Question before buying!

[nordlys]

Originally Posted by Knai

That said, there are better strategy games if you only do single player. Battle for Wesnoth for example...

You can't be serious. Wesnoth is a cheap hack of game not even coming close to the 15-20 years old classics such as Master of Magic, Fantasy General, Crusade: Fight for Power or even Swords of Aragon (and I only mention turn-based, fantasy-themed ones). Other than being a poster boy for open source champions (most of whom somehow also happen to be anime fanboys), this game has as much value as it costs.

[Fantomen]

Battle for wesnoth is pretty good I think. Very simple and well designed IMO. I don't play it much myself anymore, but my son(12yrs) loves it. And so did I for a while.

It won't compare well to Dom3 or MoM because their values are very different. If I was to compare it to anything it would be warlords.

One of the wonderful things about Battle for wesnoth is that the community provides a vast set of languages, including swedish for example. So it's accessible for kids who don't speak english so well yet.

[Ink]

I'd take BoW over Fantasy General, Crusade: Fight for Power, and even Swords of Aragon any day.

MoM, though... nobody messes with MoM

[Gandalf Parker]

In some ways this conversation has pointed out that the GOOD AI games tend to be ONLY for solo play. Peoples lists of good AI tend toward solo games. Its very difficult to mix MP and Solo-against-AI because both tend to take up alot of the code, and cpu cycles, and memory. Many cuts happen for the sake of processing time. (I did push to have cut code put aside for a server switch since server games could ignore hosting time for the sake of better AI, but it didnt get put in)

Another point that Im not sure has been mentioned or not, is that we should note that we are talking about Vanilla AI. The average AI that comes with any game only has to be good enough to cover the demo, the tutorial, and (if you re lucky) the early learning phase of the game. Thats why I feel Dom3 exceeds the average. I think Dom3 beats that (unless of course you run to the forum for a list of ways to beat the AI).

BUT it should be mentioned for those who are interested in Solo play, or playing larger games with more nations but just a couple of your friends as the human players, there are options for going beyond the vanilla AI that comes with Dom3. There are game settings, maps, mods, etc that can boost the AI beyond just setting it to Impossible.

[rdonj]

Originally Posted by Gandalf Parker

Another point that Im not sure has been mentioned or not, is that we should note that we are talking about Vanilla AI. The average AI that comes with any game only has to be good enough to cover the demo, the tutorial, and (if you re lucky) the early learning phase of the game. Thats why I feel Dom3 exceeds the average. I think Dom3 beats that (unless of course you run to the forum for a list of ways to beat the AI).

I disagree with this sentiment. The average AI that comes with any game only *is* good enough to cover demo, tutorial, and early learning phases. But they *should* be good enough to provide a challenge on the highest difficulties. I have a problem beyond that though where I want AIs to not do that in the dumbest way possible. For example, I can't be the only one who's played age of empires 3 and been annoyed with the way the AI will build nothing but cannons at the highest level. I've had to kill armies of literally 20+ cannons in some games... it's really stupid and doesn't give me anything like the experience I'm playing to have. Yes, it makes the AI hard, but the way the AI plays should also be fun and engaging. I have the same problem with pretty much all RTS games, really :/ The AIs get hard by playing in un-fun ways. Anyway, rambling aside, the AI should be hard enough to make you have to work for victory, at least on the hardest settings.

I do think that dominions succeeds at AI beyond what most games do, in that the AI can still be challenging under the right conditions. But a lot more could be done to make the AI more challenging to play against than current mods do. The problem is that I really can't see an easy way to mod a universal "fix" to make the AI better. The best fixes would all require you to make a new mod for every game you play. Unless we start developing mod nations for the sole purpose of being powerful AI combatants (which is not entirely the case at this point).

[Gandalf Parker]

True enough. I should have said that the general consideration of game devs seems to be that it only needs to cover demos, tutorial, and maybe early learning. Obviously as a solo player I myself feel that better AI is a good thing.

But Ive also hung out in AI forums for as long a the net has existed. And Ive seen the discussions. Part of the problem is that the farther into each playing of the game you get, the more branchings the AI has to handle in its logic, and the more it fails.

Dom3 AI is a pretty good example. Its total randomness was a cheap answer to providing maximum replayability. It tries to avoid predictability and MAYBE provide some surprises altho usually it amounts to stupidity. Even with the total aggregate of every AI boost we have available to us, the best I can get is early and mid-game playability. Hoping for possibly a quick win since the AI can never handle a real late-game fight unless its GM'd. The depth of the game we love works against us in hoping for an AI to handle it.

[13lackGu4rd]

if armies of 20 cannons are killing you in AoE3 than you're doing something very wrong :P speaking of AoE3, the AI there is pretty good compared to other AIs. it's alright in medium difficulty(alright as in what medium should be) and starts to get a bit tough on hard, and pretty tough on very hard. still, it's the AI so there are obviously ways to exploit it and destroy him, such as using mass Hussars or whatever those sword armed light cavalries are called, to destroy cannon heavy armies with no pike protection. but still, compared to other games in general and specifically RTS games, Microsoft actually did something right in AoE3, something I can't say about most of their other programs.

