Just why is the AI so bad!!??

ok, maybe that question is a little misleading as I think the battle AI - although obviously prone to some bizarre actions - is decent and can, on expert, make for some interesting confrontations but the strat map AI, oh dear.

I've been playing MTW solidly since it came out, although that is largely due to Wes and his excellent MedMod, which I installed after just two vanilla MTW games and have never looked back, however, even with MedMod I'm still looking for a real life and death campaign.

For ages I was playing on Early/Expert (always on GA) but after getting to that point where you just know you're gonna wipe the floor with the enemy all too soon I switched to playing almost exclusively Late era games. I've played many many times as the English (ok easy faction, but it's just my personal bias - I am english) hopeing for a real Anglo/French bust-up but it never really seems to happen. I've also played as the Spanish, and most recently the Sicillians.

In all these games I have modded my start pos so that I start with only 4 territories - in the case of the english - or only 2 - when playing the sicillians. Even though this meant that for the first 10 turns or so I was clinging on for dear life in Sicily and Malta, once I'd repelled the invaders and had time to settle I quickly built up those two provinces to the point where I was churning out gold weapon/gold armour knights that were capable of crushing anything that I chose to send them against.

In many games I mod the french so that they start with more land, more units and more money, and although they often start with uber stacks within a few decades all seems to have dwindled. Just what does the AI do with it's time and money? While I'm beavering away at making better units just what is it doing, because I almost always end up with the best turned out troops on the battlefield.

Often, if Im attacked by the AI and I lose the province, I will muster men in all my other provinces and swiftly counterattack. In many of my campaigns Ive faced vast AI empires but the AI has never done this to me. It lets me chip away at its empire, province by province, until I am the larger of the two powers. Also, multiple attacks into multiple territories is something I have never witnessed the AI do. And let's not even talk about the AI's use of ships!

While I understand the complexities of coding acceptable battle AI, I am somewhat bemused as to why the strat map AI is so underwhelming.

So to my question - how can I make this game as challenging as it can poissibly be. I'm not really keen on limiting myself to a set of self imposed 'rules' as some people do (although I never use assassins), so any other suggestions/modifications I can make would be most welcome?

I'm about to go for HRE/Expert Late in the hope of some real back to the wall life/death stuff.

oh - in my years on MTW playing I've still never finished a game! Just can't be bothered with the mop up once I know I'm the uber power - to me the challenge IS the game.

Well, think of it this way: even super-computers powered with the latest chess software are unable to consistently beat the best humans in chess... And, that's a turn based strategy game less complex (in the number of possible outcomes) than the totalwar series. Imagine the calculations involved for the AI to foresee all possible player moves in MTW and then come up with it's own optimal response to all those "maybe-situations"? It would be overwhelming, and, in the case of the tactical map, probably prohibitive since the simulation is real time rather than turn-based. Hence, a really challenging battle AI is probably a utopic dream still for some time to come...

Nonetheless, the strategic AI could be improved in some simple ways too (and this should help the tactical AI as well, providing it with better troops). It is actually easy to observe the current AI behavior by using "-ian" switch and then rotating through the different factions to see how the AI is doing. When I did that, I found that the AI is mostly running their economy to the ground (and optimizing economic decisions should not bet that hard of a programming job). Even if I leave a faction in considerable surplus and a developed trade fleet, in a couple decades, the AI has run the treasury down, bundled all trade ships together and sent them off to some far-away sea to lose money... It would have been OK, if it had achieved some territorial gains in the trade. But, no... Once, turned over to the AI, the faction actually managed to lose it's strategic fully garrisoned bottleneck provinces (Navarre and Aragon) and let crusaders take Leon by a naval attack, since the coastal fleets were sent away on a joy-ride around the Baltic... I am talking about Almohads in this particular case. Very disappointing, very!

As far as I know, you won't be able to mod the AI to your liking. If I understood well, the only mod you've been playing with is WesMod. In case I'm not wrong, I suggest you give a try to mods like BKB's, it features many more factions so it's way harder to have a "rush" tactic at the beginning. The game is very challenging, especially in the East with many new factions (Kievians, Armenians Il-Kahnate...) and buildings.
In my opinion, WesMod is based on a very good concept (homeland) but in the end, once the AI factions lose their homelands, they only produce junk units.

If you want a "brand new" game you can only try HTW (Hellenic), it's really worth playing.

Slaists - I am only really talking about the strat map AI. I do think the AI does a pretty good job in Battle Mode, especially when it has to juggle all of the factors that you've described.

No, my issue is with the strat map decision making portion of the game. Here, time is not an issue, and essentially it's a numbers game.

I think part of the problem is the "fine line" balance needed. If they go too far with the AI, it would produce perfect optimum builds and make exactly the best choices in every situation, and would be very hard to beat, and so a frustrating experience for the player. However, I don't see why this couldn't be done on something like Expert, or maybe one higher than Expert.

In RTW, there's a seperate difficulty level for the strat map and for the tactical battles. Perhaps they have addressed this issue in that game-I've only just got it and so can't say for sure. I've not played MedMod myself but I'm told it features better AI build choices in terms of not churning out rubbish units and so on. But you said you'd played that and still didn't find it a challenge.

One idea is to build up a powerful empire with large armies and money. And then switch to the worst faction you can find (using the -ian thing) and try and beat your old empire with this rubbish little one! THere you have it, one pre-made powerful opponent for you to try and take on (your own creation!). I've heard some people have had fun playing like that.

