Going from Normal AI to Impossible AI...

[Atreidi]

Im just curious what difficulty level most of you guys play and feel confortable with.

I usually play against a combination of difficulties but mostly Difficult.

1. What is the difference between difficulties?
2. What does the AI do differently as the difficulty increases?
3. Does the AI cheat in higher difficulties or is it simply more efficient?

[Atreidi]

I've been looking all over the manual for an explanation on the difficulty levels and cant find anything. I dont know if im not looking in the right place or if its just not there http://forum.shrapnelgames.com/images/smilies/frown.gif

[Loren]

The AI gets more points as the difficulty goes up.

At impossible I've seen pretenders with 4 paths at level 9.

[MaxWilson]

If by "cheat" you mean extra resources, yes, it cheats. I believe the AI at higher difficulty levels gets 200% of the gold, resources, and (I think) gems it actually "earns." I don't think it "cheats" in the sense of having extra information about your status of forces, except that it knows who owns every province whether it has a scout nearby or not. That's really the only difference between difficulty levels except for the AI getting more points to design Pretenders. I think Difficult = 130%, Mighty = 170%, and Impossible = 200%. I'm not sure, though. KO or JK posted about it once in the forums here but my memory of the details is fuzzy.

I usually play against Impossible AIs using Edi's Better Independents mod. I have found that the AI is actually more challenging with a mix of difficulty ratings (probably because the Impossibles eat the Mighties and Difficults) but I usually don't set games up that way for some reason. I also lose fairly regularly. If I really want to do some role-playing instead of feeling constrained to play "efficiently" I will play against Normal or Difficult AIs.

-Max

[MaxWilson]

Loren said:
The AI gets more points as the difficulty goes up.

At impossible I've seen pretenders with 4 paths at level 9.



You can actually do this if you take the right Imprisoned pretender and poor enough scales. EA Mictlan can get a F9W9D9B9 Smoking Mirror with Dominion 4.

Not that the AI was doing that, it just had more points. The frustrating thing for me is that it still frequently takes Death scales and I'm too lazy to mapedit smart pretender designs in for it--so I just use a lot of opponents and let the good ones eat the bad ones.

-Max

[NTJedi]

Loren said:
The AI gets more points as the difficulty goes up.

At impossible I've seen pretenders with 4 paths at level 9.



With some programming effort the AI could be designed to take smarter scales on higher difficulty. Smarter scales... such as never taking Death_2 or Death_3 or never taking Unluck_2 or Unluck_3. Human players know how to manage these scales where as AI opponents really suffer. Other useful scales for the AI would be taking Order and Production.

[quantum_mechani]

NTJedi said:
Smarter scales... such as never taking Death_2 or Death_3 or never taking Unluck_2 or Unluck_3.

Misfortune is a staple of MP players, and there are plenty of times when a build has need of points from death scale as well.

[kasnavada]

I think you missed the second part of NTJedi's reply...

What I would actually like is to be able to set the AI before starting a game...

[Edi]

I'm too lazy right now, but somebody should dig up the references to just how much bonus the AI gets at what difficulties to make sure so I could add that to the FAQ.

I'm not quite sure, but I think the bonuses to gold, resources and gems were something like
Easy -30%
Normal 0%
Difficult 30%
Mighty 60%
Impossible 100%

[Herode]

Only +30% for "difficult" ?
Considering the intense troop spamming I have to face at this level, I would have guessed almost 100% http://forum.shrapnelgames.com/images/smilies/fear.gif

I often play at difficult level, until mid game. I never went past mid game since this adverse spamming is boring in SP. On the other hand at "normal" level, the AI is far too weak and easy to beat and it is #my# spamming which becomes boring http://forum.shrapnelgames.com/images/smilies/biggrin.gif

[Dedas]

It would be really cool if you could change the difficulty when setting human players to AI when inside a new game. That way you could design their pretender gods by yourself making them much harder to beat. You could also set up challenging scenarios for yourself. What say you Kristoffer and Co?

[Endoperez]

Doesn't SemiRandom do that? Set up better pretenders for AIs, I mean.

[Dedas]

Haven't even looked at it but will now, thanks Endo!

Would it be possible to use this on the Dawn of Dominions map?

