House Rules against the AI

[Makinus]

I nearly exclusively play single player against the AI and i use some House Rules to give a "fair chance" to the AI, i´ll post my house rules here and i´m curious to other house hules:

1. Always use a Imprisioned Pretender.
2. Only attack an AI if attacked first (you are free to attack independents).
3. Only 1 researcher per lab.
4. Do not use battle spells that need gems.
5. Do not use global spells (You are free to Dispel the ones the AI casts).

By using these house rules and setting AIs to Difficult i find the SP more of a challenge...

So, any house rules against the AI that you think are fun?

[Sir_Dr_D]

Do not over bid on mercenaries.

[Sombre]

Don't bid on mercs at all :]

I also refuse to use geared out thugs (gear is reserved for ordinary commanders that have proven themselves)

I use NI maps.

I stale the first 5 turns.

I don't use archer decoys or arrange my troops in an overly clever way.

[cleveland]

Set all AIs to Impossible Aggressive, and starting provinces to 9.

Also set Supply to 300 and use Edi's Better Independents mod.

[Gandalf Parker]

There is a switch called SuperHost. It runs host over and over and over. You can add it to your dom3 icon and use it to force the game ahead many turns before stopping it, take the superhost off, and then join the game in normal mode with the AIs having a big head start.

Ive actually been outbid on mercs lately but the AI doesnt tend to bid more than a slight amount. Most of the outbids I thnk are the newer changes of "merc preference" which makes some mercs cheaper for certain nations to hire over others. My rules might include only one set of mercs at a time.

I like to RPG abit in my solo games so I treat AIs as if they were players. I even send messages. In some case, gifts. Particularly if I see that a nation I can probably beat easily is on the "other side" of some nasty nation. Such as, if I see that an undead nation is getting close then I might send "Herald Lances" to a nation on the other side of them, or even between me and them. The lances allow the AI to hold back the undeads but dont cause me much problem later.

I even form NAPs (non-aggression pacts) with AIs. On the one hand some will find that really funny. On the other hand some will say it works exactly the same as it does in MP games with humans. Everything seems good for awhile then suddenly the AI "changes its mind". At least with the AI its rarely a surprise attack. It tends to build up on your border giving you a chance to RPG an exchange of disappointment and threats of counter actions. Its good practice for the real thing.

Other things you can do:
Maps such as NI, Tower, or Pipe can give you a better game with AIs. Pushing up the resources setting for new games can help also. And lowering the strength of Independents.

You can add some commands to the end of your favorite map that makes the AIs allies with each other. Or team up. The commands are ignored for human players of those nations or if the nation isnt in the game so you can create some interesting alliances and forget you did it.

Maps can have designed AIs added. Things such as giving it only small gods, or immobile gods. The AI doesnt tend to use SC (super combatant, big mobile gods) very well unless its a small map. It also helps to use designed AIs so that it doesnt choose scales that can kill itself.

Other ways to get Designed AIs is to start the game as human for all players. Build the nations pretender and scales logically, then after a couple turns use the games Options screen to change that nation to AI for the rest of the game. Or use the SemiRand which can add random designed AIs to a map.

[Foodstamp]

Mercenary preference is definitely what causes the outbidding. For instance, if you are outbid trying to hire Hannibal, he probably went to Machaka. A year ago it was impossible to get anyone to believe this.

[pyg]

Set all AIs to Impossible Aggressive, and starting provinces to 9.

Also set Supply to 300 and use Edi's Better Independents mod.

I second this as a general setting as I dislike handicapping the way I play to accommodate the AI. I differ from these exact settings by using NI maps instead of BI mod, set starting provinces 2-5, independent strength 0-3, and use an all ages mod to select some good opponents.

[Fantomen]

I always play SP only to test out MP strategies, so crippling myself isn't really an option. I still enjoy those games and roleplay with myself on the way.

I prefer to buff the AI by designing good pretenders for it, basically maxed out scales since the AI can't handle SC or bless. With special considerations for nations that need it(LA ermor, Miasma, Mictlan). This also seems the most effective way to me.

[Sombre]

If you set the AI to impossible they get so many free design points and so much extra income that it totally outweighs economic scale design (though it doesn't change that they sometimes pick scales which are ridiculous for non economic reasons, like cold 3 with agartha or a misfortune/turmoil combo).

[Fantomen]

Are you sure?

The AI often funnels these design points in super uber magic paths that it is too stupid to use, then cheerily sends the supercharged pretender out to die. All the while succumbing under some completely ridiculus scales.

To me it has seemed more effective to design an imprisoned god, preferably immortal, with some death for skele spam and other paths suitable to cast the nations need of globals, plus good income scales and suitable other scales.

[Gandalf Parker]

Little human type pretenders tend to stay home, and of course immobile pretenders stay home. Its just the SC pretenders that tend to go out. And thats what they are meant to do.
The AI doesnt seem to have a concept of pulling back within the domain candles to recover and wait for new backup to arrive. But if the god goes too far and dies then it just does "Call God" and gets it back to start out again (losing some magic) so I guess piling magic on it fits the tactic. Not a GREAT tactic, but a tactic.

[Sombre]

Are you sure?

The AI often funnels these design points in super uber magic paths that it is too stupid to use, then cheerily sends the supercharged pretender out to die. All the while succumbing under some completely ridiculus scales.

To me it has seemed more effective to design an imprisoned god, preferably immortal, with some death for skele spam and other paths suitable to cast the nations need of globals, plus good income scales and suitable other scales.

Even with horrible economic scales, the income and resource bonus from impossible difficulty completely outweighs the negatives.

The AI gets + 100% to gold, resource and gem income on impossible. That's far more than going from all negative 3 scales to all positive 3 scales can net you.

So to reiterate: Don't start the game with human players with sensible scales then set them AI to try and give the AI an economic advantage. It doesn't make sense. You can give it a research advantage or something more marginal to do with cold blooded etc, like I said, but economically speaking the AI is always best off being set to impossible.

[thejeff]

That may be what SC gods are supposed to do, but the AI can't handle SCs.
They go out with random or no gear, run up fatigue casting marginally useful buffing spells and get killed.

