gandalf Parker

November 6th, 2003, 04:25 PM

Nice breakdown. As far as choice of targets there would also be AIs who would always choose to target the WEAKEST pretender, magic, units, etc. Especially if they are a neighbor. Of course thats kindof how it does it now but with a list of AIs those should be included.

Mortifer

November 6th, 2003, 04:49 PM

Originally posted by NTJedi:
Example=
Destroyer(Pillages lands which have been owned by opponents for long periods of time)
Architect(focuses on improving lands)
Arch-Mage(focuses on magic)
Barbarian(mainly uses brute force)
Insanity(unpredictable actions= randomly flips between all personalities every 10 turns)
Cutthroat(one out of three armies will be sneaking during movement)

With all the different levels of magic, game features and type of units... 10 very different personalities shouldn't be too difficult. If gamers could also modify these percentages with a mod tool... I'm sure we would see lots of very different ones created.

____________________________________

* On a side note:
Most of the AI personalities should be designed to attack the most powerful pretender god. One persoanlity may attack the player with the strongest domain while another may attack a player with the most research and so on. I've played lots of games where once some human player became the most powerful the remaining AI opponents would fight amongst themselves instead of joining forces. Hm nice ideas here. But still I think this would really work, if the AI would change these in the game, as someone has mentioned. Do you think that it is possible?

Gandalf Parker

November 6th, 2003, 05:36 PM

[much that isnt necessary to read again snipped here] *hint hint*

Hm nice ideas here. But still I think this would really work, if the AI would change these in the game, as someone has mentioned. Do you think that it is possible? Ive done programming like that before and the switching actually seems the least difficult. The biggest question that I would have is.... does the game programming seperate the AIs at all? The way the game runs now I see a strong possibility that the AI might be one long code string that does alot of common stuff first and then branches off (maybe) for things that apply only to particular nations. If its written with the basic AI code duplicated to each nation and then small modificatons done, then going to scripted AIs would be much easier. Thats not normal coding practices though so our chances are slim for that advantage.

[ November 06, 2003, 16:36: Message edited by: Gandalf Parker ]

Particle

November 6th, 2003, 07:24 PM

Originally posted by Gandalf Parker:
The biggest question that I would have is.... does the game programming seperate the AIs at all? Good question. Maybe JK knows the answer. The question is: We will know it? http://forum.shrapnelgames.com/images/icons/tongue.gif

johan osterman

November 6th, 2003, 07:43 PM

Eh, chances that the AI will ever have different personalities are slim, chances for the AI to change personalities during the game are even slimmer.

Any AI improvements are likely to be solutions to specific problems, not improvements on the scale an adapting multi personality AI is.

Zerger

November 6th, 2003, 10:02 PM

Originally posted by johan osterman:
Eh, chances that the AI will ever have different personalities are slim, chances for the AI to change personalities during the game are even slimmer.

Any AI improvements are likely to be solutions to specific problems, not improvements on the scale an adapting multi personality AI is. I thought so. Just update these and we will be fine:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

http://forum.shrapnelgames.com/images/icons/icon12.gif

NTJedi

November 7th, 2003, 01:06 AM

Example=
Destroyer(Pillages lands which have been owned by opponents for long periods of time)
Architect(focuses on improving lands)
Arch-Mage(focuses on magic)
Barbarian(mainly uses brute force)
Insanity(unpredictable actions= randomly flips between all personalities every 10 turns)
Cutthroat(one out of three armies will be sneaking during movement)

With all the different levels of magic, game features and type of units... 10 very different personalities shouldn't be too difficult. If gamers could also modify these percentages with a mod tool... I'm sure we would see lots of very different ones created.

____________________________________

* On a side note:
Most of the AI personalities should be designed to attack the most powerful pretender god. One persoanlity may attack the player with the strongest domain while another may attack a player with the most research and so on. I've played lots of games where once some human player became the most powerful the remaining AI opponents would fight amongst themselves instead of joining forces.

[ November 06, 2003, 12:23: Message edited by: NTJedi ]

NTJedi

November 7th, 2003, 04:11 AM

by johan osterman

Eh, chances that the AI will ever have different personalities are slim, chances for the AI to change personalities during the game are even slimmer.

Any AI improvements are likely to be solutions to specific problems, not improvements on the scale an adapting multi personality AI is.
I understand it's probably too late in the games programming design for multiple personalities to be added. It's too bad really because gamers will eventually master this one default Artificial_Intelligence. I can only recommend that Dominions_III have the multiple personalities for AI opponents.

[ November 07, 2003, 03:11: Message edited by: NTJedi ]

Aristoteles

November 7th, 2003, 09:50 AM

Originally posted by Zerger:
<blockquote>quote:<hr />Originally posted by johan osterman:
Eh, chances that the AI will ever have different personalities are slim, chances for the AI to change personalities during the game are even slimmer.

Any AI improvements are likely to be solutions to specific problems, not improvements on the scale an adapting multi personality AI is. I thought so. Just update these and we will be fine:

1. The AI is massing weak troops
2. The AI wont protect his pretender/commanders/priests/mages
3. The AI wont build forts
4. The AI's battlefield tactics could be better
5. The AI is making very odd things. IE. wandering around with his pretender alone etc.
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.

http://forum.shrapnelgames.com/images/icons/icon12.gif <hr /></blockquote>All I have to say....I agree. http://forum.shrapnelgames.com/images/icons/tongue.gif

Mortifer

November 7th, 2003, 03:42 PM

Originally posted by johan osterman:
Eh, chances that the AI will ever have different personalities are slim, chances for the AI to change personalities during the game are even slimmer.

Any AI improvements are likely to be solutions to specific problems, not improvements on the scale an adapting multi personality AI is. This was predicatble, and I agree with johan. I guess this would be hella lot of work to add. I think we can live without this in Doms II.
Maybe if the Users/players would be allowed to script AIs......they could try.
Otherwise, just upgrade those weak points in the mentioned list, step by step.
If all of those will be enchanced, the AI will rock some you know what. http://forum.shrapnelgames.com/images/icons/icon7.gif

Gandalf Parker

November 7th, 2003, 03:53 PM

All I have to sayDont worry. Im sure that the efforts to keep an un-pinned thread locked near the top of the board by the same maybe-different people saying the same things or reposting the whole list has SURELY accomplished its mission. It makes it seem almost as key a discussion as the pinned polls and neutral master lists. Good job.

