Willing to pay for better AI

Well the reason AI development has been so slow is because developers plan for their future and it's easier to show and sell what fancy graphics you can do compared to how great an AI was created. Even gaming publishers don't ask gaming companies AI questions... they just expect them to make it work. As a result we get an AI which works, yet unfortunately these AI opponents are either below average, bad, very bad or terrible.

So the question is where is the future of AI gaming... it will continue to move slow because many developers/publishers currently believe 'pretty graphics' will mean better game sales.
{as written within another forum}
I believe the next GREAT AI in a computer game will use random multiple personalities which can be disabled in case one is bugged or weak. The ability to allow gamers to add their own AI via scripts would also have to be included. Now Gal_Civ_2 comes close to providing this type of AI opponent.
The next step would be developing a learning AI which records strategies/methods used by the human and then merges it with its own strategy or even better it develops a formula to use which beats the human strategy... this last step is far beyond what developers currently have time to accomplish and it would take super genius developers to make this AI opponent a reality.

question.. are beating computers on suicidal??? i mean at that level they get like extreme bonuses and stuff and im assuming its hard.. ive been working my way up and im at the masochistic but have you beaten suicidal easily?

Galciv2 AI adready truely challenges. Of course after a number of games your brain will figure out how to beat any AI. If your PC had the ability of your brain which is wire and rewire itself we would see an AI that can adabt. AI is so predictable because a PC is nothing but a fancy calculator.

I don't buy into the assumption that most 4x strategy games fall off in popularity because of a lack of challenge from the AI. In fact, I doubt seriously that most ever beat the highest difficulty settings. Even strategy gamers run the gamut from casual to fanatic. Most are likely on the former end of that spectrum; they get bored of a game's mechanics and yes, look, moreso than from a lack of challenge.

So, in other words, developers put minimal effort into AI, because AI brings little return from the market, besides being so ridiculously difficult to do.

And you think most would be willing to wait minutes or hours between turns in a game that is expected to entertain? You think there are enough such people out there to support this type of developer interest?

Some people in this forum need to get their head out of the sand....

The real problem with the AI isn't really the high-level thinking. It's the low level stuff.

It isn't the fact that it goes to war against target X; it's usually a good move for it to do so. It's the way that it does so.

You can give the AI hardcore ships that should be able to take out its adversary easily. But it can't; it won't. It doesn't understand how to make attacks or to defend correctly. Now part of that is the fact that it's playing the game as the game was intended, not the game as it actually was written in 1.1. It thinks that defending a world means building an OFM and stationing ships in orbit, because that's what the designers intended to be the reasonable way to defend. It isn't.

Another problem of the AI is that it's just too flexible in its tactical systems. Take planet building, for example. It tries to make it's planets everything, and it will override old decisions when things change. If a planet is even slightly starting to gain outside influence, the AI reacts to it by building embassies, even if the world is fairly negligable. A smart player knows better.

I would definitely have been willing to pay for the level of continuing game support that has been going on here. In fact I waited to 1.1 came out to buy the game, so they can tell from the numbers I valued the updates.

This sort of thing doesn't usually work by directly collecting the money from the power gamers. I think what happens is great features build buzz in the power gamers that helps the game sell to normals. Like all those motorcycle jumps off fountains and stuff in Grand Theft Auto: Vice City. Played it for like 60 hours before I even discovered those, most probably never saw it, but having them in there is how it got to be such a famous game. Well, that and the hookers.

Uh..i would say it was entirely the hookers. And I doubt very seriously that powergamers have that much influence over the buying habits of many, especially when you are talking about the type who tend to hang out on message boards.

The broader 'hardcore', i.e. reviewers on gaming sites, tend to mark a good A.I. by its ability to appear intelligent at first glance, and give a challenge to most players in its semi-hardcore readership. That is what is more likely to drive sales. How do most casual gamers distinguish between a 'very good a.i.' versus and a.i. 'that is so good, it challenges the hardcore.' If anything, the latter is more likely to drive players off.

With 1.2 and the new combat system building OFM and defending your planets wont be bad for the AI, If they could lern to defend thier starrbases we would be cooking with gas.

I agree. The AI needs improvement - and it's getting it - but people who constantly, easily beat the game are the minority, and frankly they're exploiting their way through. That's not bad - I'm not going to complain about people playing the game - but I don't see why the devs should take AI development in a totally different direction, just to satisfy the powergamers. I think the continuing evolution of the AI is fine.

I just got the game about 3 weeks ago and just now entered the medaverse......and i think they developed a real solid game here..turned my self into a spokes men a little bit telling friends to go out and grab it.....but i think it would be a great challenge if the A.I. was just a little bit more tactical....a 2 ship fleet engageing a 9 ship one wont stand a chance..