It's fine to say that's intended, but it doesn't work.

[Gandalf Parker]

That happens every time? :) Or randomly?
I never said it did it well.

[Tollund]

It happens virtually every single time. The AI will take an SC pretender, not give it any equipment, have it cast flying shards once then attack.

[rdonj]

When I've had the AI draw one of the bulls as a pretender, it tends to not completely waste it. Generally, imo, good pretenders for the AI are pathless SC pretenders. The AI will usually get best use out of these, and if it dies it can just be resurrected with no penalty. Of course, it will eventually become useless anyway due to afflictions unless it has nice regeneration and or recuperation.

[Makinus]

My major gripe with the AI is that it´s incapable of keeping up in the research curve... it simply does not research fast enough.... thats the reason of my house rule of 1 researcher per lab... it´s the only way for the AI to keep up...

[Edi]

The AI seems to do fairly well with independents at strength 7 and impossible AIs, especially if it has good scales (happens sometimes). Indies 7 means you need a lot more troops to beat them than indies at 5, and the AI resource and gold bonuses mean that it gets going pretty fast. It's not at all uncommon to get swamped by two or three hordes while you're still trying to get basic things done.

I play with the Better Independents mod and difficult research, so that slows me down a lot more than it does the AI.

[Gandalf Parker]

There are pros and cons to the AI doing things totally randomly.
On one hand, it usually gets it wrong. On the other hand it can surprise us.

Good AI (artificial intelligence) is not good AH (artificially human). Good AI is the intellgent choice. "A straight line between two points" which is too predictable.

Good AH is almost achieved by "random selections filtered thru intelligent choices" but that takes alot of code. Even for ONE of these nations such a set of rules would be huge to do it with any variety for replayability.

The best bet WE will ever see are more designed AIs for each nation being submitted to the SemiRand. Or a really good mod builder creating an AI.

"AI programmers consider it a giggle that to go from AI to AH requires dumping some intelligence in favor of randoms. There must be a comment there about the human race."

For more reading: the Association for the Advancement of Artificial Intelligence
http://www.aaai.org

[Tollund]

There are pros and cons to the AI doing things totally randomly.
On one hand, it usually gets it wrong. On the other hand it can surprise us.

How can it surprise us? By doing exactly the same thing in every game?

Good AI (artificial intelligence) is not good AH (artificially human). Good AI is the intellgent choice. "A straight line between two points" which is too predictable.

And yet, the best chess programs in the world work in exactly that way.

Good AH is almost achieved by "random selections filtered thru intelligent choices" but that takes alot of code. Even for ONE of these nations such a set of rules would be huge to do it with any variety for replayability.

I'd rather worry about competence before I worried about replayability. The AI is currently incompetent at anything beyond pushing armies around the map. And yes, I mean it when I say that it's incompetent. It only has a chance against a non-beginner when the odds are more than three to one in its favour.

The best bet WE will ever see are more designed AIs for each nation being submitted to the SemiRand. Or a really good mod builder creating an AI.

How, exactly? No mod can change the AI behaviour in the game. Better pretender designs won't fix the lack of coherent pre-designed plans for the AI. Allowing the AI to cheat by giving it massive starting advantages isn't a solution either.

[thejeff]

What it really needs to be able to do is adapt, at least marginally. Your enemy's attacking with fire, give fire resistance a higher priority. Astral mages - boost mr, undead build and send more priests.

Some of this, even within the basic combat spellcasting AI would be possible. "Lightning bolts crashing all around me? Hey, I can cast Resist Lightning!" (This would help the mp game as well, if mages could react to the situation.)

Exploitable by a clever player, sure. But far less than it is now.

Some basic guidelines on gearing/buffing thugs would be nice, again modified by what's been killing them lately. Some awareness of blesses, sacred troops and the need to send priests along with them.

Basic attention in pretender design to scales. Heat preference, age of mages, resources needed for troops, etc. Keep it randomized, but weight things that make sense more heavily. Not predetermined weights by nation, but derived from the actual game data.

None of this is the simple "design a script for each nation" strawman, nor is it build a functioning human AI, pass the Turing test, win fame and fortune. It's complicated, it's a lot of work, but it's not fundamentally a hard problem.

Of course, it's not going to happen. Would require a major rewrite and the devs have moved on.

[Lingchih]

My god, is this even a worthy topic? One uses the SP AI to test out strats and summons. Playing a game against it is like playing against a six year old.

[Makinus]

Unfortunatelly a lot of people (myself included) have problems playing MP games because of variable work schedules and SP is the only alternative to playing Dominions... I have months where i can play almost every day, while other times i must stay 2 or 3 weeks without playing... this makes difficult to play MP games...

[Humakty]

Dominions AI isn't much worse than any other (ie : Civ 4 AI), it's the game which offers so many different tricks so quickly. I mean, by year two you'll have around 100 spells available, how to you want an AI to use efficiently all of those ?

The only really bad point dom 3 AI has is the propention not to buy any mages whatsoever, partially cured with NI maps or BI mod. That, and recruiting militia...*sigh*

[Gandalf Parker]

Its also not a bad AI considering how it was created. Dominions was created with a minimal plugged-in AI. And the game has grown in pieces. So even "easy" fixes dont tend to be so easy. Such as "use of lightning should mean use protection against lightning". But there does not seem to be a simplye single variable for lightning in the code. Each item and spell was coded separately without much in the way of categories. So the code would have to maintain a huge list of spells and counter items. Thats why we still find holes where some unit is supposed to be immune to something, or some spell isnt supposed to work. Because each one needs to be noticed and tagged.

We have gathered much info on what the AI does and doesnt do. What pretenders it uses well and which ones it doesnt. Units it uses well enough and which ones to avoid. Spells it prefers to use so we might be able to selectively grant more or less ability to use the ones we feel it uses well. Eventually I will use all of this to give us all a fantastic modded nation designed for no other purpose than to be a bigger better AI for the game. Of course this will be done by me creating some horribly crappy little mod that will instantly spur someone to show me up by doing it much MUCH better. :)

[Tollund]

You know Gandalf. We're all quite capable of reading this thread and don't really need you to repeat the things that other people have already said in it.