[ November 07, 2003, 14:54: Message edited by: Gandalf Parker ]

Mortifer

November 7th, 2003, 04:01 PM

Originally posted by Gandalf Parker:
<blockquote>quote:<hr />All I have to sayDont worry. Im sure that the efforts to keep an un-pinned thread locked near the top of the board by the same maybe-different people saying the same things or reposting the whole list has SURELY accomplished its mission. It makes it seem almost as key a discussion as the pinned polls and neutral master lists. Good job. <hr /></blockquote>I think, that the "AI problems" 'was?' a very important discussion on the board. I am happy that it turned out to something good. In the beginning most players were just bashing the AI without any constructive opinion. Day by day we had better and better suggestions/ideas, and now we are here...near the end of the road.

I guess it is up to the devs to do what they can.
I am sure that they will try their best. http://forum.shrapnelgames.com/images/icons/icon6.gif

[ November 07, 2003, 15:01: Message edited by: Mortifer ]

ywl

November 7th, 2003, 05:35 PM

Originally posted by NTJedi:
Example=
Destroyer(Pillages lands which have been owned by opponents for long periods of time)
Architect(focuses on improving lands)
Arch-Mage(focuses on magic)
Barbarian(mainly uses brute force)
Insanity(unpredictable actions= randomly flips between all personalities every 10 turns)
Cutthroat(one out of three armies will be sneaking during movement)

I don't know the code but such rough Category of personalities doesn't seem to be too impossible.
1) Destroyer: give the AI an option to pillage a province unless it's around the capital. Adjust the probability of how often this happened and you'd get it. This particular AI is not likely to win unless he is using dead Ermor or Carrion Wood of Pangaea. But he could be around to be a spoiler for everybody.

2) Architect: hmm... there isn't much to improve on the land other than building castles.

3) Mage or Barbarian: if there were a matrix of the AI's preference over different units, the difference would be a simple tweaking of numbers. Though such a matrix itself may not be easy to implement, because of the large number of possible units in the game.

4) Cutthroat: the AI doesn't use stealth troops particularly well. It might be hard to implement.

Gandalf Parker

November 7th, 2003, 05:47 PM

Originally posted by ywl:
4) Cutthroat: the AI doesn't use stealth troops particularly well. It might be hard to implement. [/QB]Its beginning to look as though the AI is too linear in its programming to make seperate personalities likely. Its probably one line line of instructions with some IF comments allowing for the different nations. Splitting all of that out into sperate AI files would be on the level of being a good reason for Dominions 3. The devs seem to be back to asking for specific fixes. When should an AI build a castle. When should they build a temple. When should they buy hvy troops.

However..... the seperate AIs we are talking about might be created by themes? Does a computer player use themes? I have to fall back on Dom1 experience but if Pangaea is told to build hvy units then they will lose their sneak ability with that army. Building an IF PANGAEA into the AI code might help but there is a line of play for Pangaea which does use those units.

What if a theme removed those hvy units? And maybe added an assassin? If the AI used that theme then would you have a computer player that would use those things halfway effectively?

-Storm-

November 7th, 2003, 10:04 PM

I think some games separated AI personalities, but I dunno what games. Civ3? Frankly I don't remember. http://forum.shrapnelgames.com/images/smilies/rolleyes.gif

Gandalf Parker

November 7th, 2003, 10:52 PM

Originally posted by -Storm-:
I think some games separated AI personalities, but I dunno what games. Civ3? Frankly I don't remember. http://forum.shrapnelgames.com/images/smilies/rolleyes.gif Of course it can be done. And its generally agreed that it would have some benefits. But if the AI was straight coded instead of seperate for each AI then breaking it out would be a major product upgrade. Since solo-play isnt a major part of the game it would be an unlikely project.

That means that lesser fixes should be discussed so that something can be done.

-- The word "impossible" is a term used by technicians which translates as "I COULD do it but it would take more time and effort than its worth."

[ November 07, 2003, 21:55: Message edited by: Gandalf Parker ]

licker

November 7th, 2003, 11:16 PM

One small quibble http://forum.shrapnelgames.com/images/icons/icon7.gif

Why do you assert that 'solo play isn't a major part of the game'?

The devs have said that they made Dom primarilly to be an MP game, but the number of people playing the game are likely to play more SP than MP, whatever it is that they would prefer to play.

So I would assert that 'solo play *is* a major part of the game' just that the devs may not desire to make upgrades to it as much as they desire to make upgrades to MP. Though outside of balance and bugs, there arn't that many upgrades to MP that are seemingly *that* important http://forum.shrapnelgames.com/images/icons/icon12.gif

November 8th, 2003, 12:54 AM

Originally posted by licker:
One small quibble http://forum.shrapnelgames.com/images/icons/icon7.gif

Why do you assert that 'solo play isn't a major part of the game'?

The devs have said that they made Dom primarilly to be an MP game, but the number of people playing the game are likely to play more SP than MP, whatever it is that they would prefer to play.

So I would assert that 'solo play *is* a major part of the game' just that the devs may not desire to make upgrades to it as much as they desire to make upgrades to MP. Though outside of balance and bugs, there arn't that many upgrades to MP that are seemingly *that* important http://forum.shrapnelgames.com/images/icons/icon12.gif Once again I agree, but it would be nice if we could stop this kind of arguments altogether. Either make the game completely MP, or completely SP, or there is no basis to say that one is more important than the other, and that one should receive less attention and suffer on account of the other.

Mortifer

November 8th, 2003, 08:56 AM

Yes SP is a major part of Doms II., that is for sure now.
Still I guess the devs won't be bothered by such coding like multiple AI personalities. I guess that would be countless hours to script and add.
MAYBE in the distant future, they will work on it, but maybe the AI can be upgraded a more simple way. [fix those things on the list.]
Well we shall see. http://forum.shrapnelgames.com/images/icons/icon12.gif

Gandalf Parker

November 8th, 2003, 12:56 PM

Actually I am 99% a solo player. But the game was built almost entirely around PBEM with some things tossed in for solo play. I dont forsee a major time-consuming rewrite for SP moving to the top of the list over many many MP items. Small tweaks, yes. But once it looked like it would not be simple I stopped considering it likely.

