Basicaly the people coming to defend the AI have not exposed its weaknesses yet via gameplay. You will eventualy. Then you will be in the same boat as the rest of us of being able to beat the game on at least tough every time.

I second that.

There have been postings trashing the AI which were one-sided and exaggerated, even offending to the devs and Brad in particular. This is disrespectful and nothing constructive will come out of it. But others have been flaming people for nothing more than describing AI weaknesses, and those made me think "you simply haven't seen yet what I've seen".

GC 2 makes a lot of publicity with its AI, not without reason. It is also a big argument against the "necessity" of multiplayer. The downside of this is that the AI will be watched very closely, and that people might be disappointed because they expected more than there really is.

When given the opportunity, I'll gladly try to help improving the AI by describing weaknesses I've found or gamey tactics which could be prevented - but don't shoot the messenger.

If you find the AI good, then you can express your satisfactiion, you dont need to be so rabid in your attacks on those that might not be as satisified as you. You faill to make any rational arguements and just resort to ad hominem attacks on those that arent at happy as you. You basicaly accuse me of being a liar and a spoilt brat.

As I have said several times in this thread, I like GC2 and think the AI is strong in some areas. However I point out the FACTS that it has some obvious failings. I am quite specific about what these are and am sure they can and will be attended to in future patches as that seems the intent of the developer, which I applaud. Please point out where I have demanded anything. I did say that the AI was hyped and it is. I play alot of less mainstream games so perhaps I have higher standards. I realy thought in this game the AI was going to be hard to beat wihout it cheating but that is just quite honestly not the case. Im sorry but for me that is the truth. I dont know why my description of my experirences with the game have to offend you so much. I dont explode with self rightous rage every time some one describes the contrary to my experience and says they find the AI hard to beat do I.

Stardock have my money, they have been compensated for any effort they have put into the game. The logical extension from your attitude is that we should swoon with gratitude everytime we buy something that is not a total and blatent rip off.

It seems like stardock are good devs but I can think of several devs that are just as good in there relations with those that play there games. Battlefront. Panther Games Shapnel games to name a few.

Of course this is good, but realy it is what we should expect as given. If we did expect it as a given and used out consumer power to punish those that are the worst examples of publishers such as EA games then perhaps PC gaming wouldnt have been going down the toilet for the last 5 years. If we take the attitude that one is spoilt to ask for quality then we deserve to get hosed by the industry/





Edited to Add: Also the victories I have won have all been diplomatic victories. After you have won a few miltary victories you will be so powerfull that other civs wont attack you so you can just give em stuff, anyone who dosnt become your ally. well thats a very bad short term decision by them.

Next time, I will try a medium map with only five opponents. Maybe that will help. Does the AI become tougher if it gets bigger? So far, I just kinda do whatever I want without much opposition. Also, I think the AI needs to be more aggressive. If I play peacefull, they never attack me. Maybe I need to get up to a more difficult level. And also, because they treat each other the same as they treat me, they spend a lot of time just grinding each other down while I sit back and get staged to dominate. Just some comments on these lines: I am normally a Gigantic map type person.... but I have found that those exact settings: Medium map, 5 AI tends to be an excellent game with solid wars, good AI defense, culture becomes important and the player ability to out-colonise the AI is significantly cut down. I also tend to use rare suns, rare planets, rare habitable in order to leave everyone with basically just their system. It takes a while to kick off (except the Drengin who build a large fleet) but when it does, it gets pleasantly messy. On the AI not attacking you.... if you play at intelligent and start as either a Good or an Evil race... someone will probably declare on you by the end of the 1st year..... then they will stay at war until either a) They win They get into a lot of wars c) You buy them out .. or d) You build a bigger fleet than them. If you play at lower levels, the AI seems fairly random in who it will attack from what I have seen.

Thanks you for this advice. My current game is on crippling, so I do believe the AI is not hindered in any way. I don't want to pick an evil or good race however. I like the Terrans and just want to stick to that. I will up the difficulty again next game.