[MaxWilson]

Edi said:
I'm too lazy right now, but somebody should dig up the references to just how much bonus the AI gets at what difficulties to make sure so I could add that to the FAQ.




Ask and ye shall receive, knock and it shall be opened (http://www.shrapnelcommunity.com/threads/showthreaded.php?Number=501765) unto you.

-Max

[Gandalf Parker]

Yes SemiRandom will work on the Dawn of Dominions map.

A couple of points. Firstly, SemiRandom doesnt do anything that you cannot do yourself. The Dominions 3 game comes with a doc file for map commands to create your AIs. You can use a text editor (like notepad or wordpad) on any .map file altho you probably want to copy it to a new name first. You can place AIs if you want to so that they start in powerful locations. You can set the scales. You can create the gods (even ones not on the god list such as giving Eater of Gods as a pretender). You can give the AI extra starting armies (flight of dragons, or even a squad of titan pretenders). You can give the AI specific magic sites and extra castles. You can make sure that the AI begins the game with extra provinces next to it which are high in resources, or gems, or gold. Or even that have specific independents that you wish the AI would use more such as cavemen or knights. And you can set the AIs to ally with other AIs against you (they dont share but you can set it so they do not attack each other).

Secondly its fun to point out that Ballbarian wants very much for people to send in those AI designs so that they can be included in the files that SemiRandom can choose from when scattering random surprises into a map file for people to play on. If you find the program fun then thank him for the couple dozen random things it tossed in that made you smile by sending him one he can add to SemiRandoms library.

Gandalf Parker
--
Me: "Some people play to win or lose, others just to stay in the game."
Them: "Are we talking about games? Or life in general?"
Me: "Exactly."

[theenemy]

Is it true that the AI gets better if you set research to "easy"? I've read some posts about it but never really experiensed a change myself.

[Edi]

MaxWilson said:

Edi said:
I'm too lazy right now, but somebody should dig up the references to just how much bonus the AI gets at what difficulties to make sure so I could add that to the FAQ.


Ask and ye shall receive, knock and it shall be opened (http://www.shrapnelcommunity.com/threads/showthreaded.php?Number=501765) unto you.


Thankee muchly. This has been added to the FAQ into the same section as magic research and aging mechanics. http://forum.shrapnelgames.com/images/smilies/happy.gif

[Gandalf Parker]

theenemy said:
Is it true that the AI gets better if you set research to "easy"? I've read some posts about it but never really experiensed a change myself.


I would think that might apply to certain nations more than others. Depending on their tendency to use various spells.

Come to think of it I guess it might also apply to nations which have a tendency to create items.

The problem is that its difficult to test such things because of all of the variables in any game you play with them. In the early days of beta-testing for Dom3 I would create 20 games full of AIs on my server, auto-run them thru 100 turns, save the score files then start 20 more. It eventually gave me some answers as to how well certain nations did against each other or in different game settings. I think it was helpful but I havent done any of those for a couple years now.

[quantum_mechani]

kasnavada said:
I think you missed the second part of NTJedi's reply...


Hmmm... I'm still not sure I agree with that. If the AI is going to take any negative scales at all, those don't seem at all the worst.

[NTJedi]

quantum_mechani said:

NTJedi said:
Smarter scales... such as never taking Death_2 or Death_3 or never taking Unluck_2 or Unluck_3.

Misfortune is a staple of MP players, and there are plenty of times when a build has need of points from death scale as well.



Yes... as mentioned you completely missed my second sentence.

http://forum.shrapnelgames.com/images/smilies/biggrin.gif

[NTJedi]

quantum_mechani said:

kasnavada said:
I think you missed the second part of NTJedi's reply...


Hmmm... I'm still not sure I agree with that. If the AI is going to take any negative scales at all, those don't seem at all the worst.



The unluck scale will cause more random attacks which can suffocate an AI. I've seen them sit on their hands while their capital goes under siege during early game from independents.

The death scale drains their gold income and supplies. The AI has a tough enough time not starving its armies... the death scale only makes it occur more and more often.

[quantum_mechani]

Supplies really are a problem... but just not taking death is not really enough to save them, AI games should be with maxed supplies anyway.

