Handicapping the AI

[AytchMan]

It's a fact of life that computer AI leaves a lot to be desired. To compensate for this, I'm always looking for ways to balance the game, that is, give myself a greater challenge. So, how do you guys handicap yourselves to even up the game?

My current method with the TW series is to levy a tax on my empire. I assess a certain percentage of total income (usually around 12-15%) and pay it by building and removing units from the game. By "removing", I mean that I build a fort in a remote region of the empire, march the boys there and leave them. I do not use them for ANY purpose. Their maintenance cost each turn pays the tax. For example, if total income is 20K and the tax rate is 10%, I'll assign sufficient units requiring a maintenance cost of 2000 kazoomis to Fort Apache. Since the tax is based on total income, it doesn't have to be very high to make a significant difference.

What other methods do you guys use?

[Fafnir]

Well, when I played as the Moors and Egyptians I would try to go to war with multiple enemies at once. For example, if I had two full stacks as the Moors, instead of devising a plan to go blitzkreig on Spain, I would use one stack to attack Spain and the other to attack Sicily. This makes sure that I usually don't have overwhelming strength when invading.

Another example, as Egypt, after I took Constantinople from the Byzantines I had over 6 full armies, so instead of finishing off the rest of the Empire with all armies, I sent three armies to different parts of Italy, 1 army to Hungary, 1 army to the Steppes, and one army to finish off the Byzantines in Greece. This makes resupplying your armies with elite troops complicated, making use of expensive but mediocre mercenary troops.

Oh, and I don't assault the first turn if my army is carrying siege equipment (like cannons and catapults), it doesn't allow the AI to formulate a defense.

[AytchMan]

Thanks, Fafnir. The one advantage I can see with the Fort Apache Tax is that it's self-contained. Once you move the units out of play, you simply run your empire with the forces available. You don't have to make (artificial) judgments about who to attack and where to assign stacks.

[italiano]

If the campaign AI is set up right the AI actually has a huge advantage over the player. The AI doesn't need to expand but that's probably why we play the game. If the AI is modded to form coalitions and act against aggression, it is much harder to expand. It's not unrealistic either.

[Darkarbiter]

Why not just play mods that do this for you? There are plenty of mods out there that give money to the AI or in general make the game just much harder.

[AytchMan]

Darkarbiter--

Excellent question. The answer is that, even with mods, the AI is still too weak. Obviously, a modded AI can be set up that's impossible to beat. But, for me, the key is that the game has to be within striking distance of realistic. Unfortunately, some AI designs and some mods substitute aggression for intelligent play. It gets tiresome after a while to simply confront ever-increasing numbers. Still, what's the best mod for AI play, iyo?

[Darkarbiter]

Darkarbiter--

Excellent question. The answer is that, even with mods, the AI is still too weak. Obviously, a modded AI can be set up that's impossible to beat. But, for me, the key is that the game has to be within striking distance of realistic. Unfortunately, some AI designs and some mods substitute aggression for intelligent play. It gets tiresome after a while to simply confront ever-increasing numbers. Still, what's the best mod for AI play, iyo?

Well point is it's no worse then what your doing at the moment.

[AytchMan]

darkarbiter--

You have me there, my friend.

[Fafnir]

Though, AytchMan's "taxes" can fight for him if his empire is going to the dogs, that's something useful.

[AytchMan]

italiano--

If the campaign AI is set up right the AI actually has a huge advantage over the player.

Depends on what you mean by "set up right". My idea of the best AI (which has yet to be written) is that it's "even", that is, it gives me a fight equivalent to a decent human opponent. Thus, to me, if the AI has a huge advantage, it's not set up right. But we may be talking about different things here.

The AI doesn't need to expand but that's probably why we play the game.

Here, again, I think the best AI would mimic a human player. Therefore, it would attempt to expand and win the game just as a human would.

If the AI is modded to form coalitions and act against aggression, it is much harder to expand. It's not unrealistic either.

Agreed. My only concern here is that the coalition-building not be rigidly formulaic. In M2, as with RTW, I think the designers simply chose to throw a set ratio of computer opponents at the human player (say, 150% of your empire's strength). The problem here is that it relentlessly penalizes success. Drop the anvil on your neighbor and your reward is that, two turns later, two more neighbors pile on. It seems so rigid to me that I've actually coined a term for it, Constant Pressure Syndrome. In the last few years, it seems that many game designers have thrown up their hands at the difficulty of writing good AI and settled for this second-rate compromise.

