AI, nothing more

[eidolad]

Nothing is more important than a challenging AI in turn-based strategy gaming

not graphics
not storyline
not unit or research trees
not movie-quality cut scenes
not space-babes-in-armor (unless the game is *about* space babes in armor, then i change my mind)

I have high hopes that this game can be more than just a nice graphics package wrapped about a simple puzzle of "let me figure out how to exploit the AI, get bored, and move on to another game".

Note that I'm not looking for perfect, unbeatable AI...just "good" AI. The guys at XCom:EU did pretty well with their tactical unit AI:

http://www.pcgamer.com/2013/03/26/gdc-2 ... 0&ns_fee=0


...and finally, for the love of ER-PPCs, please do not opt out of developing good tactical AI in favor of a card-based combat system a la Endless Space. Note that I luv Endless Space otherwise.

[Martok]

Welcome to the forums, eidolad!


While I won't say AI is absolutely the most important factor -- I personally find atmosphere/immersion to be more critical to a game's overall enjoyment (including strategy games) -- I definitely agree it should have high priority. Here's hoping Horizon is able to give us computer opponents that can give us a run for our money!

...and finally, for the love of ER-PPCs

Also, kudos for the BattleTech/Mechwarrior reference.

[Ashbery76]

An A.I in my view has to play the game efficiently without cheating and role play the race in question.

[Anguille]

An A.I in my view has to play the game efficiently without cheating and role play the race in question.

I see it likewise.

[Saracen]

Nothing is more important than a challenging AI in turn-based strategy gaming

So functionality, stability, ease of use etc. would be detriment unless it has challenging AI? I disagree.

I would rather prefer an AI that contain a variety of different algorithms to give me a good opposition experience and can be defeated comfortably rather than just be a complete relentless pain the the backside to defeat.

But before all that, I want a game that has everything else in it's package. Doesn't necessarily have to have a story. But replayability is down to the quality across the board. I've played many games that have had great AI's but have just been awful elsewhere because they've spent too much on the AI... I have never come back to them... The games I seem to return to are those that just have average AI with a slight challenge, But the game excels everywhere else.

One game that has definitely got it right, although it's real time... is AI War: Fleet Command. It has that package, it has the gameplay and usability... and well... the AI appears very clever. The more you attack and harass it, the more you tick it off, and the more aggressive it gets. This adds an external layer of strategy where you have to defeat the AI by finding and destroying certain structures, and selecting your attacks correctly and carefully. Carelessness makes it harder.

However if you break the game down to its individual components. The AI is actually as basic as it gets, it's nothing special. In fact it's relatively easy to defeat given a carefully selected and correct set-up. However the game's design, how you set it up, and how you play changes all that. Certain scenarios can even make completing certain games impossible due to random generation. This allows for infinite replayability and varying challenge every time you play.

I would prefer a turn based game that works on the same model. Where it's the design of the whole package that affects the whole experience, and if it does the same for the AI, then brilliant.

[eidolad]

"An A.I in my view has to play the game efficiently without cheating and role play the race in question."

...that is well said and really should would be a better subject line than the OP.

Saracen provides good rebuttal/comments.

However I hope I didn't provide the impression that I want a Chessmaster 21K, having enough trouble beating the Chess program included with Windows 7 thank you very much.

My sole objective is to bump up *significantly* the priority of A.I. and make it the number 1 priority. On websters.com, the term "priority" is defined as "something given special attention" or "highest or higher in importance, rank, privilege, etc.: a priority task.

I'm not looking to starve the budget for the rest of the game elements in favor of A.I. Say, %50 of programming focus.

but I don't want to end up with another Civ5 or Warlock Master of the Arcane...beautiful, lots of variety and immersion...but stone dumb opponents. Means once parity is achieved...game over. Blech.

Not to be a fanboy about XCom:EU..but bringing an equal squad on the field against the tactical AI guarantees nothing. The "iron man" setting (no saves, death really sucks) adds hugely to this.

[Grogger]

I'm actually setting focus on some A.I. atm, so this topic is appropriately timed
Good discussion.

My current aim is to have a balance between personality and challenge. It's my thinking that it is intensely important for the opponents to make decisions appropriate to their personality, current situation, and relationships (increasing believability and immersion potential), rather than always choosing the BEST option for every situation (increasing challenge).

[Martok]

Cool, Grogger. Here's hoping you can pull it off (and good luck)!

[jorgen_cab]

I'm actually setting focus on some A.I. atm, so this topic is appropriately timed
Good discussion.

My current aim is to have a balance between personality and challenge. It's my thinking that it is intensely important for the opponents to make decisions appropriate to their personality, current situation, and relationships (increasing believability and immersion potential), rather than always choosing the BEST option for every situation (increasing challenge).

Sounds very reasonable to me. There are no way you can protect the AI against a meticulous min/max human player anyway...

Personally I like to role play in 4x games of possible, I just want the AI to be competent and understand the game mechanics enough that it is not a push over. My role playing will usually give me some handicap as well so that should even out the odds.