As far as the misfortune thing, it can skrew particular AI nations, but it doesn't seem all that common.

[NTJedi]

quantum_mechani said:
Supplies really are a problem... but just not taking death is not really enough to save them, AI games should be with maxed supplies anyway.

As far as the misfortune thing, it can skrew particular AI nations, but it doesn't seem all that common.



Yes maxed out supplies are necessary for SP games, unless you're playing against only LA ERMOR... maybe EA ARGATHA wouldn't suffer as much as other nations. Most AI nations needs a growth scale to survive late game.

Depending on the nation it might be able to stomach misfortune 3. Ideally all positive scales work best for an AI nation. The magic paths for the pretender would be useful if the AI knew how to have priests travel with sacred units and stand near them.

[Ming]

I think the map used and whether score graphs are enabled are just as important as the difficulty level. The AI cannot judge the relative strategic importance of corner and border provinces and need a wrap-around map to be more competitive. For reasons discussed elsewhere in this Forum, turning on score graphs makes the game more challenging in the more difficult settings as the AI mistakenly see you as weak and will gang up on you.

I find the "Desert Eye" map with score graphs enabled and 12 impossible AIs to be the most challenging setting in SP(without going into any programming - as I cannot even do the simplest). The only problem is that starting positions vary in desirability a lot and one could get rather lucky or unlucky with the Desert Eye map, as the case may be.

Would be interested if anyone know of other wrap-around maps that is challenging to play in SP but with more uniform starting positions.

[Renojustin]

Wouldn't it be relatively easy to upgrade the AI to the point where it could use sacreds, scales, and supplies, everything really, in a far better fashion? I don't think SP Dominions gets enough love currently.

[thejeff]

AI behavior actually depends on the score graphs being on?

I know it's likely to attack if it thinks you are weak, based largely on army size, not quality, but I had always thought it got that information independent of the graphs being on or off.

[Endoperez]

thejeff said:
AI behavior actually depends on the score graphs being on?

I know it's likely to attack if it thinks you are weak, based largely on army size, not quality, but I had always thought it got that information independent of the graphs being on or off.



Me too?



Renojustin said:
Wouldn't it be relatively easy to upgrade the AI to the point where it could use sacreds, scales, and supplies, everything really, in a far better fashion? I don't think SP Dominions gets enough love currently.



Can you do it? At least theoretically? Unless you can, it isn't as easy as you think it is.

[Ming]

Maybe I am wrong, but I certainly found that I got ganged up on by the AI more often when I have the score graph on - in my last game 4 AI's declared war on me in year one (fortunately one was not my neighbour) and by year three I am fighting 7. that is 7 out of 10 as two have been eliminated by other AI's ( I have been unable to eliminate any of my opponents for a variety of reasons although three have been reduced to 1-3 provinces). This has never happened in my dozen SP games with score graph off. I sometimes get by without being attacked by more than one AI in year one with score graph off but with score graph on I usually get attacked by 2 AI's before year 1 is over.

Maybe it is just small sample variance. Could someone give a definite answer to this?

[Ming]

Can you tell me where to download SemiRandom? Saw it first time in your post and would like to find out more.

[Endoperez]

Ming said:
Can you tell me where to download SemiRandom? Saw it first time in your post and would like to find out more.



http://www.shrapnelcommunity.com/threads/showflat.php?Cat=&Number=523661

[Ming]

Many thanks.

[Zeldor]

Impossible AI with better independents mod can make really good use of sacreds, but only if it has H3 priests [though Divine Bless should have really higher priority]. With all that points it gets and negative scales it take you can meet triple 9 blesses there.

[thejeff]

You can meet triple 9 blesses, but you'll also meet big armies of expensive sacreds that the AI doesn't bother to send a priest with. Or Cast bless if it does.

[MaxWilson]

I also did not think the AI needed the score graphs. I suppose I can try playing some more without the graphs and see if there's a perceived difference--although buying decent PD also keeps the AI off your back and I don't think that shows up in score graphs.


Endoperez said:

Renojustin said:
Wouldn't it be relatively easy to upgrade the AI to the point where it could use sacreds, scales, and supplies, everything really, in a far better fashion? I don't think SP Dominions gets enough love currently.