[Sombre]

Gandalf: A mod won't be made that makes the AI a worthwhile opponent. What you're talking about could already be done easily by any number of modders on these boards and you could easily create an AI nation that would destroy even the best players in SP, no problem. But it would still suck to play against because we can't mod the AI. Giving it nothing but the incredibly simple stuff it can actually cope with requires making this stuff ridiculously powerful for it to stand a chance. That is zero fun to play against.

You go on and on about this all over the place, talking about AI mod nations and mods designed to improve the AI and blah blah blah. It's all just wishful thinking. Oh and shill work of course.

[Gandalf Parker]

Well shill is one thing I do well. :)
But you are probably correct that its not possible to mod a worthwhile AI. Even if we could mod the AI I dont think thats possible. Its not even possible to program a worthwhile AI. Especially not for this game.

But since there are people who play the AI, and there are known features about the AI, then it should be possible to mod a better AI than the random that most people are playing against now. It is likely though that like other projects it wont happen without a shill effort first.

[Sombre]

Good luck with that. How many years have you been talking about it again?

[Tollund]

Of course it would be possible. Just create a nation where the AI can recruit juggernauts as troops and vampire lords as commanders for 1 gold and 1 resource each. It would easily crush any human player then, but still wouldn't be a very interesting opponent.

Anyways, you're still conflating the AI with nation/pretender design. The AI is the decision making engine. It cannot be modified, changed, or affected in any way by the players. One can create better pretenders and cheat in other ways for the AI, but this is _not_ AI modding.

[kenmtraveller]

I haven't played MP, but have played SP exhaustively.

My latest rule , is

1) all AIs on impossible and aggressive
2) I'm not allowed to recruit troops or mercenaries. I only get starting troops, commanders, troops from random events, and nationals summons.

Ken

[Gandalf Parker]

Good luck with that. How many years have you been talking about it again?
Heehee. Yeah I end up doing about 1 a year. Not like some of the fast burner mod makers.
And dominions has dropped on the list. As far as this project goes I havent gotten very far in even play testing my own efforts from the other thread, or the mods done for AI such as NTJedi. On the other hand mentioning it shouldnt hurt since there seem to be people who speak up that have much more specific info on what the AI does and doesnt do well than I have.

[Edi]

Since the AI can't be modded directly, about the only way that it can be affected is by limiting its choices, which is the driving force behind both the Better Independents mod and the No Independents maps.

They work fairly well given the limitations, but fighting the AI is always dealing with hordes and fairly little magic (in light of the enormous potential that you see wielded in MP games). With the BI and NI solutions, you just face hordes of quality troops and more national troops instead of identical militia/light infantry hordes.

Something that would help if it were possible to set non-random AIs, ala creating an AI pretender with the point bonuses that could only be assigned to an AI nation (just like you can create and load pretenders for human controlled nations). Sure, you'd know what kind of build you're up against, but it would make for a better SP game experience when you don't have half the AIs turmoil/sloth/death/misfortune 3, which causes them to fall apart after around twenty turns.

[Foodstamp]

I hear a lot of people say the Better Independents mod works well but I have to ask how you got around the huge flaws that the mod had before.

In the past the Better Independents mod either increased the gold cost or the resource cost of the bad Independent units. This created two scenarios that pretty much destroyed the AI.

If the gold cost was raised, Militia events would devastate the AI's economy.

If the resources were raised, the AI would not be able to produce units in any castle except the capital because it would queue up the independent units that had the resource cost it could never hope to meet ultimately only allowing armies to be built at the capital.

How did you get around these limitations?

[Edi]

The current version fixes it so that units generated by events don't have altered gold costs but the same units in national rosters are replaced by a copied unit that does have higher cost. I never ran into the queue problem myself.

[Gandalf Parker]

Something that would help if it were possible to set non-random AIs, ala creating an AI pretender with the point bonuses that could only be assigned to an AI nation (just like you can create and load pretenders for human controlled nations). Sure, you'd know what kind of build you're up against, but it would make for a better SP game experience when you don't have half the AIs turmoil/sloth/death/misfortune 3, which causes them to fall apart after around twenty turns.
Thats part of the SemiRand program. Run your map thru it before playing and it will attach a randomly selected pre-built pretender/scales for the AIs you select. Most are versions of player builds but some by Stavros are quite cranked and well thought out.

[Foodstamp]

The current version fixes it so that units generated by events don't have altered gold costs but the same units in national rosters are replaced by a copied unit that does have higher cost. I never ran into the queue problem myself.

What about the standard militia event?

To test the resource hang up I am talking about, you only need to take over any AI controlled castle besides their capital. I am pretty sure the AI makes the same decisions in provinces without castles but you just don't see the queue hang up because that bug only allows you to see the queued units when you conquer a castle.

If the AI doesn't do this anymore, I may give it another shot!

[Gandalf Parker]

Does it show up if you open an AIs turn file?

[Lingchih]

Dominions AI isn't much worse than any other (ie : Civ 4 AI), it's the game which offers so many different tricks so quickly. I mean, by year two you'll have around 100 spells available, how to you want an AI to use efficiently all of those ?

The only really bad point dom 3 AI has is the propention not to buy any mages whatsoever, partially cured with NI maps or BI mod. That, and recruiting militia...*sigh*


Really, Humakty? Dom AI is about the worst there is. Civ, GalCiv, a dozen other games, beat it by a mile. This game is all about MP.

[NTJedi]

My House Rules against the AI:
1) Typically I provide every AI opponent with Order_3, Growth_3, and Luck_3 using the mapedit commands. This really makes a massive difference.
2) The game is setup with at least Independents of 7 or higher.... and Supplies at 300.
3) I never build the unique artifacts and never bid on mercenaries.
4) I use the AI Opponent Balance Mod: http://forum.shrapnelgames.com/showthread.php?t=42679
5) For large maps I provide one AI opponent(usually Niefelheim or EA Argatha) with at least a dozen SCs, an extra province with a fort, immobile mages and library(!!research!!), then last a super powerfully equipped pretender.