Dominions' AI is nothing like AoE3's, it's pathetic. sure, it amasses armies of 300+ men, but it has piss poor magic support, and generally a single army killing spell, from Earthquake to Rain of Stones to Flame Storm, etc, completely annihilates those massive armies.

the main problem is that Dominions is so complex and with such a variety of both units and spells, that no AI can actually handle it properly. even those super genius chess AIs won't do well in Dominions due to the sheer amount of options and variables upon which said options are evaluated on. in chess you have a lot less moves, and the success/failure of said move is a lot easier to calculate. in Dominions there are a ton more options, and the results said actions lead to(mainly on the strategic map but also during battles) are very hard if not outright impossible to predict.

with that being said, there are still a lot of things that *can* be improved with Dominions' AI, such as combat scripting(or well, lack of, hence automatic behavior), actually casting/dispelling globals, actually using rituals, site search, not stack a ton of useless PD, etc.

the main question for the developers is whether all the work that needs to be done to improve the AI will be worth it. is there enough demand for a better AI in Dominions and in other TBS in general? personally I don't think so, as no AI will reach the level of multiplayer, as I've explained above, so multiplayer will still remain the favored options by seasoned players. those who love single player and endless hours of AI annihilation will continue to do so with or without an improved AI. thus I don't think the AI will get much better than it already is, which is unfortunate. the only field which I think might(or at the very least should) actually see some work is the battlefield spell choice, as that affects both single player and multiplayer(when you forgot to script mages, and more commonly when they run out of their 5 turn script).

[TheConway]

Epic bump much?

Anyways, Blackguard more or less hit the nail on the head.

@rdonj: Good AI nations? Try Amos mods

[Gandalf Parker]

The AI is one long sequential run. To improve it would take many many IF statements.

To fix it to a more updated tree-structured AI would take a major rewrite of the code which is not going to happen.

So we are left with what is at our disposal. We have found game settings (such as resources and indepts) that improve the AI. We have found MAP commands which improve the AI. And we have MODDING which can improve the AI. The maximum of all of these tends to me more AI than even the solo lovers can wear out.

Its still not equal to a Human player in an MP game that goes all the way to EndGame. But for a thread that started out asking if the AI was adequate to Solo play, it is an answer.
IMHO of course.

[rdonj]

I'm going to be doing a few quotes here...

Originally Posted by Gandalf Parker

Dom3 AI is a pretty good example. Its total randomness was a cheap answer to providing maximum replayability. It tries to avoid predictability and MAYBE provide some surprises altho usually it amounts to stupidity. Even with the total aggregate of every AI boost we have available to us, the best I can get is early and mid-game playability. Hoping for possibly a quick win since the AI can never handle a real late-game fight unless its GM'd. The depth of the game we love works against us in hoping for an AI to handle it.

I am pretty fond of the dom3 AI. It handles early game and midgame at least to a reasonable extent, even if it stands no chance in late game. So I actually agree with you there. The way the AI works is good for making it challenging early on... if only the AI knew how to use powerful units.

if armies of 20 cannons are killing you in AoE3 than you're doing something very wrong :P speaking of AoE3, the AI there is pretty good compared to other AIs. it's alright in medium difficulty(alright as in what medium should be) and starts to get a bit tough on hard, and pretty tough on very hard. still, it's the AI so there are obviously ways to exploit it and destroy him, such as using mass Hussars or whatever those sword armed light cavalries are called, to destroy cannon heavy armies with no pike protection. but still, compared to other games in general and specifically RTS games, Microsoft actually did something right in AoE3, something I can't say about most of their other programs.

No, I was killing them :P I just hated it because it was stupid. 20 cannons vs massed cavalry doesn't make for very fun gameplay. Also I prefer long games, so that sort of thing happens to me a lot :P. I do agree that AoE 3 had a pretty effective AI, I just wish that it could be effective without resorting to silliness like that. I'd like to see the AI make use of more different types of units and create real armies to attack you with.

the main problem is that Dominions is so complex and with such a variety of both units and spells, that no AI can actually handle it properly. even those super genius chess AIs won't do well in Dominions due to the sheer amount of options and variables upon which said options are evaluated on. in chess you have a lot less moves, and the success/failure of said move is a lot easier to calculate. in Dominions there are a ton more options, and the results said actions lead to(mainly on the strategic map but also during battles) are very hard if not outright impossible to predict.

with that being said, there are still a lot of things that *can* be improved with Dominions' AI, such as combat scripting(or well, lack of, hence automatic behavior), actually casting/dispelling globals, actually using rituals, site search, not stack a ton of useless PD, etc.

There's a really, really easy way to make the AI 100x more effective. Take options away from it. The problem is just that you'd have to have either a different mod for every nation you want to play, or you'd have to create a mod nation/nations for the sole purpose of having a challenge in SP. The AI will never be intelligent, so the best way around that for us is to remove potential idiocy.

I'm always surprised though when I hear people say that the AI doesn't cast spells/globals etc, because I've had them do that plenty of times with me. I've even had global slot battles with the AI before. I once had to fight an AI marignon army in SP that summoned summer lions and fire snakes, and spammed prison of fire at me. Sometimes the AI can be surprisingly effective.


The Conway - I know Amos has some pretty unbalanced mods out there... I'd say I've played about half of them! But if I wanted a truly effective AI, I'd go a bit more in the direction of warhammer chaos. A lot of Amos' nations have weak national units and... weird commanders, for one thing. One of the most important things to do is ensure that the AI has effective troops and commanders available to it. Kind of like the hellgate mod, but even more so