Can you do it? At least theoretically? Unless you can, it isn't as easy as you think it is.



Bear in mind that the "upgrade" has to be smart enough to work with modded units and nations, too. It can't be a bunch of special-casing like "If you're Mictlan, send a priest to bless F9W9 jags but if you're Marverni don't bother blessing your Boar Warriors." You'd need to either *always* bless blessable troops (I actually think that would be a pretty decent fix) or figure out a way to evaluate when it's worth it.

-Max

[Loren]

Renojustin said:
Wouldn't it be relatively easy to upgrade the AI to the point where it could use sacreds, scales, and supplies, everything really, in a far better fashion? I don't think SP Dominions gets enough love currently.



Yeah--I've taken multiple air-breathing items off the AI in a current game. There are no water races in the game!

[NTJedi]

Loren said:

Renojustin said:
Wouldn't it be relatively easy to upgrade the AI to the point where it could use sacreds, scales, and supplies, everything really, in a far better fashion? I don't think SP Dominions gets enough love currently.



Yeah--I've taken multiple air-breathing items off the AI in a current game. There are no water races in the game!



I would be happy to organize a complete shortlist of AI improvements, if KO and JK have an interest for improving the AI.

[Gandalf Parker]

Well I know that I have asked for such a list so that the modders and mappers could improve things. But it would tend to be a very different list.

For KO/JK the AI is linear in one section of code. It tends to operate in general rules without alot of "if" code. So it would be something like purchase infantry before cavalry instead of specifying IF Ulm then purchase specific unit before other specific unit. Or consider xxxxxx spell better than xxxxxx spell but not IF pythium or lines like not to cast xxxx spell on xxxxxx specific unit.

But for modding and mapping purposes the specific failures of specific nations can be addressed. Such as the fact that Mictlan tends to kill itself off early in solo games by not providing for its own dominion can be fixed by setting specific pretenders and specific scales.

[Loren]

Gandalf Parker said:
Well I know that I have asked for such a list so that the modders and mappers could improve things. But it would tend to be a very different list.

For KO/JK the AI is linear in one section of code. It tends to operate in general rules without alot of "if" code. So it would be something like purchase infantry before cavalry instead of specifying IF Ulm then purchase specific unit before other specific unit. Or consider xxxxxx spell better than xxxxxx spell but not IF pythium or lines like not to cast xxxx spell on xxxxxx specific unit.

But for modding and mapping purposes the specific failures of specific nations can be addressed. Such as the fact that Mictlan tends to kill itself off early in solo games by not providing for its own dominion can be fixed by setting specific pretenders and specific scales.



Assuming the code tolerates it I would think allowing modders to raise/lower the priority given various things should help with AI tweaking.

[NTJedi]

Gandalf Parker said:
Well I know that I have asked for such a list so that the modders and mappers could improve things. But it would tend to be a very different list.

For KO/JK the AI is linear in one section of code. It tends to operate in general rules without alot of "if" code. So it would be something like purchase infantry before cavalry instead of specifying IF Ulm then purchase specific unit before other specific unit. Or consider xxxxxx spell better than xxxxxx spell but not IF pythium or lines like not to cast xxxx spell on xxxxxx specific unit.

But for modding and mapping purposes the specific failures of specific nations can be addressed. Such as the fact that Mictlan tends to kill itself off early in solo games by not providing for its own dominion can be fixed by setting specific pretenders and specific scales.



Sounds like it would be easy to stop the AI from sending pretenders into the arena death match since they have a unique tag.

[HoneyBadger]

One really easy (I think...) stopgap solution to the bad AI sp problem would be if we were able to assign greater or less points to all players at the beginning of the game, and then ofcourse switch them to computer controlled-this would also be an interesting handicap feature for MP games.

[hunt11]

though this would only help bless nations, a mod that made all the computers holy units blessed all the time could help make the AI tougher

[Renojustin]

NT, I think an AI Improvement Shortlist is a fantastic idea.

[capnq]

HoneyBadger said: One really easy (I think...) stopgap solution to the bad AI sp problem would be if we were able to assign greater or less points to all players at the beginning of the game, and then ofcourse switch them to computer controlled-this would also be an interesting handicap feature for MP games.

