AI, Am I wasting my time?

Well, AI is still penalized when set to bright. If you want an equal level field, you should try setting all AI to "intelligent".

BTW, you may want to try smaller galaxies (like large) with 9 opponents to spice things ...

You won the game, no point in playing on. Altough the AI is better in this game than other games it still has many shortcomings. Such as it is not capable of ganging up on a stronger opponent, not that Ive seen anyway. On the contrary if your fleet is bigger than theirs they are more likely brown nosing you which makes victory even easier.

Also, even if their fleets are bigger than yours then they will still not be capable of winning over you by military means. The AI is quite poor at organising it fleets and will most likely leave half of his fleet in orbit around his planets (often without orbital fleet management) and the other half zerging you (mass attack) once you survice the first waves of attacks it has no other tricks in his sleeve.

Furthermore it is not very good at designing ships. It puts way too many points in one particular defence area and all you need to do is figure out which area that is and then put your attacks in another area which makes all his defence modules worth nothing and the AI will often not react to that until is too late. Which makes your ships, if you know what youre doing, alot more powerful than his.

Try playing above BRIGHT.   

And if you haven't allowed the AI to use more CPU, it will design new ships on a yearly basis.

Thanks for the replies, especially Yamoda. That's exactly what I needed to know.

When the AI is running at full (with the intensive CPU option on), does it do better at "tactical" things, or, if I lose, is it mainly going to be because it's beating me with it's "economy"? Simply getting crushed by numbers isn't always fun.

Oh, and if someone knows, if "blind exploration" is on, is the AI also "blind"? I guess even if it isn't I might turn that on. I kinda' like not knowing what's happening on the other side of the galaxy. You never know if there's some monster nation forming in the far reaches of the universe waiting to gobble you up.

sfc

actually, they will gang up on opponents under the right circumstances. if everyone already likes you, the AI players probably won't gang up on you. but if you start trouncing them and aren't too far ahead, they'll develop of fear of "i'm next." in this, they consider killing off minor civs as well. so if you want to aggrviate them, play with minors and kill a bunch of them. and do it quick. the AI players will start attacking you. otherwise bide you time until you can take them all on.

yes, i believe.

Up to and including "Tough" difficulty ("Intelligent" intelligence level), the AI doesn't get any bonuses, so if you lose there, it's because you just plain got beat. It isn't until "Painful" difficulty ("Gifted" intelligence level) that the AI gets boosts.

So going from "Bright/Challenging" to "Intelligent/Tough" WILL make the AI a bit smarter.

I don't know about blind exploration, though, so I'll let somebody else answer that bit.

Actually, for the AI, blind exploration is always on.

I will totally second this. Even in 1.4 DL the AI could sometimes be very opportunistic. The very last game I played before switching over to DA, I was the second most powerful militarily next to the Drengin, who were a very real threat I had to deal with now. I had commited nearly all of my forces to the Drengin front and was a few turns into a pretty tense fight with them, when all of a sudden...Treachery! Two of the Three remaining powers in the galaxy both declared war on me simultaneously from the other side of my empire...and I had been at "neutral" or better relations with both of them. Some of my few remaining obsolete ships fought valiantly to delay them while I redeployed.

It was DL, so I had no idea if it was part of some backroom diplomatic deal or not, but it does happen. If youre playing on the higher difficulties, and leave yourself undefended too long, you can very easily start a "feeding frenzy" of races declaring war on you as well.


A couple points to the OP:

The game doesnt really begin until the "tough" difficulty( I think thats the one), when the AI is fully enabled.

I dont know which version youre playing, but from what I understand the +cpu option in Dread Lords just increases the frequency with which the AI will redesign ships. In Dark Avatar the +cpu option has a much more significant and broader reaching impact.

If you havent picked up DA yet, you definitely should...its much more challenging all around, and due to the speed nerfs the AI has a much better chance of "understanding" the context of player moves and acting accordingly.

Your doing well to dive in and play galciv like that!

i started playing galciv2 as an experienced galciv1 player, even so, i lost the first couple of games after my government spending ended up going out the back door!! I'm like, "what the f--k is wrong with this economy!!!!".