** And to increase the challenge one only needs to setup two, three or four AI opponents as allies.

But it would still suck to play against because we can't mod the AI.
That's why I'm looking forward to playing Elemental: War of Magic when its released during August of 2010. Besides regular modding the game will also allow modding of the AI opponents.

[Foodstamp]

Yeah it's been very difficult for me to want to mod Dominions anymore with Elemental on the way. My mod time has gone to learning Python better. I wrote Pong, Hoorah!

[Edi]

The current version fixes it so that units generated by events don't have altered gold costs but the same units in national rosters are replaced by a copied unit that does have higher cost. I never ran into the queue problem myself.

What about the standard militia event?

To test the resource hang up I am talking about, you only need to take over any AI controlled castle besides their capital. I am pretty sure the AI makes the same decisions in provinces without castles but you just don't see the queue hang up because that bug only allows you to see the queued units when you conquer a castle.

If the AI doesn't do this anymore, I may give it another shot!
The standard militia that appears by event is unmodded in terms of gold cost. It will not cause bankruptcy. It will also be recruitable in the gold cost version from those poptypes that have it (two or three, so you see some of it in the AI armies). Nations that had other crap in their rosters still have them, but as copies. So e.g. if they had light cavalry 26, they now have other units (unused) that had light cav copystatted on them.

If the queued up units don't show outside castles, then yes, that would be a problem for the resource mod. I use the combo or the gold cost version myself most of the time. So the resource version may or may not suffer from that problem.

The only problem with gold cost mod is that enslaving indies via astral magic can be hazardous to your finances, so a look at the mod code and the poptype list in the Dom3 DB is advisable before experimenting.

[Humakty]

@ Lingchih :

Civ 4 AI is all about cheating, and much more so than dom 3 AI. Even on noble difficulty, it gets bonuses, and, mind you, there isn't anything too complicated to do in civ, well nothing as complicated as dom 3 spell system.

Civ 4 AI, at emperor difficulty, has at the very least ten different bonuses, ranging from troop XP to Great People generation, really every kind of bonus you could think about is covered.

So to evaluate Civ 4 AI, you'd have to play on warlord difficulty, were you could realise it is utter crap, not able to do anything consistent, not even by accident.

In conclusion, the only merit of civ 4 AI is to have much more bonuses than dom 3 AI, and I would in no way compare it to Galciv AI, which is able of a somewhat consistent gameplay.

[Makinus]

I recently bought Sins of a Solar Empire (the Trinity pack) and the AI seems to be very good.... i am mostly playing the AIs in Normal difficulty yet, but according to the developers, up to the Hard AI, it does not have any kind of bonuses, only it uses more advanced tactics, only in the difficulties above Hard the AI start gaining bonuses...

Seems to me to be a very good AI design... spceially considering that i still didn´t beat the AI on Normal difficulty yet....

[Humakty]

I only have basic SoaSE game, but I do get regularly kicked over by normal AIs too ( I love playing the cool psychic ones ). I don't play much though, base game is kinda limited, do you think the expansion pack is worth it ?

[Belac]

Actually I think the AI is pretty decent as games go. Much better than the CIV4 AI in that it is willing to attack other AI nations and can master a few basic battlefield strats (even if all that is is artillery+waves of chaff+flanking cavalry, that's something).

My house rules involve massively rich settings and hard research. The time of game when the AI is most challenging is early midgame, when as a player you want to be building forts+mages and racing up the research tree, but instead you have to keep armies around designed to beat waves of chaff. I just give up on reaching the endgame and play like there's nothing above level 6 magic.

[Saarud]

Well I only play SP since I can't be sure to keep up with the turns in a MP game even if I would prefer that. I only have a few houserules I follow and those has already been mentioned earlier in this thread.

My favorite way of playing SP is simply trying to break my own records in winning as fast as possible. While the AI won't challenge me, my earlier win records will. My favorite map are Dawn of Dominions (I think that's the name) where about 10 or so EA nations are preset on a very nice map. The starting locations are not balanced but very thematical placed which makes me more immersed into the game.

[Gandalf Parker]

Ive seen some great advances in the subject of improving the Computer Players in Dominions 3. We still could use a thread specifically listing the settings, maps, mods, and 3rd party programs that are directed toward AI improvement. Possibly sticky, altho we might have to wait until a new regime for that.

Some levels for Indepts above 9 tend to go AI'd, Allied, SemiRand, Chaos, Designed, GameMaster.

Some MODs would be Better Independents or the AI Opponent Balance Mod.

Some MAPs such as the Tower Map seems to help, and the NI (No Independents) maps. And now Im adding another map. Based on the idea that the NI maps help the AI use more elites, and cuts micromanagement, I have created a PI (Partial Independents) map. Approx 1 out of every 10 provinces will allow their usual recruitment of independent units. The others still force you to fight for them but do not allow recruiting (unless a magic site provides something). To keep the locations from getting memorized from continual replay, my server will re-randomize this file daily. Check out the PI version of the Parganos map called Partial_anos on my server.
http://www.dom3minions.com/maps.htm

[JonBrave]

I'm deliberately reviving this thread! :eek:

When I first joined this forum, I tried to follow this guy's advice:
I nearly exclusively play single player against the AI and i use some House Rules to give a "fair chance" to the AI, i´ll post my house rules here and i´m curious to other house hules:

3. Only 1 researcher per lab.


And later he confirms:

My major gripe with the AI is that it´s incapable of keeping up in the research curve... it simply does not research fast enough.... thats the reason of my house rule of 1 researcher per lab... it´s the only way for the AI to keep up...

Now I'm still a simpleton a while on, and this is depressing. I don't know who this Makinus is, but has anybody actually tried beating some (Impossible) AIs with restriction of you can only have 1 researcher per lab all game? Because this has haunted me, I certainly couldn't achieve it...

[krpeters]

I don't know who this Makinus is, but has anybody actually tried beating some (Impossible) AIs with restriction of you can only have 1 researcher per lab all game?