However, there are always major points to be made in that scoring process for personal interest by a dev.

[ November 08, 2003, 13:01: Message edited by: Gandalf Parker ]

DominionsFan

November 8th, 2003, 04:50 PM

Hey! The AI is NOT THAT BAD Gandalf, it need lot of tweaking but that's all. I think the devs can update those things from the list, without too many problems. http://forum.shrapnelgames.com/images/icons/icon6.gif

Gandalf Parker

November 8th, 2003, 05:46 PM

Originally posted by DominionsFAN:
Hey! The AI is NOT THAT BAD Gandalf, it need lot of tweaking but that's all. I think the devs can update those things from the list, without too many problems. http://forum.shrapnelgames.com/images/icons/icon6.gif The tweakings is what I mean. That IS what we should pursue. Its the projects such as seperate AI scripting or tactics specific to a nation that we might not see soon.

As for the tweakings I think we will need to get more specific. Just saying "more hvy units" by itself isnt a fix. The AI is no better with a few hvy units than it was with alot of cheap ones. When we put up some specific percentages that helped although that would hurt some races. When we gave specifics on when the AI should consider another castle that was closer.

[ November 08, 2003, 16:47: Message edited by: Gandalf Parker ]

Aristoteles

November 9th, 2003, 09:26 AM

Originally posted by Gandalf Parker:
<blockquote>quote:<hr />Originally posted by DominionsFAN:
Hey! The AI is NOT THAT BAD Gandalf, it need lot of tweaking but that's all. I think the devs can update those things from the list, without too many problems. http://forum.shrapnelgames.com/images/icons/icon6.gif The tweakings is what I mean. That IS what we should pursue. Its the projects such as seperate AI scripting or tactics specific to a nation that we might not see soon.

As for the tweakings I think we will need to get more specific. Just saying "more hvy units" by itself isnt a fix. The AI is no better with a few hvy units than it was with alot of cheap ones. When we put up some specific percentages that helped although that would hurt some races. When we gave specifics on when the AI should consider another castle that was closer. <hr /></blockquote>Well, than what would you suggest? Since the AI won't use too many heavy units in the demo, it is clear that we must make a suggestion about it. I think this is a 'quite simple' example, we cannot post too many details about this one...or can we?

Aristoteles

November 9th, 2003, 09:40 AM

A little update:

THE "AI WEAKNESSES" LIST:
1. The AI is massing weak troops
2. The AI won't protect his pretender/commanders/priests/mages
3. The AI won't build forts, maybe 1-2 max, but that is very rare as well.
4. The AI's battlefield tactics could be better
((Example: You can use a few units and put them to the front, the AI will attack them with full force more than likely. Usually the AI's own missile units will kill some or more of its melee units.))
5. The AI is making very odd things. IE. wandering around with his pretender alone from province to province ((the pretender is surely not doing anything, since it is moving in all turns. - and it is doing this in the border provinces! - the AI pretenders are all dead before turn 20-40 ususally.))
6. The AI won't stay in a province, it is always moving his armies.
7. The AI won't make complex strategic maneuvers. IE. Making a focused attack against a province.
8. The AI won't equip his supercombatants..at least I've never seen that in the demo.
9. The AI is building too many temples ((??????? This is good or bad, I cannot decide.))
10. The AI won't search for magical sites? ((I am not exactly sure, but I have never noticed, that the AI searched for magical sites.))

[ November 09, 2003, 08:44: Message edited by: Aristoteles ]

Nerfix

November 9th, 2003, 02:04 PM

One thing i noticed today:
The AI doesn't play Rainbow mage type pretenders very well:
I saw a Frost Father with 7 in Astral, 3 in Earth and 1 in Water and a Arch Mage with 6 in ir, 3 in Fire and 1 in Earth.
The thing is, those pretenders were very expensive in design points. The AI had poor scales and Mausoleums as forts.

Gandalf Parker

November 9th, 2003, 02:05 PM

Originally posted by Aristoteles:
Well, than what would you suggest? Since the AI won't use too many heavy units in the demo, it is clear that we must make a suggestion about it. I think this is a 'quite simple' example, we cannot post too many details about this one...or can we? WHAT AI doesnt use enough hvy troops? WHAT hvy troops do you wish he would use? What percentage? Maybe we could use the mini map to figure out if that is even a good idea and what nations its a good idea for. Definetly not all of them. Some nations would definetly go from decent to sucky if they used more hvy troops (Pangaea being a big example). And most nations would regret it as a general rule if it was too expensive a move. Is a few hvy troops better than alot of cheap ones in most cases?

The Posts talking about setting hvy troops to 10% of the army made some sense. But Im taking for granted they meant hvy infantry to be mixed in with the lite infantry. Does it also mean that they should have 10% hvy cavalry? Is that 10% of the whole army or 10% of a mounted army? Should they have a walking army where 10% of it is hvy troops AND a mounted army where 10% is hvy troops (with... what was it? 60% lite mounted?)

Gandalf Parker

November 9th, 2003, 03:29 PM

Originally posted by Nerfix:
One thing i noticed today:
The AI doesn't play Rainbow mage type pretenders very well:
I saw a Frost Father with 7 in Astral, 3 in Earth and 1 in Water and a Arch Mage with 6 in ir, 3 in Fire and 1 in Earth.
The thing is, those pretenders were very expensive in design points. The AI had poor scales and Mausoleums as forts. What meant they didnt do it well? It wasnt a good choice for a demo game since Rainbow Mages are a strategy that bears fruit late in the game. How do you feel the AI should have done it. Oh and what race had this pretender since that can make a big difference.

Nerfix

November 9th, 2003, 03:39 PM

Originally posted by Gandalf Parker:
What meant they didnt do it well? It wasnt a good choice for a demo game since Rainbow Mages are a strategy that bears fruit late in the game. How do you feel the AI should have done it. Oh and what race had this pretender since that can make a big difference. They spent huge amounts of design points on those pretenders. They had poor forts and negative scales.

Uhhhh, sir, we are talking bout' RB Mages with 7 in one path... Getting to 4 in one is hard and expensive enough with a RB Mage pretender
I usualy take 1-3 of every path with my RB pretenders. It may not be the best choice, but it is less expensive that taking 7 in one path with pretender like Crone.

