has the AI been upgraded?

[Taqwus]

Scenario-based games frequently pose more of a single-player challenge, as (a) the author frequently has the tools to write a customized plan for the AI, and (b) single-player scenarios tend to be grossly unbalanced in terms of advantages granted to the AI(s) to the point that were the computer replaced by even a novice human opponent, defeat would be all but automatic. In addition, the human player frequently needs to fight the interface itself and (relevant in RTSes) the fact that his units tend to be utter imbeciles who need excruciatingly exact, explicit instructions every step of the way forces micromanagement.

Random-map TBS games have a harder time giving a good game to the single-player -- the player might actually have the time per turn (if not necessarily the motivation...) to micromanage, and random setups prevent exact scripting.

That said, some more configurable games have done somewhat decently without massive scenario-specific scripting. The tactical AI of the Combat Mission series, for instance, does at least try to avoid some large possible errors (such as firing too early and accomplishing nothing but revealing one's location, or sending units piecemeal) and employs some reasonably healthy habits (like sometimes towing a gun away after its location is known, so that it's gone before your artillery responds). It may even be capable of blocking your LOS with smoke rounds, IIRC. It's not a match for an experienced human player, particularly on offense, but it tends not to make one scream at its idiocy quite as much as does, say, an AoK AI which cheerfully stays put on its tiny island and waits until the Imperial Age to build a dock. I once played a random map AoK game in which one AI somehow managed to mine *zero* gold during the entire game -- and this was a game in which I'd taken a absurd (and were it multiplayer, surely a guaranteed-losing...) amount of time building economy and teching up before bothering to assemble an attack force. How? Heck if I know...

[NTJedi]

In my opinion:

Multiplayer games with humans has the most fun yet most people play the single player game of Dominions_2. The biggest improvement for single player games would be improving the AI. Most game developers just focus on getting the AI to attack/defend and expand... then when human players discover a pattern its easy to win.

The biggest improvement for multiplayer games would be having an option for maps to be randomly generated and/or a more user friendly map editor. (Hopefully the developers can glance over the map editors from Heroes_III and AgeofWonders:ShadowMagic for ideas)

Also I'm praying the arena death match gets improved where human players can consider it worthy enough to send someone. Currently the prize is not worth risking the life of a good or even average commander.

[PrinzMegaherz]

There is another disadvantage of single player dominions 2:
You do to many turns in a row. The fun thing with multiplayer is that you are eager to do the next turn because you have to wait for it. It makes every turn special.

[Agrajag]

PrinzMegaherz said:
There is another disadvantage of single player dominions 2:
You do to many turns in a row. The fun thing with multiplayer is that you are eager to do the next turn because you have to wait for it. It makes every turn special.

On the other hand, when you have to wait for a really long time for your turn, you forget what happened so far and what you have planned for the next turn.

[FarAway Pretender]

Mercurycs, I play Dom2 exclusively single-player (just don't have the time to make a commitment to play MP with folks I've never met, and am still trying to get friends to play it with me!), and I am finding, after about 9 months of an hour or two a day that I'm only finding the limits of the AI. Some of that came from reading stuff here where people talk about the limits of the AI, I probably never would have figured it out on my own.

The gameplay is different enough for each faction (and each Pretender design) that there's a significant learning curve, which is fun whether you're playing AI or not.

There are certain tactics in use among MP (primarily involving a "Rush" to get certain spells and/or magic items fast) that the AI doesn't rely on. But I'd play it yourself and see if any problems with AI performance slow you down.

Bottom line: Most good AI is not a function of programming, it's a function of game design. Game systems built to allow easy modeling and decisionmaking by a computer allow for good tactics. On the other hand, computers have a hard time modeling wacky and non-statistical combinations and permutations (e.g., "Am I better off casting Haste and Giant Strength on the same unit? What if it's a flyer?").

Just one example from another TBS: In Heroes of Might & Magic 3, the AI did a lousy job with spell selections. BUT, if I faced a Barbarian opponent (Barbarians don't use spells, but rely on combat troops that are more magic-resistant than others), the AI "got" a lot smarter.

One benefit of not allowing super-scripted combat is that the AI doesn't enjoy such a huge handicap in combat. I'm hoping that, in Dom 3, the designers might be able to tweak the Castle-building system in a way that subtly changes game play, but allows for more gradual castle building (and thus allows for more effective modeling of Castle-building decisions in the AI). At this point, that's just a dream of mine, though, and I don't think it's shared by that many.

[Boron]

You can do a few easy to do things to increase the AI's performance :

- Vh opponents
- Vh magic research
- Taking yourself a drainscale
- Mod supply to 5-10x as much as atm , then the ai will attack you with 500-1000 men armies where not 90% are starving
- Play against some strong but interesting modded nations
- Play yourself "weaker" nations if you enjoy that

If you have some time you can do some maps/mods on your own where the ai gets some boni .
Imo the ai is anyways quite ok , at least as you said compared to HoMM or even more to AoW which are very similiar games .