Was his pretender a Great Sage, and was his one mage per lab a Starets?

[JonBrave]

He was suggesting generic house rules. I don't think the point is what sages you have!

[Edi]

Depending on what nation you play and what you set the indies level to, chances are that several of the impossible AIs will be ahead of you on the research curve for a fairly long time. Especially if you get a slow start.

[Knai]

Depending on what nation you play and what you set the indies level to, chances are that several of the impossible AIs will be ahead of you on the research curve for a fairly long time. Especially if you get a slow start.

Yes, but you can probably keep up well enough when you consider the difference in efficient research as opposed to wasted research. This becomes particularly notable when you start handing out research boosters.

Besides, one only has to beat the AI. At that point, how you use your research is far more relevant than what you have. An impossible AI is pretty easy to drop with only one researcher per lab, particularly as it doesn't bar mages being used for non research purposes.

[JonBrave]

An impossible AI is pretty easy to drop with only one researcher per lab

Well, there you are. So you agree with the original poster.

For my part, currently happening to play Dawn map against all Impossible AIs, I have spent the first 8-odd turns hiring researchers at my single lab, and looking at the chart I'm about half way up the scale compared to other AIs. So how I would beat them with just one researcher is beyond me.

Like I asked, has anybody actually (regularly) done this, as opposed to saying how easy it would be?

[thejeff]

I haven't done this, but I'd assume the technique is to have fast expansion which doesn't rely on research, then lever that expansion into lots of labs.
Unless you start with an awake research pretender your research will suck at the end of 8 turns. You'll have to make up that deficit later, with more labs & castles.
I'd say the easiest way would be to start with a high bless nation, Mictlan or Niefleheim, for example. You hardly need researchers at all to beat the AI. Do that on tight map to let you fight and win earlier in the game.
I have my doubts about pulling it off with a mage (rather than sacred or thug) based nation.

[Gandalf Parker]

One of the options on AI games on my server is to run the game X turns before letting the players in.

Altho the if its a server game there are better boosts for AI. Allying them with each other, giving them equipment at the start, or an extra starting castle. And of course the best thing is assigning them a well-made pretender

[Knai]

For my part, currently happening to play Dawn map against all Impossible AIs, I have spent the first 8-odd turns hiring researchers at my single lab, and looking at the chart I'm about half way up the scale compared to other AIs. So how I would beat them with just one researcher is beyond me.

Dawn of Dominions is fairly stacked against you. That said, against the AI there are a few keys.
1) Province Defense. The AI is monumentally stupid, and will routinely split large forces, kill them off against PD individually, and not learn from its mistakes. PD is a good investment against the AI, and can buy you time.
2) Thugs and Supercombatants. These are amazingly good against the AI, which tends to be pretty bad at anything beyond troop swarming. An army with three banes with a fire brand, vine shield, and amulet of luck stuck way out front will be surprisingly effective, and this requires surprisingly little research.
3) With the 1 researcher per lab requirement, you will want most provinces to have labs. This includes those without forts.

I have killed a bunch of impossibles with the 1 researcher restriction using these keys multiple times, on varying maps. That includes Dawn of Dominions, playing as Caelum.

[triqui]

I don't remember which difficulty it was, becouse it happened a lot of time ago. I doubt they were Impossible, but probably Mighty. But I once defeated 4 computer nations, with no research at all. W9F9B4 Jaguar Warriors, that's all you need.

The easiest way to defeat the computer is using PD smartly. Having triple blessed PD helps a lot too...

I once beat AI using nothing but the starting army and independent troops (and a SuperCombatant Pretender), plus national mages.

The AI has several big holes. Not only they aren't very original or smart in general, but they can't face some strategies at all. SC for example, is something that often defeat it. Scripting decoys to retreat to disable archers, using mass remote spells, that kind of things are very hard for AI to face.

[Knai]

The AI has several big holes. Not only they aren't very original or smart in general, but they can't face some strategies at all. SC for example, is something that often defeat it. Scripting decoys to retreat to disable archers, using mass remote spells, that kind of things are very hard for AI to face.

There are a handful of spells that will just wreck the AI. Anything that summons an SC, Flames from the Sky, Mind Hunt, Enslave Mind communions, so on and so forth. They can't counter SCs, they won't make smaller armies to mitigate Flames From The Sky, they don't hand out magic resistance gear and Mind Hunt works beautifully as a result, so on and so forth. Then there are the globals, where the player can easily have five up.

[JonBrave]

I have killed a bunch of impossibles with the 1 researcher restriction using these keys multiple times, on varying maps. That includes Dawn of Dominions, playing as Caelum.

I bow, scrape & grovel :D Fair enough, that was what I asked for. You are a much better player than I am ;)

I do take your points --- still not sure how you manage it early --- thanks.

I am retrying against AI @Impossible, but this time not losing my Pretender on turn #3 and then having to pray for her return, and I must admit it is going better......

[thejeff]

I don't remember which difficulty it was, becouse it happened a lot of time ago. I doubt they were Impossible, but probably Mighty. But I once defeated 4 computer nations, with no research at all. W9F9B4 Jaguar Warriors, that's all you need.

The easiest way to defeat the computer is using PD smartly. Having triple blessed PD helps a lot too...


In other words the strategy to beat the AI with limits on research (whether none or 1/lab) is to use strategies that don't rely on research. High bless sacreds or maybe SC pretenders.
It probably also helps to use a fairly small map. With ~20 opponents you might have had more trouble by the end.

The other easy choice for a nation would Nieflehiem. E9N4+ bless. Buy nothing but Neifel Jarls, and Giants when you have cash left over. Maybe 1 PD, just for visibility.

A interesting challenge I played with a few years ago was LA Ermor with an imprisoned pretender. No research, or extra reanimators, until he breaks free. Though you could get lucky with Etimmu or hire mercs to build labs for indy mage recruitment.

[Makinus]

I never managed to defeat Impossible AIs using the 1 researcher/lab rule, but i can half the time do it against Mighty AIs and consistently against Difficult ones...