Also i think it would be good if the A.I. concidered a little bit before it declared war...(dont know if thats real possible ) but in my current game one declared war on me and because of my alliances was at war with 5 other races.....

AllowCPUIntensiveAlgorithms=1

Setting this line (in the Prefs.ini) to "1" might help with the AI, but it will take more CPU, so more time for turns to complete.

Probably shouldn't do it on a slow machine.

It's about half way down, in the GalaxySetup section.

...but I don't see why the devs should take AI development in a totally different direction, just to satisfy the powergamers.

Isn't that like saying, yes chess is a tough game, but this chess program, though easy to beat, seems okay to me (I can win by doing silly things like moving all my pawns up and the program is baffled by that). Why are they going to make it harder just for those who are good chess players?

Well, to make the game better? I also don't understand what you mean by "totally different direction." I thought the direction was to make AI improvements. I also don't understand "powergamer" as you use it. I don't see this as a powergamer game, but I do see that the AII becomes predictable, therefore eventually easy to beat. You want to see a powergamer game I suggest WoW or the other online games. Those are powergamer games. This is a stroll in the park, not that a stroll in the park is a bad thing.

You eat white bread, don't you? =D I tell you, hemp bread or sprout bread. That white bread has nothing in it but empty calories. But my nephew still eats white bread because the better breads are tougher to chew. Now "that's" lazy.

I don't "get" your post. I don't understand what you're trying to say, unless my first impression is correct and you're stomping your feet because you won't be able to move all your pawns up anymore and win. I understand that some people like the embassy way to win. That's great. I never tell a person what hand to use to wipe their behind. I don't expect them to tell me. I'm not at all suggesting that they ways to win be changed. But if I select a conquest victory I expect the AI to come and "conquest" me

Of course this is a difficult thing to do. That's a hard program to write. And there are the simple things that slip by, like you can no longer build the Galactic Stock Exchange because in the Tech tree.xml the Culture ID/ Display Name (I forget which is used) = "Advanced Trade" but in PlanetImprovements.xml the
"AdvanvedTrade" doesn't have a space in it . How hard is that to change? White bread, white bread. Eat white bread. Easier to chew. Wonder Bread. We used to make spitballs out of it as kids.

So if it's taking the developement team a long time to put a space in a piece of code (you can do it yourself actually) what does that tell you? Frankly I'm not sure. Maybe they can't read this or other posts about it because there are so many posts. I know they want to succeed. There are more games to create, more money to be made if they succeed here. I have no problem with that. What if the AI is easier to beat because some races are going for an achievement they can never get, like, say, Galactic Stock Exchange? I suggest fixing the spaces first. Make sure the game works like it should right now, codewise anyway, then see what the AI can or cannot do.

But to the first poster, yes, I'm willing to plop down the extra bucks for "our" kind of AI--and that will be a tough thing to create--and I will plop down the bucks for multiplayer, all upfront too. The problem is that my bucks may not be enough bucks. And I understand that. Besides, I think Frogboy is secretly, slowly but surely, it takes time, working on his pet project and making improvements to it. Will it be Gal Civ 3 or a totally different game? I don't know. But he did make a reasonable game here and he knows it, despite my cleverly concealed whinning.

I agree with SIDDII
Nothing more to say.

Citizen Corinthian, are you to a point where you can beat GalCiv2 everytiime on the highest level? What about scores? Alot of times, I just play games like this on a normal to just above normal level and I will try to beat my score and do better each game, even if I win most of the time. That said, I find this game to be extremely challenging, especially if you do this with the scoring like I do.

No, the point of the game is not to be 'tough' or to satisfy powergamers. Rather, it is to satisfy as many gamers as possible within the genre market its going for. That's a big difference, and should be easy to understand.

Most people saying the game is too easy are not at maximum difficulty and are using settings they use to win and then they say its too easy. If you feel the game is too easy then set it to suicidal and set all other races as a team at war with you. If you can beat that then you can say its too easy. You can easily make this game easy to win or impossible to win but you have to find whats right for you. If you find the game too easy then set it to a harder setting because your not at it if it's easy.

To the OP: The first expansion won't be multiplayer, they're working on new diplomacy things, new espionage, and a new Drengin campaign. Oh, and the AI to use Terror Stars. They aren't giving up on it yet

Galciv2 AI adready truely challenges. Of course after a number of games your brain will figure out how to beat any AI. If your PC had the ability of your brain which is wire and rewire itself we would see an AI that can adabt. AI is so predictable because a PC is nothing but a fancy calculator.

Wrong, if a neural net is used, the AI can adapt itself in-between games, and learn what you're doing. But galciv2 doesn't do this, the AI adapts, but only during the game.

And you think most would be willing to wait minutes or hours between turns in a game that is expected to entertain? You think there are enough such people out there to support this type of developer interest? Some people in this forum need to get their head out of the sand....