What do you mean, "if"? The AI difficulty setting, which affects how many points it gets to design its pretender, can be set individually for each AI nation, and a human-built pretender switched to computer-controlled doesn't have to spend all of the assigned points.

The only complication in using this in multiplayer is enforcing the point limits.

[Atreidi]

Does the computer ever cast Master Enslave?

I've never seen them do it in all the SP games I've played. Maybe I need to set all the AI to Impossible...

[NTJedi]

Atreidi said:
Does the computer ever cast Master Enslave?

I've never seen them do it in all the SP games I've played. Maybe I need to set all the AI to Impossible...



Well the spells which require many magic paths are never casted by the AI because it doesn't place magic boosters correctly and it doesn't use communion master/slave. Also the powerful spells like master enslave require lots of research and the AI usually does poor at researching in comparison to the human player. One of the best computer researchers is Tien Chi.


Renojustin said:
NT, I think an AI Improvement Shortlist is a fantastic idea.



I agree, but I know KO and JK are currently working a new project, plus the current bug shortlist plus the new mod shortlist so I doubt a computer AI shortlist would recieve any attention.
I'll organize the shortlist if they first show interest.

[Atreidi]

waa, people playing in Easy Dificulty? Why? :S

[chrispedersen]

I find you get more reliably challenging games if you pick stronger nations, rather than accepting random nations with impossible difficulty.


For example, in glory of the gods, if you pick one water nation, and then the following land nations -
tien, lanka, van, niefle, mictlan - I believe you will get a pretty challenging game (so long as you are not playing a water race)

[Humakty]

I play difficult but the only independants I recruit are the mages I need to diversify magic, every commanders in my army are national ones. It makes the game really harder.

[Sombre]

I play difficult but the only independants I recruit are the mages I need to diversify magic, every commanders in my army are national ones. It makes the game really harder.

Sounds just like a NI map.

[Gandalf Parker]

There are many AI fixes available. Lowering resources in the game so only castles can recruit much, turning the independents strenth down, NI maps (no independents), creating super powerful AIs with selected gods and scales, Allying the AIs with each other, mods which add new powerful nations or change present nations to play AI abit smarter.

On the subject of "create powerful, start human, then go AI" it is true that the AI settings do give the AI more points than you get to create a god. But by turning off cheat-detection, and using map commands to create the god, you can exceed the player limits to make the best use of both worlds.

Gandalf Parker

[Humakty]

I never tried NI maps, but how does the AI uses its excess gold if it can't recruit indeps ?
I think putting some additional limits on what I do recruit is an easy way to make the game harder, without fighting the endless hordes you encounter on impossible difficulty (I'm always crushed on the big maps I love if I don't have luck on site searching(ie : finding the gems for your mage's main paths) + a good startpos).

The NI mod/maps seems good, but is it compatible with CBM ? I barely can play a game without it, too much reference switching.

[Edi]

The AI will buy staggering amounts of good quality stuff if the low quality chaff is unavailable. Try the Better Independents mod, which disables the crap infantry types by skyrocketing their gold costs. It leaves archers available and in addition to those, you'll start seeing a lot more nationals among the AI armies.

Nothing like seeing a scouting report that reads that there is an army of 750 units in the neighboring province, consisting mostly of Principes, Hastati, Equites of the Sacred Shroud and Archers and when the 400 Principes 70 Hastati and everything else actually begin to roll over you...

EA Ulm is similarly brutal with that mod, it will bury you if it gets to expand early on. The AI will also build a lot of castles, labs, temples and other stuff with the extra gold.

BI is not compatible with the full CBM, but you cna use parts of CBM such as the gods with that, since BI only modifies units that appear in independent poptypes.

[Zeldor]

Better Independents mod works really nice.

So difficulty would be that:
vanilla Impossible
BI mod Difficult
BI mod Mighty
BI mod Impossible
Multiplayer :)

[chrispedersen]

How do you use the map commands to make a god gandalf,.. brief overview?