[CSS]

A.I is Crucial as most of us will be playing this solo I recommend A.I. Levels easy normal hard expert very hard suicide just like Moo2

[FirroSeranel]

Hmm... My feeling is that it's a lot less critical for AI to follow the rules, as it is for it to make sense from an immersion perspective.

AI War did that part right. What matters is the player's experience. In a game that, at least from what I've seen and read so far, the player is playing a newcomer to the galactic society (humans), with other, older races already well-established, I think to make the AI play by the same rules as the player would be a big mistake. Humans shouldn't be seen as an equal threat, at least not at first. An older race might try to befriend and uplift us, while another might try to enslave us... but why would they even bother trying to 'destroy' us early on?

What AI War did wrong was, IMO, well... everything else. It doesn't have "the total package". Its graphics suck, its "tactical" combat is a joke (thousands of ships? Please... I'm a human. I can't even micro-manage dozens of tanks in Command & Conquer, let alone 3,000 ships that seem to think that a giant circle is the optimal tactical formation...), and frankly, it isn't even in the same league or classification of games, as this seems to be trying to become. It describes itself as a puzzle game with tower defense elements, and that's exactly how it plays. It isn't a 4x at all.

I guess what I'm trying to say is, I would hope that given the premise of the game, the other races aren't starting with a single homeworld, at the same time as the player, like in MOO2. It should be more like AI war, in that in the beginning, the player's empire is so pitiful that its safety lies in simply not being a credible threat. I'd rather be ignored or patronized while I'm a low-tech empire just colonizing its first new systems, by vast empires that are busy with other things, than do the typical early game frantic land grab that plagues nearly every other strategy game ever. Expanding out into the galaxy should be about cautious exploration, careful diplomacy, and secretive research into technologies that can make my empire more competitive, not recklessly colonizing three dozen systems in the first fifty turns, which will then determine whether I win or lose. That's exactly the problem with the entire genre, by which I mean grand strategy, not just space-based.

Civ V's AI is far from the biggest issue it has. Every single game I've ever played of that, ends in one of two ways. Either I get wiped out early by a luckier civilization with a better starting setup... or I wind up getting intensely, mind-numbingly bored of steamrolling guys with spears or maybe muskets, with my stealth fighters, aircraft carriers, and nuclear weapons. While that sounds fun, it really, really isn't.

I would argue that rather than AI, which I think only needs to be good enough to make us -think- it knows what it's doing (immersion over trying to be better than a human at grand strategy, which at this point is impossible, and why pretty much all strategy game AIs cheat like mad), the primary focus should be on making the tech tree feel important and varied, while not actually completely breaking the late game.

Frankly if I have a single problem with the entire strategy genre, it's that for me at least, there's one strategy that is absolutely unbeatable: turtle and research. Even in MOO2, if you built a custom research race, and got a decent start, you could just sit ignored in a little tiny corner of the galaxy with 2-3 systems, research the majority of the tech tree much faster than other races that are focusing instead on expansion, war, etc... and then steamroll the entire galaxy with ease, as if the Vorlon declared war on the rest of the galaxy just for fun.

So no... I don't think the AI needs that much more attention. I think tactically it needs to present a challenge, but strategically it needs to simply not need to cheat, on account of not being in at all the same situation as the human player in the first place. Is it cheating when in Endless Space, the other empires have already claimed the entire galaxy and started to explore my constellation with wormhole tech, before turn 30? Yeah, I think so.

But is it 'cheating' for Earth to get their butts kicked by the Minbari in Babylon 5? Of course not! So just... don't start the other civs out on a single world at the beginning of the game. Make the challenge be to expand into a galaxy that's already well in advance, and to find a way to catch up, not to face off against yet another AI that can out micro-manage me, but is necessarily dumb as a stump when it comes to actual strategy, because it isn't really AI at all (because real AI doesn't exist yet, let alone in a video game).

[CyberMage]

While I won't say AI is absolutely the most important factor -- I personally find atmosphere/immersion to be more critical to a game's overall enjoyment (including strategy games) -- I definitely agree it should have high priority.

Of course, any strategy game which puts emphasis on the single-player experience rather than on PvP multiplayer must offer reasonably smart and competitive A.I. players ... but I wholeheartedly agree :

" I personally find atmosphere/immersion to be more critical to a game's overall enjoyment "

I would not want to have the cold and dry feeling that I was playing chess against a Grandmaster-level A.I. program -- superficially dressed-up in a science-fiction costume.

I mostly adopt a RPG-style perspective when playing strategy games : atmosphere and immersion then acquire more crucial value -- more than adopting rationally optimized builds and pathways in order to defeat purely rational A.I. programs.

[Lithari]

Well, the only Challenging AI I ever faced was on homeworld 2 and that blatantly cheated, hell it didn't even hide the fact it cheated....it knew where I was while cloaked, it just didn't shoot at me, it knew where I was from the get go, like I had location beacons on my ships.

Also, at the end of the mission, the AI calculated what I had, rounded it up to the worth of the ships and spawned a fleet that 100% countered my own fleet and sent it against me, while still building ships, going WAY over the ship limit.

If that is your idea of a challenging AI, then the game is already dead, cause I surely wouldn't play it.