HicRic - To be fair, I have had a challenge playing MTW, but frustratingly it never seems to last.

I see your point about it being a fine line although I reckon you could code in simple rules which would give each faction some individulaity and make it look like they were not robots making perfect decisions whilst still improving things...eg have the AI make perfect economic decisions but impose penalties on the income it has to work with to simulate wastage, bad management etc.

Building up better enemies via the -ian cheat or pre modding them via the data files doesn't really seem to help either.

I guess I could get 10 mates or so to come round, and using the -ian cheat have them control the economies of all enemy factions, while I keep my eyes closed.

There is one very simple change they could make to the AI that might make things better. I saw it as an option in computer Risk (must have been years ago), and it was simply that if you took a territory off an AI player it remembered that was "its" land and basically threw the kitchen sink at getting it back. (And for MTW also code it so that there is none of this "The XXXX decide they cannot win the battle" nonsense.)

It doesn't sound like much but believe me you could beat the AI with your eyes shut with that option turned off, but with it turned on it was almost impossible to win.

I think its right that the AI's economy is half the problem, to be honest although its a total cheat I wouldn't mind a crude system that just gives the AI the average of what a human player would get from a given province in a given year regardless of what it does with its trade ships or whatever. Also a complete ban on the AI building peasants would be no bad thing.

Some of the AI problems can be solved, also the trade "cheat" that the player can use by building a huge trade empire. In my XL mod, I reduced the trade value of goods, increased farm income by 30%, changed AI building setup, increased unit morale and most of all removed peasants. The AI now builds a lot more HQ units (large stacks!), battles are harder and the player dosn't have the trade advantage. My mod also adds more units and factions to the game. Try it out. ~;)

The problem w/ an AI - and I don't think there will be a way around this until HAL comes along - is that it can't learn, it will only ever know what was tediously programmed into it. I mean, I made a lot of the same dumb mistakes that the AI makes during my first campaign. I thought about it, read the manuals and checked the boards, and made a lot few mistakes the second time. The AI however didn't learn a damn thing and never will.

My imperfect solution has been careful use of the -ian switch to give the AI a hand here and there, and to use a few self-imposed rules to keep me from mauling the AI too badly and ruining all the fun.

Quote:

---

Originally Posted by Procrustes

The problem w/ an AI - and I don't think there will be a way around this until HAL comes along - is that it can't learn, it will only ever know what was tediously programmed into it. I mean, I made a lot of the same dumb mistakes that the AI makes during my first campaign. I thought about it, read the manuals and checked the boards, and made a lot few mistakes the second time. The AI however didn't learn a damn thing and never will.

My imperfect solution has been careful use of the -ian switch to give the AI a hand here and there, and to use a few self-imposed rules to keep me from mauling the AI too badly and ruining all the fun.

---

The day someone makes a complex AI that can learn and improve in complex areas such as a Total War game will be a fine day indeed-it will provide and endless challenge as we constantly must find new ways to defeat an opponent who learns and counters our tactics.

Until then-wait for an MP campaign in a Total War game. Human players on all factions would make things difficult-and I doubt an AI will come along any time soon that would be better than playing real, skilled people. ~;) (Of course, there are difficulties in making such a thing, like the time it would take to play through and so on.)

Quote:

---

Originally Posted by Procrustes

The problem w/ an AI - and I don't think there will be a way around this until HAL comes along - is that it can't learn, it will only ever know what was tediously programmed into it.

---

I can't begin to imagine how it's possible to boil down something as complex as strategy into a series of binary yes/no decisions, so I don't envy the task of the AI programmers one bit. (oops, unintentional pun there)

However, there's plenty of room on most people's HDDs in which to create additional data files so I don't see any major obstacle to a system whereby the AI can 'learn', after a fashion. For each AI faction, it records the strategic decisions they made in a campaign and the ultimate consequence of what it did in each province it owned, economic improvements added, buildings constructed, the troops it built, the moves it made with them and so on.

If it never lost a particular province throughout a campaign, it knows it can follow that same event path for that province in a subsequent campaign, 'expecting' some degree of success. Where a province ended up being lost, or the faction got totally wiped out the sequence of decisions leading to that are on record and next time, this can serve to inhibit it from going too far down that same path a second time.

Better still, a record of what the player did at any given time - especially useful where the player ultimately won - since this behaviour can be copied when you play as a different faction next time. Ultimately, you'll end up playing against 'yourself', in a way... ~;) On the other hand, the AI resolves all its battles (with other AI) by autocalc, where numbers seem to count for more than quality of troops and skill in their use. No doubt the fate of the AI trying to exactly copy what you did (including attacking large forces with small ones) will rapidly diverge from your previous success and the rest of the event tree becomes inapplicable, leaving it on its own, perhaps carving out an alternate route to ultimate failure...

With or without the ability to learn, one thing which pure applied logic can't do, which people can, is those 'what the heck' decisions about things to build or whether or not to launch an attack, say. They may have more to do with emotion (revenge for example) than reason but sometimes can still pay off. No AI would be able to anticipate such actions, nor emulate them.

you want tough, go play multiplayer.

Quote:

---

Originally Posted by solypsist

you want tough, go play multiplayer.

---

But noodle said it was the strategy side that he was interested in.

I generally agree, keep in mind though that if the campmap ai was really good, some factions would be unplayable, and there would be no other strategy except the turn 1 blitz.
I have come to prefer eastern factions, especially high, because of the constrained positions and pressure from the horde. Often, a big power will emerge in the west or in Africa while you fight to survive. Try the Turks or Russians.