Pythium had the Frost Father and Vanheim had the Arch Mage.

[ November 09, 2003, 14:42: Message edited by: Nerfix ]

Taqwus

November 9th, 2003, 06:43 PM

The AI should pay more attention to supply lines and the risk of strategic encirclement.

If it's going to risk being cut off, it really should make an attempt to forge supply items and bring them along (which requires nature magic; perhaps AIs should make more use of indy druids et al). This is particularly true when using troops with a low base morale, because of the 4 point penalty...

e.g. Last night playing the demo solo as Jotunheim (Utgard theme), I found and attacked the Machakans. At one point, they attacked a forward fort of mine. Their sieging force, mostly Machakan light foot with spider knights, was adjacent to only one Machakan-held province, which in turn was adjacent to their only castle (Machaka itself). In this single-province bridge lay another Machakan force led by the spider lady herself. Next to this bridge lay my main force.

The next turn, my main army (led by my Son of Niefel, fairly small but composed mostly of giants and with a number of Skratti and Godes et al) attacked -- not the besiegers, but the bridge. The Machakan force on that bridge, before the attack hit, joined the besiegers. That meant that the besiegers, while they were now making decent progress against the walls, were completely surrounded -- and I annihilated them the following turn, as the morale-4 troops were trivial to rout. Well in excess of half of the total Machakan armed forces, including a decent national hero, were obliterated at little cost since they had nowhere to run... That made it far easier to take Machaka a couple of turns later.

November 9th, 2003, 08:25 PM

Originally posted by Taqwus:
The next turn, my main army (led by my Son of Niefel, fairly small but composed mostly of giants and with a number of Skratti and Godes et al) attacked -- not the besiegers, but the bridge. The Machakan force on that bridge, before the attack hit, joined the besiegers. I would say that this is another example of the AI constantly pacing back and forth with his armies, and not a deliberate move. If you wanted to fight that army, you should have just waited and attacked the next turn - they would have walked back from the besieged province. It's a very predictable behavioral pattern for some reason.

Mortifer

November 9th, 2003, 08:43 PM

Originally posted by HJ:
<blockquote>quote:<hr />Originally posted by Taqwus:
The next turn, my main army (led by my Son of Niefel, fairly small but composed mostly of giants and with a number of Skratti and Godes et al) attacked -- not the besiegers, but the bridge. The Machakan force on that bridge, before the attack hit, joined the besiegers. I would say that this is another example of the AI constantly pacing back and forth with his armies, and not a deliberate move. If you wanted to fight that army, you should have just waited and attacked the next turn - they would have walked back from the besieged province. It's a very predictable behavioral pattern for some reason. <hr /></blockquote>Yup one of the many problematic points of the AI.
This is already a well known AI weakness btw...

[ November 09, 2003, 20:33: Message edited by: Mortifer ]

NTJedi

November 9th, 2003, 09:49 PM

Originally posted by Mortifer:
<blockquote>quote:<hr />Originally posted by HJ:
<blockquote>quote:<hr />Originally posted by Taqwus:
The next turn, my main army (led by my Son of Niefel, fairly small but composed mostly of giants and with a number of Skratti and Godes et al) attacked -- not the besiegers, but the bridge. The Machakan force on that bridge, before the attack hit, joined the besiegers. I would say that this is another example of the AI constantly pacing back and forth with his armies, and not a deliberate move. If you wanted to fight that army, you should have just waited and attacked the next turn - they would have walked back from the besieged province. It's a very predictable behavioral pattern for some reason. <hr /></blockquote>Yup one of the many problematic points of the AI.
This is already a well known AI weakness btw... <hr /></blockquote>This is why I suggested multiple personalities for the AI opponents because then we would not see this problem. Even as the developers improve the AI for this game as time passes more and more weaknesses/flaws will be discovered thus only leaving a challenge for human multiplayer games. If the AI opponents had multiple AI personalities it would keep gamers guessing... especially if the personalities would randomly change every 40 turns.
Hopefully Dominions_III they will be able to provide this or a miracle patch. I've suggested the same to the developers of Age_of_Wonders series and most of the gamers there also seem to agree.

Chris Byler

November 9th, 2003, 10:25 PM

Originally posted by Taqwus:
The AI should pay more attention to supply lines and the risk of strategic encirclement.

If it's going to risk being cut off, it really should make an attempt to forge supply items and bring them along (which requires nature magic; perhaps AIs should make more use of indy druids et al). This is particularly true when using troops with a low base morale, because of the 4 point penalty...

e.g. Last night playing the demo solo as Jotunheim (Utgard theme), I found and attacked the Machakans. At one point, they attacked a forward fort of mine. Their sieging force, mostly Machakan light foot with spider knights, was adjacent to only one Machakan-held province, which in turn was adjacent to their only castle (Machaka itself). In this single-province bridge lay another Machakan force led by the spider lady herself. Next to this bridge lay my main force.

The next turn, my main army (led by my Son of Niefel, fairly small but composed mostly of giants and with a number of Skratti and Godes et al) attacked -- not the besiegers, but the bridge. The Machakan force on that bridge, before the attack hit, joined the besiegers. That meant that the besiegers, while they were now making decent progress against the walls, were completely surrounded -- and I annihilated them the following turn, as the morale-4 troops were trivial to rout. Well in excess of half of the total Machakan armed forces, including a decent national hero, were obliterated at little cost since they had nowhere to run... That made it far easier to take Machaka a couple of turns later. Also, if it is caught in that situation, it should attempt to cut back to friendly lines. Usually in such a situation there are several directions the beseigers could have gone where they would have encountered little or no resistance and could have reestablished their supply lines within a few turns (although of course it would have meant abandoning the siege, this is surely far preferable to having your beseiging army wiped out).

An inferior force (and while they may have had decent force, they were definitely inferior while starving) should almost always try to avoid battle. (The exception is when they are in a desperate enough situation that avoiding battle will do no good - for example, if the entire army was already diseased, then making it back to friendly lines would be of little value.)