[italiano]

The advantage is in that the human player is trying to expand, while the goal of the AI may just be to preserve it's nation until the end of the game. The human wants to win the game while winning for the AI is irrelevant. There's no unfair advantage, just a difference in goals

Drop the anvil on your neighbor and your reward is that, two turns later, two more neighbors pile on. It seems so rigid to me that I've actually coined a term for it, Constant Pressure Syndrome.

It's fine providing there is reasonable logic behind it. Example, the AI forms some alliances to keep a potential threat in check, instead of something like AI forms alliances to gang up on human.

As far as the m2 vanilla process goes, the AI seeks alliances with nations that share a border with a potential enemy. From what I remember it's not set to a certain ratio, the AI is just given a value of how important it is to seek an alliance in that situation.

[AytchMan]

italiano--

On a set ratio of opposition: I have no hard evidence other than my own experience of playing the entire TW series a great deal. But I've seen many cases where a new enemy will pop up right after I've dealt a severe blow to a current enemy. Could this just be some smart alliance-building by the AI (or even random dumb luck)? Sure, but most of these attacks are awfully convenient. Often, the new enemy is an ally on good terms with a good reputation who just happens to change sides when I get ahead a bit. And, for what it's worth, no actual enemy alliance has been created; my ally simply turns on me. I've found it to be so predictable that I almost never have to declare war on anybody. If I just wait a couple of turns, somebody will do me the favor.

As further evidence, I'll note that there are many cases where I've beaten my only two neighbors to a pulp yet they refuse to surrender. If I gain a border with a new powerful neighbor, the new guy will immediately attack. Then, suddenly, the weak sisters are much more receptive to a ceasefire. I attribute this to the AI deciding that the ratio has now been met. Again, there are other possible explanations but it's very suspicious.

I have no problem with (and, indeed, welcome) intelligent alliance-building by the AI that, at times, targets me. The current design just seems a little too rigid, game in and game out and, thus, awfully suspicious. I'd like to hear one of the AI designers put my suspicions to rest.

[Evolution]

Aytchman must have some sort of Uni Degree in his English Hooah for him.


wow you guys have no idea. There is no AI (artificial intelligence) in this game.
they are all retards. they should just be known as A (artificial)



|MeMw|Obi-wan.
Has hated reading long forum posts since 1964.

[AytchMan]

Fafnir--

But I would consider that a cheat, though. The purpose of Fort Apache (and the tax) is to remove a portion of my strength from the game to balance it. Coupled with some mods that tweak the AI up a bit, it gets the job done. Plus the tax is easily adjustable from game to game, even between factions to balance the relative difficulty.

[AytchMan]

Obi-Wan--

Aytchman must have some sort of Uni Degree in his English Hooah for him.

Ha! It's actually my second language. I was raised by Armadillos.

[Evolution]

You were razed by armadillos!!!

you wouldn't happen to have a tatoo on your body saying 666 and for some strange reason you freak out when you go near a church?

[AytchMan]

I, uh, I'm not at liberty to say, Obi-Wan.

[homefry]

Handicapping would not be in my gameplan.

For me, the idea is to do the best I can with what I have. That doesn't mean to use any exploits such as leave dozens of merchants in a fort on a high-money-making resource. But if I'm fighting the enemy who doesn't have any missiles and my missile units are just tearing them apart while the enemy just stands there, then I won't charge in with my troops.

To make the game harder, send your diplomats everywhere and don't establish trade rights with anyone. Have your diplomats be bullies and request 100,000 florins or else you attack. Do this every turn with every faction. Sooner or later, you'll be fighting on every front and you will need every troop and florin at your disposal.


The other handicapping you could do is to have the AI manage your empire, but that's no fun. You could set up your building queue to build the buildings in order from left to right without any strategy like how the AI builds.

[AytchMan]

Handicapping would not be in my gameplan.

But doesn't this make the game a walkover? It does for me -- like most humanoids, I can tune the AI's carburetor with little trouble in a fair fight. Handicapping is necessary to make the game a challenge. Once in a great while, it's fun to trash everybody in sight but, for general play, that gets old fast. Thanks for the suggestions on handicaps, though.

[italiano]

AytchMan, the vanilla AI is set to gang up on the player as they become more powerful. It's not a set ratio as far as I'm aware, just a faction by faction thing. It isn't set up to be a fair campaign; CA is just using "ganging up" as a way to try to challenge the player. The AI is set to gang up on any powerful opposition as well as to act aggressively against the player.

The problem is if the AI is set up to act more realistically, war becomes much less frequent as the AI won't go to war against a superior enemy. The game then becomes boring for many players who prefer a more arcade style game, or at least something more action-oriented.