I like to play with all empires in a given era, so the maps are normally large (i specially like the Latus one, it is good for some epic games).

I rarely finish the games i start with, normally ending the game when i´m at 1st place with a considerable lead on the 2nd place empire.

As others have said, some empires/tactics make it easier than others... bless factions have a clear advantage here, also a good SC pretender from the start (or at least dormant) will make you go a long way.

Even with the 1 researcher/lab rule i notice that the AIs rarely get ahead of me (with some exceptions when playing AIs above Difficult level), specially since i use all research boosters possible. Lately, instead of building labs everywhere i´m also restricting myself to build labs only where i can recruit a mage type commander (either from a castle, indys or sites).

Finally, i noticed that a higher number of mages in combat will make, at least in the beggining, a good boost to conquering indy provinces... massed fire flies/darts are fun...

[triqui]

I like to play with all empires in a given era, so the maps are normally large (i specially like the Latus one, it is good for some epic games).

Very big maps benefit the AI in my experience. If only becouse boredom makes micromanagment very unfun at large scales, so I tend to overlook a lot of things in the late game, do half-arsed turns and so on.

[Gandalf Parker]

I like to play with more than all the empires of an age and maps larger than large. Maximum on both even. (didnt finish the game of course, but it was fun not to know what nation you might run into)

In case anyone is wondering:
the maximum map is 1500 provinces.
All of the nations in the game is 73
the maximum number of nations in a game is 99

If you chaos the map, throw in random alliances between the AIs, and add in the Special Monster AIs as Independents then you can get a game which can continue much farther than usual.

[Knai]

In other words the strategy to beat the AI with limits on research (whether none or 1/lab) is to use strategies that don't rely on research. High bless sacreds or maybe SC pretenders.
It probably also helps to use a fairly small map. With ~20 opponents you might have had more trouble by the end.

Don't forget about scales. Taking something like Caelum, Niefleheim, or Abyssia you can take Order 3, Luck 3, Growth 3, Cold/Heat 3, and Magic 3 if you don't bother with a bless. Early on, focus on troop use - the AI is good at massing troops, but their ability to pick them is substandard at best - use heavy PD, and just expand outward leaving no province undefended. You will want some low level of research, focused towards getting thugs and battlefield mages in play, these back your armies up. Eventually, the battlefield mages will be far more impressive, and the thugs will have transitioned to SCs, at which point you basically win.

Against a player, this strategy is a good way to die horribly. It only works against the AI because the AI is incredibly dumb, and even then you have to be careful.

[Starbelly Geek]

Against the AI, I like to set Research to Very Difficult and pump up the independents to slow things down, and set up large games on larger maps, because a lot of the fun for me is finding stuff randomly dispersed around the map. Hidden Kingdoms and Towers of mages and stuff are awesome. Slowing down the game lets me play around with all of that.

I also like to go hard on the roleplaying aspects, so, for example, try EA Pangaea with 3 Turmoil, 3 Sloth, 3 Growth, 3 Luck and 3 Magic and never hire human troops (except maenads - lots and lots of maenads). Play MA Marignon without ever using undead or blood. Try always to rely most heavily on your sacreds as EA Agartha, eschewing inferior surface dwellers. That sort of thing can be amusing, although the AI folds pretty quickly anyway.

[Makinus]

I too like to RP a lot in games... like never using death/blood magic with "good" nations and so on...

I find also challenging to try to play as a "good guy" in a more "evil" faction... like using blood oriented nations without using blood hunting (i only use the natural blood supply from sites, etc.)... the main advantage is that i reduce the horrible micromanagement of blood economies and at same time give the AI a small advantage...

As from my original house rules, i also limit myself only to attack AIs that attacked me first (simly declaring war does not count)...

[triqui]

I wasn't aware of any good nation in Dominions 3 :p

[Makinus]

I wasn't aware of any good nation in Dominions 3 :p

The majority of the factions can be played as neutral or evil empires, but only a few can be roleplayed as a good nation... from the top of my head i can point out Man and Arcoscephale (except for the slavery thing)... most of the other nations can be considered neutral at best, with a few (MA and LA Ernor, Ryleh, etc.) being outright evil...

I believe there was a thread about good/evil factions some time ago and the general consensus was that all factions could be played as "Evil", about half as "Neutral" and only a handful as "Good" nations...

[triqui]

MA Ermor is not evil, just misguided :P

[thejeff]

Isn't a lot of the fluff for MA Ermor about using Death magic to hold back the Death unleashed at the end of EA Ermor? You could definitely spin that into at least a neutral rp.

Of course, everyone just unleashes as much death as they can, cause that's the most effective way to play it, but that doesn't mean you couldn't.

[Gandalf Parker]

Here are a couple of examples of boosted AIs.
This text was added to a map creating better AIs

a game with lvl 3 boosted AIs
http://www.Dom3Minions.com/~gandalf/NewbsandVets1_boosted_AI.txt (http://www.Dom3Minions.com/%7Egandalf/NewbsandVets1_boosted_AI.txt)

a game with lvl 6 boosted AIs
http://www.Dom3Minions.com/~gandalf/NaVII_AI.txt (http://www.Dom3Minions.com/%7Egandalf/NaVII_AI.txt)

the next game was supposed to be lvl 10 but the players opted to lower that after trying the previous two. I need to dig that one up

[triqui]

Isn't a lot of the fluff for MA Ermor about using Death magic to hold back the Death unleashed at the end of EA Ermor? You could definitely spin that into at least a neutral rp.

Of course, everyone just unleashes as much death as they can, cause that's the most effective way to play it, but that doesn't mean you couldn't.

Yep, that's what I mean. Besides that, Death magic does not mean "evil" magic. Tien Chi Ancestors are summoned by death gems, and that do not qualify them as "evil"

[Makinus]

Thinking about you said mad me realize you are right... Death magic is not necessarily evil except for some spells/summons/itens, but other paths also have "evil" spells/summons/itens without making the path itself evil...

The only really "evil" path in itself is Blood...

[triqui]

Yep, it's hard to argue that Blood is evil.