MStavros

November 10th, 2003, 12:03 PM

Originally posted by NTJedi:
<blockquote>quote:<hr />Originally posted by Mortifer:
<blockquote>quote:<hr />Originally posted by HJ:
<blockquote>quote:<hr />Originally posted by Taqwus:
The next turn, my main army (led by my Son of Niefel, fairly small but composed mostly of giants and with a number of Skratti and Godes et al) attacked -- not the besiegers, but the bridge. The Machakan force on that bridge, before the attack hit, joined the besiegers. I would say that this is another example of the AI constantly pacing back and forth with his armies, and not a deliberate move. If you wanted to fight that army, you should have just waited and attacked the next turn - they would have walked back from the besieged province. It's a very predictable behavioral pattern for some reason. <hr /></blockquote>Yup one of the many problematic points of the AI.
This is already a well known AI weakness btw... <hr /></blockquote>This is why I suggested multiple personalities for the AI opponents because then we would not see this problem. Even as the developers improve the AI for this game as time passes more and more weaknesses/flaws will be discovered thus only leaving a challenge for human multiplayer games. If the AI opponents had multiple AI personalities it would keep gamers guessing... especially if the personalities would randomly change every 40 turns.
Hopefully Dominions_III they will be able to provide this or a miracle patch. I've suggested the same to the developers of Age_of_Wonders series and most of the gamers there also seem to agree. <hr /></blockquote>Hehe, dominions 3.......if there will be Doms 3. at all.
The idea is good, but I guess, that the devs can update the AI very well, without adding major content -like the personalities.
I think that list made by Aristoteles is very valid, and all of those problems must be fixed.
Gandalf, your question was that what heavy units the AI should build.
I think that it is not our job to make suggestions about this one. The devs will know that what -and- how, since we do not know that AI scripts.

[ November 10, 2003, 11:04: Message edited by: MStavros ]

Gandalf Parker

November 10th, 2003, 03:19 PM

Gandalf, your question was that what heavy units the AI should build.
I think that it is not our job to make suggestions about this one. The devs will know that what -and- how, since we do not know that AI scripts.Yes it is. The Devs have asked for specific examples. It may give them ideas of where the AI should be modified. Or it may give us some more awareness on the problems here. People say that the devs are ignoring these threads but this is a 15 page thread that was answered numerous times starting with the second post. We seem to be ignoring their answers.

We could just start with one nation. When people say that the AI isnt using enough hvy units, what AI are they referring to? What units? If we can come up with a formula then we can compare that formula to other nations to see how it would go.

Illwinter listening to players debate formulas is what made Dom1 what it was.

DominionsFan

November 10th, 2003, 10:08 PM

Hey, I think that the AI should mix the heavy units, IE. it should use all of them. I think that all of the heavy troops are very good.
I am always using lot of different heavy troops and it is working very well. http://forum.shrapnelgames.com/images/icons/icon6.gif

NTJedi

November 11th, 2003, 07:00 AM

The developers want specific examples...

I believe the developers should have the AI create armies which they feel would be effective during gameplay. Hopefully they've played the game enough to know what combinations make a powerful attack force. http://forum.shrapnelgames.com/images/icons/icon10.gif
The current build of having massive amounts of small troops should be left for the 'easy AI' setting.
-On the same note: If NASA asked how to build a good space shuttle... I would look them in the eye and say 'Your the Experts' .

___________________________________

The only suggestion I highly recommend is having the AI opponent able to build multiple/random types of army combinations. If only one default army design is built the human players will quickly learn then master it.

-Storm-

November 11th, 2003, 10:02 AM

I don't get it. We had a suggestion that the AI should build more Heavy troops. So than? It should build HI, HC, summon powerful monsters etc. etc.
Why is this so complicated??

Gandalf Parker

November 11th, 2003, 03:41 PM

Originally posted by -Storm-:
I don't get it. We had a suggestion that the AI should build more Heavy troops. So than? It should build HI, HC, summon powerful monsters etc. etc.
Why is this so complicated?? Its like saying that people should own more porsches. Specific formulas could be tested for whether it would be workable more situations than it would be stupid. This is getting nuts. Out of all of these "different" people saying this needs done SURELY one of them can come up with a few specifics.

Everything has its pros and cons. Usually the disadvantage of having hvy units is that you dont get very many of them compared to lighter units. Building too many, too soon, and definetly thinking that ALL the AIs should do it would not be a good idea. (and Im suspecting its one that the Devs have already come up against). But I agree that all lighter units is not a good idea. I dont think thats whats being done except in the early game when gold/resources are limited.

And the devs have ASKED for specifics. Id say that unless we try to work this further we can probably consider the matter dropped.

SO what AI? What units? What percentages? Lets test the Groups on the "battle simulator" (mini.map) and pin this down.

licker

November 11th, 2003, 04:11 PM

Heh, this is exactly why I hoped that the army compositions could be externalized, much easier (and quicker) to test that way http://forum.shrapnelgames.com/images/icons/icon7.gif

For my part I'm not giving suggestions as to what the best composition is because I don't know what it is, I'm too new to the game to provide meaningful commentary on this issue. I was hoping that the Gandalfs and other vets would provide the army compositions that they find succesful in their games, and use that as a starting point...

Anyway, the issue is more complicated than just 'whats the best army composition' as you have to be able to take into consideration the factors in game that control that. The income, resources, gems, available mages (and magic for the '?' mages) are all controling factors in what the 'best' army compostion will be. Also certain balance or bug issues may be negatively impacting the AIs army compostions, issues with building forts and labs come foremost to mind.

So it may be too eary to start tweeking the AI, the balance and bugs need to be resolved first. But it's not too early to provide specifics as to what an ideal army compostion should be (scale it for turn#, or gem income, or whatever if you can). Once other issues have been sorted out, we'll be that much further along to getting a more robust AI (if it's needed that is http://forum.shrapnelgames.com/images/icons/icon7.gif )

I'll try a specific if you like...

C'tis (non Tomb theme)
20% LI (for the javalins)
40% Falchioneers or Elite Warriors (or both)
20% Swamp Gaurds
10% Longdead + Skeletons
10% Other (poison slingers, Serpents, Dancers)

The other is somewhat based on effectiveness of blessing and availability of Undead to use with the slingers.

The % of SG should drop as more and better Undead become available. If gem income and mages allow the % of Falchioneers and EWs should drop to allow for more death/nature (ideally poison immune) summons. The % of Slingers can increase as more poison immune units are created.

Comanders...