I'm fairly sure they do believe that, and they'd be right. A lot of people have CPU's so powerful it won't make a noticable difference, and quite a lot are happy to wait for a game that challenges them, the same way people would wait for another player to take his turn in multiplayer. A good AI is like multiplayer, without the whiners, disconnects, cheap tactics, and other things that turn off a lot of gamers.

To those that say it's too easy even on suicidal, you have to ask what settings are they on? Frogboy himself said yesterday that playing on something like gigantic map, tight clusters is like saying "I want the AI to be at it's absolute worst, since it can't use a human tactic like turtling in as efficiently as a human can". Settings like that give the human player a close set of planets (in a game yesterday I used that setting and got a group of planets which had five +14 planets in the radius of 3 econ and one military base. One was a class 33 with 100 bil pop, my econ capital, and one was my homeworld, 14, manu capital).

Go set up a medium map, scattered stars, slow tech, common planets, and rare anomalies. This is, IIRC, the absolute best AI available, since it's using it's number crunching ability and the human gets no advantage, other than pure skill.

Oh, and every time you use ctrl+n to get a decent set-up and complain about how the AI couldn't beat you, an angel dies. Painfully.

Most people saying the game is too easy are not at maximum difficulty and are using settings they use to win and then they say its too easy. If you feel the game is too easy then set it to suicidal and set all other races as a team at war with you. If you can beat that then you can say its too easy. You can easily make this game easy to win or impossible to win but you have to find whats right for you. If you find the game too easy then set it to a harder setting because your not at it if it's easy.

This to all that harbor similar thoughts...

It would take them a minute or two to add an even higher difficulty setting, or at least to make the current ones be tougher. Change a multiplier here, one multiplier there, and presto. Imagine if they doubled all multipliers the AI gets in the next patch, would you then say: Wow cool, they made the game even tougher! I would think not. It'd be a very cheap thing to do. And though it is acceptable and desirable to have the possibility of unbalancing the game in favor of the AI, that can only go so far before it becomes just a gimmick at best. That's why having an "impossible" or whatever difficulty setting can't counter the argument for having an AI with sound fundamentals.

This to all that harbor similar thoughts... It would take them a minute or two to add an even higher difficulty setting, or at least to make the current ones be tougher. Change a multiplier here, one multiplier there, and presto. Imagine if they doubled all multipliers the AI gets in the next patch, would you then say: Wow cool, they made the game even tougher! I would think not. It'd be a very cheap thing to do. And though it is acceptable and desirable to have the possibility of unbalancing the game in favor of the AI, that can only go so far before it becomes just a gimmick at best. That's why having an "impossible" or whatever difficulty setting can't counter the argument for having an AI with sound fundamentals.

He meant that some settings (like, as I said, gigantic + tight clusters) give the AI a disadvantage because the human player can use specialised tactics while the AI has to use an adaptation of what it normally does. My playstyle wildly varies on every setting, the AI doesn't have the luxury of this, since the code for making every single AI act differently for every single map would be immense and illogical. It's not about multipliers, just cheese.

He meant that ....

Unless you're his brother and he's sitting next to you, I don't we need to argue about what he meant. I understood it differently than you and I'd wager I'm not alone in that. Besides I was also making a general point.

Unless you're his brother and he's sitting next to you, I don't we need to argue about what he meant. I understood it differently than you and I'd wager I'm not alone in that. Besides I was also making a general point.

Granted, but so was I. A lot of people play with settings the AI isn't good at because it crunches numbers. It doesn't have a long list of pre-defined tactic styles. I've never seen it rush, for example. Or turtle properly. Humans can do things that an AI can't because it's a piece of code that analyses and reacts, and can't think creatively. Anyone who isn't playing the AI at it's absolute best really shouldn't make comments about why it isn't strong enough, but I've seen a fair few people who say "The AI sucks because I used a cheesy exploit tactic and beat it!"

they're exploiting their way through.

Ridiculous statement. Plenty of people beat the game at very high AI levels without "exploiting their way through". Saying that is just bloody ignorant, or trolling.

This thread is trash. The AI in GC2 are the AI that other games dream of.

There is a mod to an old game called Ascendancy that truly amazes, and is the reason a lot of people still play such an ancient game. This mod, without "cheating" by giving the NPC races unfair bonuses, manages to challenge even the best of players, but the reason for this is that as far as I know, it was produced by the player community, much like Apolyton's version of Call To Power II.

Much like GNU/Linux, many very intelligent and skilled people put many hours into its development, and the end result is impressive; however, Stardock just does not have the resources to put that much work into the GC2 AI. For a commercially produced game, GC2's AI is far above average in the 4x genre, but depending on how the modding enhancements go I think we can expect to see some very nice AI mods comparable to Ascendancy, SimCity 4 (user-made mods vastly improved this game, especially the pathfinding ability for sims) and CTP2.