[Gandalf Parker]

Nation number 11 is early Pangaea...
#god 11 812 -- Lord of the Wild
#comname "Green Mind"
#allies 11 26 -- likes to ally with Oceania if they are both AI
#allies 11 48 -- likes to ally with Oceania if they are both AI
#allies 11 40 -- likes to ally with mid_Pangaea if they are both AI
#allies 11 61 -- likes to ally with late_Pangaea if they are both AI
#allies 11 48 -- likes to ally with Oceania if they are both AI
#clearmagic
#mag_death 4 -- needed for Pangea spells and a bless
#mag_nature 4 -- needed for Pangea spells and a bless
#mag_blood 4 -- good combo with the above and a bless
#additem "Treelord's Staff" -- 2 more nature
#additem "Skullface" -- helmet with 1 more death and support
#additem "Armor of Souls" -- armor and 1 more blood
#additem "Boots of the Behemoth" -- goes well with his size 6
#additem "Brazen Vessel" -- misc slot and 1 more blood
#additem "Ring of Sorcery" -- 1 death and 1 nature and 1 blood
#dominionstr 11 6
#scale_chaos 11 2
#scale_lazy 11 2
#scale_cold 11 -1
#scale_death 11 0 -- this line might need removed
#scale_unluck 11 -3
#scale_unmagic 11 -1

You can also give him experience (stars). With these commands you can make anyone a god. A slinger, a turkey, a mummy, a horror. Giving Ermor a god of "Eater of the Dead" is fun.

The game will add up the points and if you go over then it will announce to everyone in the game that you have cheated and downgrade you. Unless you turn off ceat detection of course. It isnt a strict with AI gods. Giving AIs an unmoving god can help keep it from doing stupid things and killing itself off too early in the game.


Gandalf Parker

[Epaminondas]

The AI will buy staggering amounts of good quality stuff if the low quality chaff is unavailable. Try the Better Independents mod, which disables the crap infantry types by skyrocketing their gold costs. It leaves archers available and in addition to those, you'll start seeing a lot more nationals among the AI armies.

Nothing like seeing a scouting report that reads that there is an army of 750 units in the neighboring province, consisting mostly of Principes, Hastati, Equites of the Sacred Shroud and Archers and when the 400 Principes 70 Hastati and everything else actually begin to roll over you...

EA Ulm is similarly brutal with that mod, it will bury you if it gets to expand early on. The AI will also build a lot of castles, labs, temples and other stuff with the extra gold.

BI is not compatible with the full CBM, but you cna use parts of CBM such as the gods with that, since BI only modifies units that appear in independent poptypes.

Edi, I like your mod over the default game--by a lot. Still, the AI seems to prefer to build the cheapest units, rather than massing formidable sacreds (even when the AI has given itself a nice blessing scheme!).

[Edi]

Edi, I like your mod over the default game--by a lot. Still, the AI seems to prefer to build the cheapest units, rather than massing formidable sacreds (even when the AI has given itself a nice blessing scheme!).
Thanks. :D

The AI has never known how to use a bless effectively, so you can't ever count on it doing that. It will tend to go for a quantity of lesser troops over a quality of high end troops if the cost differences are big, but it gets markedly better still.

Picking the right nations to oppose you with BI will lead to really brutal games. The EA especially is bad. Choose one of the weaker nations, throw in Ulm, Ermor, Vanheim, Helheim, Tir na n'Og, Marverni, Hinnom, Agartha and C'tis as the opponents on a big map like Glory of the Gods and see what happens. Putting supplies at 150% will also give the AI a bit of a boost as it will not starve so easily and research to difficult so you don't get powerful magic too soon. You'll be in for a rough time.

I lose more games with that setup than I win, even when playing co-op with a friend, because the AI will simply roll over you if you don't get to a flying start.

[Epaminondas]

Edi, I like your mod over the default game--by a lot. Still, the AI seems to prefer to build the cheapest units, rather than massing formidable sacreds (even when the AI has given itself a nice blessing scheme!).
Thanks. :D

The AI has never known how to use a bless effectively, so you can't ever count on it doing that. It will tend to go for a quantity of lesser troops over a quality of high end troops if the cost differences are big, but it gets markedly better still.