One LK per army mininum. As the army fills out one Shaman and one Commander or Lord to help ferry units. Undead can be led by MMs or Sauromancers, or summoned (Banes or Mound kings early). Add more mages as they become available to summon more hoards of undead.

Well there you have a stab at some concrete numbers http://forum.shrapnelgames.com/images/icons/tongue.gif

Gandalf Parker

November 11th, 2003, 05:44 PM

Originally posted by licker:
I'll try a specific if you like...

C'tis (non Tomb theme)
20% LI (for the javalins)
40% Falchioneers or Elite Warriors (or both)
20% Swamp Gaurds
10% Longdead + Skeletons
10% Other (poison slingers, Serpents, Dancers)
OK that looks good. So this is what you feel would provice you a challenge to go up against?

Do you see this as an army you would meet at their castle? Or that it would show up coming at you? Since its all home units I take it that this is something for early in the game.

How large an army do you see this as being? This is an ongoing build order? When these max do you see them walking out and starting another one?

I take it for granted that this is supposed to be built in this configuration right from the start? Not cheap units first then fill in later with others (I think thats whats going on now)

So I brought up my Ctis game and bought 2 LI, 2 Falconeers and 2 Elite Guards, I didnt see swamp guards but I may have the wrong theme, and I bought 1 slinger. That worked. I ran low on resources long before money but that would allow me to buy some nice expensive mages to keep up on the need for the undead units.

It didnt look like very many troops though. Many cheap troops are usually what I would start with to get a cheap efficient patrol force. Maybe for a walking army. If I use the battle simulator map I could put in 20, 40, 10, 10 of all those troop types and test them in battle.

November 11th, 2003, 06:03 PM

Well you have to decide on whether or not you want the AI to expand early and at what attrition rate.

I would contend with Indy's being set on 7 (50+ - 80 member armies per province) that the AI would react differently. As the only real efficent way to deal with the masses of LI, with a few support units (HI and Cav) is to have a tougher attack base to keep your army healthy and not waste money.

I would hope that the AI would buy the cheapest most survivable unit earliest.

Example

Ulm (Non Iron Faith Theme)
For the first 10 turns

40% of Production of LI (Maul, no Shield 18 Resources)
50% of Production of HI (Morningstar work)
10% of Production of Cav

All Commanders Master Smiths.

If you have them place them in the right formations with the right starting positions they can effectively still field an army that can kill nearby indies to upgrade their gold output.

After a certain point is reached (Perhaps 50 LI) the focus could switch more to the HI and Cav.

60% HI
40 Cav

Maybe this isn't the best strategy around but if you could put a cap on certain weaker units that they produce or maybe a turn cap; it would allow the progress of the game to be gauged by the AI somewhat generally.

Gandalf Parker

November 11th, 2003, 06:27 PM

Originally posted by Zen:
Well you have to decide on whether or not you want the AI to expand early and at what attrition rate. Thats part of the problem. At different indep settings this can change the reqquired troops alot. And if you expand with alot of attrition then you fall to the first player that reaches you. If you are too careful then you fall to the players who expand rather than rush.

much good stuff snipped

Maybe this isn't the best strategy around but if you could put a cap on certain weaker units that they produce or maybe a turn cap; it would allow the progress of the game to be gauged by the AI somewhat generally. That all looked good. So a rush on getting light troops made, patrol, raise taxes, then start filling in with heavy, elite, cavalry?
A number like 50 LI sounded good.

Gandalf Parker

November 11th, 2003, 06:41 PM

OK here would be my own effort. Just in general terms (needing tweaks for specific nations)

Buy a non-special commander with high leader ability. Able to lead 50 or 75 troops? Then each turn buy one cheap non-commander type (scout, spy, assassin, priest) and as many of the cheapest units possible (slingers, light infantry). That is the patrolling army. Each round you also raise the tax level 10%?

When the patrolling army is maxxed then start a moving army. Heavy commander and a high troop leadership ability. Start building the 5 different unit Groups to assign to him. Section 1 should have a mix of all infantry and be 50% of the army. Section 2 should be shooters for 20% of the army (archers, crossbows, javelin, slingers) to be positioned directly behind section 1. Sections 3 and 4 are the flanks. Large units or cavalry making up 20%. Section 5 is longrange such as longbows or flyers.

I would say that filling up the LI, then the shooters, then section five, and Lastly the section 4 (hvy cavalry/elephants/hydra/chariots) would keep you from having to support the higher upkeep-cost and supply eating units for a long time before marching them out.

But how does that compare with what the AI does now? Should I scatter the purchases so that a low-independent small-map rush game doesnt take me too soon? If I do that then Im not playing very good for the longer games.

licker

November 11th, 2003, 06:46 PM

Originally posted by Gandalf Parker:
<blockquote>quote:<hr />Originally posted by licker:
I'll try a specific if you like...

C'tis (non Tomb theme)
20% LI (for the javalins)
40% Falchioneers or Elite Warriors (or both)
20% Swamp Gaurds
10% Longdead + Skeletons
10% Other (poison slingers, Serpents, Dancers)
OK that looks good. So this is what you feel would provice you a challenge to go up against?

Do you see this as an army you would meet at their castle? Or that it would show up coming at you? Since its all home units I take it that this is something for early in the game.

How large an army do you see this as being? This is an ongoing build order? When these max do you see them walking out and starting another one?

I take it for granted that this is supposed to be built in this configuration right from the start? Not cheap units first then fill in later with others (I think thats whats going on now)

So I brought up my Ctis game and bought 2 LI, 2 Falconeers and 2 Elite Guards, I didnt see swamp guards but I may have the wrong theme, and I bought 1 slinger. That worked. I ran low on resources long before money but that would allow me to buy some nice expensive mages to keep up on the need for the undead units.

It didnt look like very many troops though. Many cheap troops are usually what I would start with to get a cheap efficient patrol force. Maybe for a walking army. If I use the battle simulator map I could put in 20, 40, 10, 10 of all those troop types and test them in battle. <hr /></blockquote>Whether it would be challenging or not depends on many different things... In general I'd say it would be more challenging than an army composed of 80% LI and Militias though...

I wasn't real specific with the when or the wheres of that compesition. Ideally that would be an army that would be used later in the game with 80+ units I think. Initially you'd have to go with more like 70% LI and 30% 'other' probably the EWs and SGs. I didn't add any kind of logic to tweek the %s for the outside factors though, but ideally they %s would be tweeked depending on income/resource levels, gem income, mages...