However, some nations might use Blood for reasons they think are not devil. LA Marignon comes to mind. Inquisitors think they are doing "the right thing"

[Gandalf Parker]

One of my favorite nations is Pangaea. I would consider that to be an example of "we are not evil, we are right" nations. I think of Pangaea as the defenders of Nature. Using human sacrifices to get more defenders is considered necessary.

I think this is something that plays into the RPGers. The strategists might have quick pat answers for evil that the RPGers can role play out of

[Marmaduke]

I am not a powergamer, but I generally use:
- all nations from the chosen era, mighty
- large map
- indy 9 (I like it when Independents stay on the map very long)
- magic sites EA 65 / MA 70 / LA 75 (you guess it I dislike gem poverty)
- hof 15
- very difficult research
- victory points : 7 required, 1/capital (I can't be bothered to finish a game with full conquest)
Also, I don't bid on mercenaries but I might go back to that (no overbidding though).

[triqui]

Even for some "clearly" evil nations like Hinom, things are debatable.

I wouldt say jews are evil becouse they sacrifice lambs, or christians are evil becouse they eat turkey in christmas. They dont find themselves doing evil things, simoly becouse turkeys and lambs are inferior beings and they feel those are there to feed on them or force them to work (like horses). For Hinom, it is the same. Humans are for them so inferior as turkeys or lambs are for human beings.

[JonBrave]

I never managed to defeat Impossible AIs using the 1 researcher/lab rule, but i can half the time do it against Mighty AIs and consistently against Difficult ones...


Well hello there, Makinus! I resurrected this thread because, ever since reading your starting post when I started playing Dom, I have been very disturbed by my inability to win Impossible with just 1 researcher/lab..... I had assumed you had (shame on you!), now I hear you haven't I'm a lot happier ;)

[Makinus]

The main trick is to use massed mages with low-level battles spells (fire flies/darts, that one that trows rocks, etc.)... 7 or 8 mages spamming low-level attack spells can do a suprisingly high amount of damage needing only a small screen (20 shield troops or so)...

[Knai]

I think this is something that plays into the RPGers. The strategists might have quick pat answers for evil that the RPGers can role play out of

Strategists and RPGers are mutually exclusive categories now? That's new.

Also, it isn't RPGers. It's roleplayers.

[Gandalf Parker]

No I dont consider them exclusive. But they do represent different takes on the game, and SOMETIMES at odds.

And I did mean RPGers. I was referring mostly to the game styles. As in "people who play RPGs". If anything I guess that Strategists was wrong since "people who play strategy wargames" would best be referred to as Grognards but I wasnt sure how many here would get the reference. Googling RPGer or Grognard gives results which match what I had in mind.

I guess for your sake if I was answering officially it might have been best to say "This is something that plays into the RPGer in us. The Grognard in us might come up with a pat answer for evil that our RPGer can role play out of."

[Rookierookie]

Very large map
- Custom AI pretenders
- Indie strength 9
- Very difficult research
- Resource: 100 EA, 150 MA, 200 LA
- Supply 300
- Recruit exactly 1 of each kind of mage per province

The AI may not know how to deal with SCs or PDs, but given enough time to develop they will have multiple 500+ armies and 10+ mages that eat the average well-geared SCs. (Note to self: never send SCs against Ri-yeh)

[krpeters]

So what strategy do you employ against an AI horde with 500+ troops and 10+ mages?

[Doo]

Burden of Time

[Knai]

So what strategy do you employ against an AI horde with 500+ troops and 10+ mages?

Remote assassination spells are your friend here. Mind Hunt, Earth Attack, so on and so forth. Plus, you can just field much smaller armies backed by far fewer mages, and actually spend gems intelligently to win the fights effortlessly. Army of Gold, Earthquake, Earthquake, win for instance.

[JonBrave]

So what strategy do you employ against an AI horde with 500+ troops and 10+ mages?

Ah, but what about following the rules from the OP?:

3. Only 1 researcher per lab.
4. Do not use battle spells that need gems.
5. Do not use global spells (You are free to Dispel the ones the AI casts).


So far, we only discussed 1/lab.

Burden of Time

What about with no Globals? :)

Plus, you can just field much smaller armies backed by far fewer mages, and actually spend gems intelligently to win the fights effortlessly. Army of Gold, Earthquake, Earthquake, win for instance.

What about no battle gems? :)

[Knai]

What about no battle gems? :)

There are plenty of spells that are pretty nasty without battle gems. Smaller buffs, stuff like blade wind, so on and so forth. Plus, against the AI you can have most every magic path, and between assassination spells and Flames From The Sky armies are fragile, fragile things.

[rdonj]

So what strategy do you employ against an AI horde with 500+ troops and 10+ mages?

The last time I had to fight AI armies of such size, I would maybe throw a flames from the sky in that direction, then drop a mage SC in on them who can stand in melee while spamming aoe spells.

If you don't have access to such things, then you'll need to deploy an army that can hold the AI army down while you use some BE/battlefield wide spells to bring them down. I use the word army loosely and it doesn't need to resemble an army at all. But some useful spells could include blood rain, wailing winds, fire storm, rain of stones, earthquake, soul drain... the list goes on. And of course things like skellyspam are very handy to keep the opposing army occupied while you slowly chew through them.

[Rookierookie]

By the way, how does the AI use its gold? I was wondering if increasing the income would help the AI, since I rarely if ever feel in need of money in SP.

[rdonj]

The AI uses its money to buy anything and everything it can afford. Increasing the income *would* help the AI, but it would help you just as much. If you're trying to boost it that way, taking poor scales to compensate for the increased money would be necessary to avoid negating the advantage to the AI. It will recruit even more chaff that way though.

[Gandalf Parker]

There are also no-independent maps which forces the AI to spend money on elite troops and mages at castles instead of cheap independents all over the map.

Personally I consider those abit extreme altho a fun game variant. I prefer the partial NI maps where something like 1 out of 10 provinces is left with whatever random recruitables the game gives it. Thats the same effect of forcing the AI to better spending but it still allows for the joy of exploration, finding, and securing prime locations

[Knai]

There are also no-independent maps which forces the AI to spend money on elite troops and mages at castles instead of cheap independents all over the map.