If there are continued reports that the AI is significantly better I guess I will be buying the expansion soon. It'd be great to actually play against an AI with a real tactical brain though. In about every game I've ever played I've overcome my mediocher economy (I don't go crazy crunching the numbers) by fighting sometimes very tricky (or "cheesy") wars. Most AIs fail badly at concentraing their force in the right area, and so you can defeat a huge army with a smaller since most of the huge one is just sitting around doing nothing. Another dumb AI move is constant attacks against highly defended areas. I think one of the quirks of planetary invasion is to help the AI: the fact that the SIZE of the attacker doesn't matter. That way it's not so much of a disaster for the AI to use a force that's too small in an invasion, which it's probably fond of doing.

Of course, the tactical bit would be much more difficult, and the number crunching economy a more obvious place for a computer to take the advantage. I guess what you want is, at least, an AI that gives a semblance of tactical ability so that when you pull off something really impressive, you think it's because of your own genius, and not the AIs bumbling.

sfc

Mystikmind,

Well, I didn't just dive in. This game is really terrible for "just diving in" as the manual is useless(ly out of date) and there's no bloody "civilopdedia" (in game). I nearly never got going on it 'cause I couldn't figure out how things worked. The Wiki saved the game for me. It's REALLY frustrating to research Alliances and then not be able to offer another civ one. Hmm, why isn't is on the trade screen? Maybe it's somewhere in the Foreign Relations tab. No.... what's the manual say... nothing... hmm.... ahh, finally, on the Wiki it says I have to at least be at level 7 of friendship...back to the game... how do I know what exactly what level I am in a relationship?? Do I just count the plusses and minuses...hmm... back to the Wiki...now Forums..hmm. Repeat that excercise for about 10 to 20 different aspects of the game. Population growth and battles are real head spinners too, Yikes. Oh, and a favorite, the hardcoded Range limit! I built a colony ship named Columbus that could go across the entire map but for some reason it can't. Why not?? Back to the Wiki! Ah, great, that was a real waste of resources!

Luckily, things have generally been worth it.

sfc

Yes, the game does have a bit of a steep learning curve, but that is part of it's attraction. The complexity of the game leads to an almost infinite variety during gameplay, and should keep you interested for months to come.

Frogboy has rewritten the manual himself. It is night and day compared to the first one. GalCiv2DL has been updated, with a few bug fixes and DA compatability stuff, but since the .pdf manual is for both games combined, perhaps it is in there as well. I have DA already, so no way for me to check, but it sure wouldn't hurt to update and see for yourself.

Take my word for it, friend. It is better. Prepare to have your *ss handed to you.

Well, while DA is a much more challenging game (you could easily find yourself playing a couple levels of difficulty lower in DA than you were in DL), I also think I detect some unrealistic expectations here. There are very pronounced limits (derived both from current hardware and finiancial practicality) to any game AI (look at the kind of super-hardware required to match a human player in chess, and that game has a very limited and predictable scope compared to the near infinite number of possibilities in a single turn of a game like GC)


Any human player is going to discover weaknesses in any AI. That being said, there is no current game that Im aware of that is any better in an overall sense, and Brad never stops improving it. Some one above described GC2 as "one of the better (AI's)", but I would challenge them to name a current game thats really any better in this regard. The very best one they could name would only be a different set of vulnerabilities, with most of their challenge coming from "cheating". The GC 2 AI at least puts up a decent show well before it resorts to cheating, but eventually youll have to allow cheating from any AI to provide challenge to an experienced player.


You havent even played the best that Dread Lords has to offer though.("Bright" enemy intelligence level is still the AI playing with its arms tied behind it back), and if its challenge youre after, even DL will spank you soundly (at first) on the three highest difficulty settings.


tip: Dont play "random Intelligence", at least not while youre learning the game. The AI races already have hardcoded individual personalities that make them unique in different situations, all randomizing intelligence will do is ensure that some of them are gimps just talking up space for the AI monsters to quickly gobble up, as well as making it difficult to get a real idea of the challenge level of each of the difficulties.

This does not show that AI gang up on stronger opponents. If ot would then it would make more sense that they attacked the Drengin since they was the most powerful. This was probably just a coincidence.

But this could easily be confirmed by the devs, if AI gangs up on stronger opponents. For all the games Ive played the AI has never ganged up on me because I was stronger, quite the opposite: if I was alot weaker than the AI then it frequently would declare war on me.