I don't see patroling (to counter unrest from raised taxes) to be as useful in Dom2 so I'm not sure that I'd focus on patrols and high income early, over expansion to simply get more province, but that's a bit of a different question. I don't konw if the AI is built to up taxes and patrol or not, and I'd rather not complicate the issues at hand with that.

It would be interesting to run some tests with that basic army compostion and say a couple Sauromancers (or Marshmasters) with evocation 3 or 4 and Alteration 2. Use an LK and one other Infantry type comander. What you pit that army against... I dunno, maybe some combination of similarly gold/resource Ulm army (since that's the other nation that's been discussed).

I have to say though that as the game progresses those %s need to change, and probably heavilly. Especially if there is a good nature/death gem income to create armies with more summons. If that's the case then the poison slingers need to go up along with the poison immune units, while some amount of either blessable or heavy hitters (falchioneers and EWs) are kept around.

Gandalf Parker

November 11th, 2003, 08:43 PM

I have to say though that as the game progresses those %s need to change, and probably heavilly. Especially if there is a good nature/death gem income to create armies with more summons. If that's the case then the poison slingers need to go up along with the poison immune units, while some amount of either blessable or heavy hitters (falchioneers and EWs) are kept around. I think it would be best if the formulas handled the game progression. Either by using the number of turns as a factor (LI should be 50 - turns) or by having them in the %'s with a clause to ignore it if its not possible.

Such as... If I said HI 20%, undead 20%, Hvy Cavalry 10%, non-undead summons 20%, LI 50% (in that order) then of course thats more than 100%. But early in the game the undead and summons cant be done which would cause LI to be built. If only undead at %5 can be done then the rest would fill in. But even late in the game when it can make ALOT of undead it still wont end up with a completely undead army. My math sucks so Im hoping the concept is clear here. The %'s dont have to add up to 100% if the sequence is considered.

The same could be done for making magic items or casting global spells (probably before unit building). Telling the AI to create something wont matter if it cant afford to create it. As soon as it can afford it, it gets made.

I still think this might be a job for a real-life budget manager to tackle http://forum.shrapnelgames.com/images/icons/icon7.gif

[ November 11, 2003, 19:44: Message edited by: Gandalf Parker ]

licker

November 11th, 2003, 09:32 PM

I like your suggestion Gandalf, seems more easy to implement than using a ton of if then else structure.

Hopefully once the full game hits we'll have ample time to see if the AI is actually doing things better after turn 40 than it appears they are before. Also there was mention of a patch, so maybe some of the formulae were already tweeked to improve the AI...

Only time (or a dev) will tell http://forum.shrapnelgames.com/images/icons/icon7.gif

Mortifer

November 12th, 2003, 09:07 AM

Originally posted by licker:
I like your suggestion Gandalf, seems more easy to implement than using a ton of if then else structure.

Hopefully once the full game hits we'll have ample time to see if the AI is actually doing things better after turn 40 than it appears they are before. Also there was mention of a patch, so maybe some of the formulae were already tweeked to improve the AI...

Only time (or a dev) will tell http://forum.shrapnelgames.com/images/icons/icon7.gif Yes, but if the early game of the AI is so weak.....anyways wait for the full and we will be wiser. http://forum.shrapnelgames.com/images/icons/icon10.gif

As for this suggestion by Gandalf. Do you think, that it is a good idea to post deatiles like.. XY would like to see 10% LI, 30% HI etc. in the early game, this and that in the late game....I think this won't work.
Someone have said that the devs will know that what and how. I agree with this. They've scripted the AI, they know the major AI weaknesses now. We've made our suggestions already. Just my 2 cents.

[ November 12, 2003, 08:09: Message edited by: Mortifer ]

Gandalf Parker

November 12th, 2003, 02:32 PM

Originally posted by Mortifer:
Someone have said that the devs will know that what and how. I agree with this. They've scripted the AI, they know the major AI weaknesses now. We've made our suggestions already. Just my 2 cents. OK we can back off on the discussions if everyone REALLY feels its left in the Devs hands. But the devs havent said they agree with all of those suggestions.

I thought the people who kept raising the list to the top wanted to pursue the subject along the lines that the devs had mentioned in their response. If the thread keeps jumping up then I will jump back in to discuss things further.

NTJedi

November 12th, 2003, 03:57 PM

Originally posted by Gandalf Parker:
If the thread keeps jumping up then I will jump back in to discuss things further. 50% of the recent Posts here are from you jumping back in keeping the thread at the top. I think the devs got what they needed... did you? http://forum.shrapnelgames.com/images/icons/icon10.gif

Mortifer

November 12th, 2003, 04:17 PM

Originally posted by NTJedi:
<blockquote>quote:<hr />Originally posted by Gandalf Parker:
If the thread keeps jumping up then I will jump back in to discuss things further. 50% of the recent Posts here are from you jumping back in keeping the thread at the top. I think the devs got what they needed... did you? http://forum.shrapnelgames.com/images/icons/icon10.gif <hr /></blockquote>Well, IF the devs want us to make detailed examples like what should the AI build, than I will post MY ideas. But those will be my ideas, prolly we all have different opinion about this topic. That is why I've said that details like this are not vital for the devs...IMHO.
I guess they will know how to fix those things from the list. BUT if they really need such details, I will be glad to help in that. http://forum.shrapnelgames.com/images/icons/icon7.gif

[ November 12, 2003, 15:18: Message edited by: Mortifer ]

licker

November 12th, 2003, 04:17 PM

I don't know if the devs got what they needed or not. Anyway, its silly to just stop the discussion, though you are free to stop reading it if it bothers you for some reason.

What is unknown at this point is how the bug and balance changes will influance the AI's army composition (since that's mainly what we are dealing with here). There's no reason to stop debating what a good army mix should be, even as we hope that the simple fixes (coming in the patch I assume) may have fixed some of these issues.

Besides Gandalf is playing an important role in these discussions, he's challenging everyone who steps in and says 'this isn't good' to give some details on how they would make it better. The more specifics you give the better off the discussion will be. If you arn't interested (or cannot supply) the specifics then don't bother complaining when someone else tries to provide some.