Picking the right nations to oppose you with BI will lead to really brutal games. The EA especially is bad. Choose one of the weaker nations, throw in Ulm, Ermor, Vanheim, Helheim, Tir na n'Og, Marverni, Hinnom, Agartha and C'tis as the opponents on a big map like Glory of the Gods and see what happens. Putting supplies at 150% will also give the AI a bit of a boost as it will not starve so easily and research to difficult so you don't get powerful magic too soon. You'll be in for a rough time.

I lose more games with that setup than I win, even when playing co-op with a friend, because the AI will simply roll over you if you don't get to a flying start.

I was a bit surprised that you mentioned that Ulm was a strong EA nation early in the thread. Given that it lacks powerful battle mages and super-units, how is this so? I've found the nation frustrating to play, and the AI rarely does much with it either. I also tend to think Ermor would be weak, because of the old age issues with most of its top-tier casters.

But again, I am a beginner, and I look for any enlightenment on the issue.

Also, what difficulty setting do you recommend? I played 1 with Difficult, and I am gradually winning--though I suspect the fact that I have greatly strengthened PDs for every nation has something to do with it (the AI doesn't seem to use PDs much).

And yes, I hiked up Supplies to 300 percent--and Hinnom still has problems with starvation, LOL! :)

[Edi]

I mean that with the BI mod, Ulm is a really brutal AI enemy, especially with difficult research and impossible AI. It has good troops and with the BI disabled ones, it concentrates on its better nationals, expands fast and will come at you with massive hordes of good quality stuff.

Ulm is not the most powerful nation magically, but depending on pretender pick and other things, it can be successful in MP.

[Edratman]

I have just posted a mod called "Improved Computer Opponents" in the modding section of the forums.

The mod has five MA nations, T'ien Ch'i, Caelum, Arco, C'tis and Man modified to be better opponents. These are the 5 nations with the least amount of wins in the MP section, which is the sole reason they were chosen.

The mod also includes a version of the "No Independents Mod".

I cannot figure out how to make a link to the mod, so you will have to go there yourself.

With my testing, on Impossible settings, these nations are pretty close to impossible to beat (not that I claim to be an expert).

Try it and see.

[Epaminondas]

I have just posted a mod called "Improved Computer Opponents" in the modding section of the forums.

The mod has five MA nations, T'ien Ch'i, Caelum, Arco, C'tis and Man modified to be better opponents. These are the 5 nations with the least amount of wins in the MP section, which is the sole reason they were chosen.

The mod also includes a version of the "No Independents Mod".

I cannot figure out how to make a link to the mod, so you will have to go there yourself.

With my testing, on Impossible settings, these nations are pretty close to impossible to beat (not that I claim to be an expert).

Try it and see.

T'ien Ch'i and Arco are two of the least competitive nations in MP? I am surprised, given that both nations have a wealth of high-powered Astral mages. I suppose most MP games are rush-type of games where high-powered starting sacreds are over-emphasized?

[Epaminondas]

I have just posted a mod called "Improved Computer Opponents" in the modding section of the forums.

The mod has five MA nations, T'ien Ch'i, Caelum, Arco, C'tis and Man modified to be better opponents. These are the 5 nations with the least amount of wins in the MP section, which is the sole reason they were chosen.

The mod also includes a version of the "No Independents Mod".

I cannot figure out how to make a link to the mod, so you will have to go there yourself.

With my testing, on Impossible settings, these nations are pretty close to impossible to beat (not that I claim to be an expert).

Try it and see.

T'ien Ch'i and Arco are two of the least competitive nations in MP? I am surprised, given that both nations have a wealth of high-powered Astral mages. I suppose most MP games are rush-type of games where high-powered starting sacreds are over-emphasized?

Oops! I only play EA nations. So I don't know if my comments above apply to MA T'ien Ch'i and Arco as well!

[Epaminondas]

I would also have expected Ulm to be among the least competitive, but I suppose their starting troops are not so bad for rush-type of games.

[Zeldor]

MA Arco is one of the strongest nation in that age, not much weaker than Ermor or Pythium. TC is one of the weakest though when it comes to multiplayer. Man is without doubt the worst nation now.

[Edratman]

MA Arco is one of the strongest nation in that age, not much weaker than Ermor or Pythium. TC is one of the weakest though when it comes to multiplayer. Man is without doubt the worst nation now.

Man is very powerful in the mod.