That makes perfect sense since weak empires are easy pickings. But it should consider the fact that the strongest empire is a threat so it should either try to ally with him or try to gang up on him with other empires to either take him down or atleast weaken him. But I have never seen any evidence that it does either of these.

This is one of the improvements the devs should look into, along with the design issue of not crippling itself by putting too much defence modules in one particular area.

For me, a good AI isn't simply one that plays the game farely well, but one that makes an interesting game. The AI should have built into it a trigger to try and keep the status quo. The history of Europe shows how natural this sort of thinking is. If someone starts to get a real advantage, it's obviously in the best interest of everyone else to stop them.

Oh, and I know I haven't experienced the full AI yet. I'm gonna' start on that now  

I did find one thing interesting in my last game. The Korx were really weak and it looked like there'd be a "feeding frenzy", so I jumped in. I was closer and had the troops ready and captured the first planet and was poised to capture a couple more. But the Korx suddenly capitulated, but not to me! They gave like 7 planets to one of my opponents. That was a fun shock. Like when someone gives all their property to another player in monopoly just to spite you    I wasn't sure if they simply gave up to the first country to declare war on them, or if it was specifically to annoy me.

Well obviously the AI is capable of noticing that an empire is growing too strong because of the messages they keep sending me that empire XX is growing too strong and that we should do something about. But the second step, to actually do something about it, seems is not working or not implemented.

It doesnt happen "often", but it does happen. Two races, at neutral or better relations, do not "coincidently" both declare war the same turn, right after youve started an invasion against someone else. Races in good diplomatic standing generally dont declare war on you at all, baring some diplomatic doublecross (payoff). Surely in your dreadlord games, youve gotten the Diplomatic message from someone youre one your friendly neighbors about the "growing threat" of race X, and how you should "work together" to combat their power. So clearly its something the AI is aware of, and the potential is there.


The "most powerful" statement was based on my own approximation, and not what the game was ranking military strength at, which I mostly ignore as being wildly innacurate. Its entirely possible I was the #1 military. Its also possible the Drengin just paid them both to attack me. Maybe they were allied ( though I dont think so..it was a bit early for that).


Either way, it just shows that all sorts of stuff can and does happen. It may not have happened precisely with the AI motivation of "weaker races ganging up against you", but the net result is the same. Certainly it could improve, but I just wanted to provide a counterpoint to some of the more "overstated" points that were made, and show that not all game experiences are the same. There are many well documented Galactic events and phenomena that I have never experienced in all my games.


Im all for continuing to improve the AI, as are Stardock. I too would love to start seeing more diplomatic events like explicit military alliances formed against nation x. But a sense of perspective is neccesary. There is no turn based strategy ( or probably strategy game in general) game in recent memory with a better "overall" AI, and that was true of Dread Lords, even with its vulnerabilities, let alone DA.


Agressiveness does not equal intelligence, or you have a situation like in M2TW, where even the puniest nation will attack your vastly more powerful nation without provocation, and not accept a free ceasefire even when you have reduced them to a single minor city. And if you ARE the most powerful military force in the game, depending on force comparison you may not even blink at two minor races declaring war on you, and they'd still be destroyed. In which case the smarter thing for them to do would not be to hasten their destruction, but change tactics and start aggresively pursuing a diplomatic or tech game victory (which also could stand to happen more often).

hehehe ok.

Just sounded like in your original post that you picked up the game and started playing sucessfully at bright level!!

I don't even know if that would be humanly possible without first researching the game??

It doesnt happen "often", but it does happen. Two races, at neutral or better relations, do not "coincidently" both declare war the same turn, right after youve started an invasion against someone else. Races in good diplomatic standing generally dont declare war on you at all, baring some diplomatic doublecross (payoff). Surely in your dreadlord games, youve gotten the Diplomatic message from someone youre one your friendly neighbors about the "growing threat" of race X, and how you should "work together" to combat their power. So clearly its something the AI is aware of, and the potential is there.

No doubt it can happen but the reason why it happened doesnt neccesary has to be because "AI gangs up on stronger opponents". It can be that the empire you decided to attack had good standings, or maybe even allied, with those other two and he asked them to attack you.

Im all for continuing to improve the AI, as are Stardock. I too would love to start seeing more diplomatic events like explicit military alliances formed against nation x. But a sense of perspective is neccesary. There is no turn based strategy ( or probably strategy game in general) game in recent memory with a better "overall" AI, and that was true of Dread Lords, even with its vulnerabilities, let alone DA.