Personally I consider those abit extreme altho a fun game variant. I prefer the partial NI maps where something like 1 out of 10 provinces is left with whatever random recruitables the game gives it. Thats the same effect of forcing the AI to better spending but it still allows for the joy of exploration, finding, and securing prime locations

There is also the no-indpendant, special sites that give recruits option. Basically, a bunch of new sites are modded in, and finding them gives access to some amount of independents, often including multiple troop types.

[JonBrave]

By the way, how does the AI use its gold? I was wondering if increasing the income would help the AI, since I rarely if ever feel in need of money in SP.

If the AI earns enough money, it will actually buy itself an ice cream :)

Not sure what level you are playing it, and we've all claimed we can beat the AI anyway, but I would not say I am not "in need" of cash in SP. It is true that later on you seem to get more than you can want, but in the earlier stages I do not have the money to buy mages at every castle plus all troops each turn. If you truly have money to burn, don't forget that filling your homeland with temples never does any harm.

[Rookierookie]

I don't pretend to play optimally or even try to. It suits me perfectly fine that the income scale benefits me early and the AI late.

[Makinus]

If you're trying to boost it that way, taking poor scales to compensate for the increased money would be necessary to avoid negating the advantage to the AI. It will recruit even more chaff that way though.

Good idea.... but what would be considered "poor" scales in that situation? If i get Trumoil3 instead of Order3 would i really be handicapping myself or, in a game with plentiful money, it would be irrelevant?

[Knai]

If you're trying to boost it that way, taking poor scales to compensate for the increased money would be necessary to avoid negating the advantage to the AI. It will recruit even more chaff that way though.

Good idea.... but what would be considered "poor" scales in that situation? If i get Trumoil3 instead of Order3 would i really be handicapping myself or, in a game with plentiful money, it would be irrelevant?

Take the wrong temperature scales, don't use the extra points. The most blatant method is Heat 3 Niefleheim or EA Caelum, or Cold 3 Abyssia, but you can always get at least 3 away from ideal. Take Misfortune and Drain, don't use the extra points. So on and so forth.

[rdonj]

Let's say you take turmoil 3 and heat/cold 3. That's 33% less gold that you get. And if the AI has just order 3 and nothing negative, it's more like a 55% difference. With double money that's huge, and will make things a fair bit more challenging, especially in the early game.

[BewareTheBarnacleGoose]

Taking the wrong temperature also gives your units encumbrance from "severe heat/ Cold", I think it's +2 enc. Makes life especially difficult for nations that use high enc units/thugs.

[Rick L]

The AI can be terrifying at the beginning of a single player game, a patsy in mid game, and boring, why bother, in late game. To make the late game more challenging, play two strong units as human. When you get to the lat game, turn whichever is stronger over to the AI (the become computer controlled command in the options menu).

I just did this playing Helheim and Hinnom. I avoided all the "Hinnom eats everything" units by not taking any luck and managed to do okay against the AI. When I turned it over to the Ai, I had all the Lords of Civilization summoned. It made for a pretty challenging late game. Nethgul was the key to victory.

When playing the AI I also try not to give my commanders any gems, which does make it harder in late game as the AI does do that at least.

[Gandalf Parker]

In reading back thru the previous posts....
about giving the AI gold but the drawback is that it aids you.
Most of the nations have room for more capital magic sites. Some well chosen sites can add gold, gems, even special troops. Of course it would require a map and specifically placing the AI on the map so you know where to put the extra sites.

And if that wasnt enough, since you know the location you could always give it another province next door with more specifically chosen magic sites. And add a specifically chosen populace for recruiting and PD. Trolls maybe? Cave men?

Even more fun, if you give the AI two provinces apart from each other. Maybe at opposite ends of the map. Maybe two castles with selected sites. The first thing the AI does is consolidate its position so it would concentrate on fighting to connect the two sites. You can give the AI an early game direction and goal by doing that.

[Makinus]

I like the idea of giving the AI 2 separated starting provinces so it will try to connect the territories... is it possible to add a copy of the nation capitol sites to the 2nd province?

[Soyweiser]

about giving the AI gold but the drawback is that it aids you.
Most of the nations have room for more capital magic sites. Some well chosen sites can add gold, gems, even special troops.

There is one little problem with this. Giving them those kinds of sites aids the player A LOT when he finally conquers his first capital.

Beter would be to give the AI a starting commander that domsummons good troops, or a national spell that summons a monster that has no slots but has default equipment and stats that mimic popular SC loadouts and blesses. (Or give them slots and let the AI randomly dump even more powerful stuff into it).

[Gandalf Parker]

IIRC its treated like a cap site. It would give gold and gems but not troops.

[majama]

AI has always problem with supply for their big armies. They dont using magic items with supply bonus. AI army always has lower morale by supply shortage and diseases. Here is problem.

[Soyweiser]

IIRC its treated like a cap site. It would give gold and gems but not troops.

Yeah, but that gives anybody that conquers the cap site (Ea: the player) a huge bonus.

[Makinus]

Just to add a recent set of house rules i´m playing with that is generating some very fun and challenging games with the AI:

New house rules:
- do not construct any new castles - you only keep your starting castle and any conquered or generated by events.
- Always use max Drain scale.
- AIs at least in Mighty

Old Rules:
- Always use an Imprisioned pretender
- Only attack AIs if attacked first (just declaring war is not enough)
- Do not use battle spells that use gems
- Do not use global spells (using Dispel to bring down ones used by the AI is fine).


I´m not using anymore the house rule of only 1 researcher per lab, but using the max Drain scale and the castle limit i got a similar effect, and the limit on castles really make the game challenging...

[JonBrave]

I´m not using anymore the house rule of only 1 researcher per lab
I'm very glad to read this, as I've commented on it before :)

- Do not use battle spells that use gems
Of course it's matter of what you want to achieve in a game, but I did try this for a while and for me it took out a lot of fun.