Anyway, we all seem to have faith in the devs, but even they benefit from reading these discussions, as I said, if you don't benefit from it, stop reading it.

Mortifer

November 12th, 2003, 04:20 PM

Originally posted by licker:
I don't know if the devs got what they needed or not. Anyway, its silly to just stop the discussion, though you are free to stop reading it if it bothers you for some reason.
You are right licker...but I hope that you understand my point about this. http://forum.shrapnelgames.com/images/icons/icon7.gif
{{This is a very complicated thing, thus we all have different ideas. Now if we will post 20 different things, that surely won't help.}}
Hm however it won't cause any probs at all, so yes why not.. http://forum.shrapnelgames.com/images/icons/icon6.gif

licker

November 12th, 2003, 04:40 PM

I wasn't actually replying to your post Mortifer http://forum.shrapnelgames.com/images/icons/icon7.gif

It just got in the way, that's all http://forum.shrapnelgames.com/images/icons/icon12.gif

I think all of this discussion is worthwhile and interesting. If the devs agree, great, if they don't... it's still not hurting anything, and some of us can read between the lines and improve our own games from this discussion http://forum.shrapnelgames.com/images/icons/icon7.gif

LordArioch

November 12th, 2003, 05:22 PM

Gandalf Parker

November 12th, 2003, 05:31 PM

Originally posted by Kristoffer O:
The AI is not cheating on normal level. At higher levels it gets more design points and earns more money. Thats about it.

I got an idea regarding AI dependancy on LI when I was away. Vacations are 'foyson' for the mind. http://forum.shrapnelgames.com/images/icons/icon7.gif

Regarding fort construction. I'm not sure how the AI works, but where would you build a fortress? Consider fort type, nation, geography, income, resources etc.

What is the most important matter? How should they be weighted? How much shall current wars affect the spending of time and money. What army should build the fortress?

I wouldn't mind a numerical evaluation of this such as:
Castle cost / 2 < Income + res + gem income x 25 + neighbors x 5
Add a couple of other conditions.

Just to make you think. http://forum.shrapnelgames.com/images/icons/icon7.gif Eventually it might result in something good. What I saw when I dove in here strongly was a series of Posts putting the list back up to the top of the thread. My efforts were trying to get things back into the track requested by the devs in responses in this thread.

We can tackle any of the things that are on the wish list of this thread but the goal should be to give testable examples and suggested formulas. Im not one to jump on people for saying "its broke" and not helping out. But saying its broke over and over doesnt do much.

Shall we continue work discussing army formulas or shift to something else on the list?

Gandalf Parker

November 12th, 2003, 05:41 PM

I'd advise testing the AI on different difficulty levels. The easiest game I had against the AI was against all impossibles, about default settings. I personally defeated 2 ai players by turn 20, which is unheard of by me on even normal difficulty. Admittedly jotunheim with a blade wind casting pretender didn't hurt...but even so. [/QUOTE]

That is part of the problem. Telling the AI to build armies a certain way only works best for certain settings. The AI now seems to do best when the game is set at independents 5 or 6. I usually set it at 7. The AI then seems to properly seek expansion with mass troops, then builds decent armies when it can afford to, then work on magic.

Indeps set at 3 (the default) tends to create a "rush game" but the AI set on difficult makes it think its in for a long hard game. The harder AIs definetly perform better in harder games.

http://www.otakurevolution.net/otakurevolution/idea.gif
I havent tested it but Im rather curious if the easy AI might not actually do better in a low indep game than the hard AI does.

November 12th, 2003, 08:31 PM

Does anyone know if the AI is one broad spectrum or if it's individualized?

Just for my benefit for thinking about it:

Maybe I haven't been able to find it on the Boards or on any website, but if the AI is simple in it's parameters (Simple as in it meets a certain condition and performs) and not geared towards a race then we can come up with formulas based on generalizations.

To better explain. Does the AI have a parameter: (to use my own brand of dumbass programming language)

IF Jotun, Turn 1
IF Have Castle, Hill Fort
DO Produce Jotun Woodcutter 100%

or is it that certain things are flagged as a certain Category like say all Cavalry are flagged in every race or province and are created when a certain command is given no matter the race.

So when we say "The AI should produce 50% HI unless they are X, X, X, race where they should focus 20% more on Ranged/HI" are we making comments for only one race or do you want broad generalizations. As in the Castle and placement debate.

November 12th, 2003, 09:13 PM

To try to contribute a little to the discussion.
I was thinking about the AI pacing back and forth in the border provinces. I don't know how the AI handles that situation now, and whether it's actually reacting to the border and neighbouring armies by amassing troops but then running out of supply in that particular province and hence moving the troops out. But to make it a little less predictable, maybe something like this would help:
IF (enemy army present in neighbouring province)
THEN (stay in the province for 1+x turns if not starving, where x is a random dice number)

It would seem that the AI knows when it's threatened, judging by the amount of local defense it puts into threatened provinces, and it should know whther it's out of supply as well, so maybe this would work if it's not already working this way, and the only problem is the supply route, which brings us back to the castle issue.

licker

November 12th, 2003, 09:39 PM

Originally posted by HJ:
To try to contribute a little to the discussion.
I was thinking about the AI pacing back and forth in the border provinces. I don't know how the AI handles that situation now, and whether it's actually reacting to the border and neighbouring armies by amassing troops but then running out of supply in that particular province and hence moving the troops out. But to make it a little less predictable, maybe something like this would help:
IF (enemy army present in neighbouring province)
THEN (stay in the province for 1+x turns if not starving, where x is a random dice number)

It would seem that the AI knows when it's threatened, judging by the amount of local defense it puts into threatened provinces, and it should know whther it's out of supply as well, so maybe this would work if it's not already working this way, and the only problem is the supply route, which brings us back to the castle issue. That seems like a reasonable idea. I'm wondering if the AI is getting caught in some kind of a yo-yo effect where it keeps seeing different threats and keeps moving to react to them, only to see them disappear in the same way that it's forces are disappearing... http://forum.shrapnelgames.com/images/icons/icon7.gif

The AI should be able to keep a properly suplied army in a province as well, I don't see why it's moving in too many units anyway, unless it understands the abusability of the supply rules... which is something else that could be fixed... but I digress http://forum.shrapnelgames.com/images/icons/icon7.gif