True but with almost all games having multiplayer cappability and with that the option of playing opponents with vastly superior cappability than any AI, that is kinda moot isnt it?

GC II does not offer multiplayer cappability so it must offer excellent AI or the game would get boring, real fast.

Agressiveness does not equal intelligence, or you have a situation like in M2TW, where even the puniest nation will attack your vastly more powerful nation without provocation, and not accept a free ceasefire even when you have reduced them to a single minor city. And if you ARE the most powerful military force in the game, depending on force comparison you may not even blink at two minor races declaring war on you, and they'd still be destroyed. In which case the smarter thing for them to do would not be to hasten their destruction, but change tactics and start aggresively pursuing a diplomatic or tech game victory (which also could stand to happen more often).

True but remember that I said that the AI should gang up or try to ally with a superior empire for diplomatic victory. Going for influence or tech victory might or might not be viable but most likely would someone going for military victory squash you like a bug if you tried going for tech or influence victory.

The AI should do some kind of probability check for which action it should take agaimst a superior opponent trying to win. For example it could add up its military power with all the other AIs military powers ganging up on the superior opponent and then compare it with his military power and any allies he may have. If they are vastly inferior then ofcourse it shouldnt declare war on the superior one but if there is a reasonable chance of taking him down (or weaken him) then it should.

I confess I have no idea how hard this is too code, Im just saying that this is one of the things AI need to do if its gonna be a challenge for veterans.

But as both you and I have said the AI can notice a powerful opponent but can it properly react to such a situation?

Also Ive read a previous post from saying something of the effect that for "roleplaying" reasons the AI wont do this and that just for the cause of winning. And this is bad because then, if you want a challenge, then you as a player must do the same. That is, even if you could win, you choose not to because you are roleplaying a peaceloving nation going for tech victory and for me that means trying to win with a hand tied behind my back. And is a reason that I believe that there needs to be a level where the AI plays to win, disregarding roleplaying.

Roleplaying is overrated btw. Ive seen several polls for people playing MMORPGs that only a minority actually do roleplay. Other people just want to play and have fun and for many people that includes being challenged.

I dont see how defining the top of the curve for AI in single player strategy games can be "moot". It seems you are basing your statements on the pursuit of some theoretical ideal that doesnt exist anywhere outside of you own mind, rather than any practical metric.

Single player games dont, cant, and wont for the forseeable future, compete with human opponents in terms of "opponent intelligence", which doesnt mean they cant be challenging...but you knew that going in, and GC never claimed to be anything but a single player game. The multiplayer argument is out of place here.

Why is that?

I mean there is no better measure for an AI than what human players do!

Arguments for and against aside, a multiplayer would certainly provide more intriguing challenge than the AI. I can just imagine all the dirty tricks you could be suprised by, that the AI currently would never do to you! That would be a weird new universe!!

Im all for multiplayer. Yes, yes, a thousand times yes. But the above poster suggested that the fact that GC2 is as good as it gets for SP strategic AI was inconsequential because a human would be smarter. Human player level IQ is the holy grail for any AI, but that aint happening any time soon. And until that day in the future when my games are outhinking me, when Im flying to work in my shiny flying delorean time machine , multiplayer games are always going to be against "smarter", less predictable (though not always more challenging, depending on how badly the AI is cheating) opponents.


But any reasonable person knows this, and if they wanted a multiplayer game, they probably wouldnt have purchased a decidedly single player game. And overall, as far single player strategy games go, GC2 is about as smart as you can currently get. There are things Id like to see better..things can always get better...but thats a truism. If Brad worked on the game non stop for 50 years, it would be just as true.


Whats important is to keep a sense of persepctive with your expectations, and measure it against the AI intelligence of other SP strategy games. If youre going to measure the game and find it significantly lacking based on some imaginary picture in your head, or against human level intelligence, you may as well just throw out all your SP games. The Multiplayer point was only broached when the poster had to acknowledge that he was making "it still has a ways to go" arguments against something that arguably is already better than anything else around at what it does. In which case, what is this theoretical peak its supposed to reach? It IS the peak.

Why did man learn to fly?

Because he saw birds doing it!

How do AI's get better?

Whatch what human players do!