Also, I think the AI needs to be more aggressive. If I play peacefull, they never attack me. Maybe I need to get up to a more difficult level. And also, because they treat each other the same as they treat me, they spend a lot of time just grinding each other down while I sit back and get staged to dominate. What the *(%#? I initially thought the AI is TOO aggressive, until I learned what makes it tick and learned how to make the AI leave me alone, eventually! Do you guys invent this stuff or what?! It's the same "MOO2 had the best AI ever, amazing 3D battles and even GODZILLA, and it was giving you money from the CD-ROM drive, too" syndrom Stop slinging mud for the sake of slinging mud, what's up with you?

Hey, I am just posting my own experience. Like I said above, I am playing on crippling. So far, I haven't been seriously threatened. As the above responder suggested, perhaps I need to not be so neutral in order to be attacked more readilly. But no, I am not making it up. And my intent is not to bash the game. The more I play it, the more I understand it. It IS an enjoyable game. I just want what many other people want I think, a more improved AI. The Devs do respond often and that is great. That is prehaps the same reason that so many people complain. Because they think their voice counts!!! If the game is hard for you, and you can't keep the AI off your back then that is great. But so far, it hasn't happened to me, so back off.

I'm not a comp. sci. major, more of a general sci/math major, but I took a couple of basic comp. sci and logic classes for fun. And one thing I understand is that programming AI is one of, if not the, most difficult undertakings in programming. Humans are best at using heuristic solutions through abductive reasoning, and the worst (in general) at deductive reasoning. Computers, OTH, use deductive reasoning (the type of reasoning which they are spades better at than humans) and make use of algorithmic solutions, and are poor at abductive reasoning.

Think to yourself and see if you can come up with some algorithms for playing this game, and you will see how difficult the task is.

Brad, I understand the AI is your baby, while some people (some probably unintentionally, some not) are insulting it, try not to take it too much to heart. Understand I'm not trying to insult you or your AI (as I have some appreciation of the difficulty involved in its creation). However, I am confident you can still strive to improve upon the infrastructure you have built and can make the AI even better.

However, know ahead of time there will still be those that aren't satisfied, but I am glad you are willing to do your best for the love of your product.

Hope this post doesn't look as smarmy as I think does. It is really meant to be genuine.

pasty11, calm down. No one is accusing you of being satan, and I doubt you have murdered any Kurds lately (or have you? ) I hope you realize that I'm joking. I agree that constuctive criticism of the AI can be good for the game. But try to understand Brad is willing and trying to improve it the best he can, and it won't happen over night. As I said before AI is very difficult, but rest assured, Brad is on the case. Please share any concerns with the behavior of the AI that you see as well as any ideas you have to improve it and hopefully we can all make this game more enjoyable to everyone.

And this last part is not directed to anyone in particular, so please don't fly into a rage upon reading this, but remember the phrase you catch more flys with honey than vinegar, and express your concerns in a non-accusatory or insulting manor.

Oops, I meant to say manner, not manor in my last post! I tried to edit, but I kept getting errors.

Guys...

Let's try to keep this discussion civil.  There's a lot of very good constructive criticism here.

I am definitely happy to hear from people with strategies and ideas that would make the AI more effective.

The only place I tend to get off the train is when people start asserting that the AI is "broken" or "bad".  Every game has its strengths and weaknesss and bad AI is not one of this game's weaknesses.  So it's hard not to just start tuning people out who start out by saying how "bad" the AI is because I know it's not bad.

The second issue is that people need to realize that the AI is NOT scripted. That's one of the things that makes GalCiv II's AI special.  It's is NOT told to do X then Y, then Z.

Instead, it evaluates a bunch of conditions and designs its own solutions.  So it's easy to sit back and say "Good grief, that stupid AI build a technology capital on a planet that had few research labs."  I can't solve that with a simple IF THEN loop. The AI needs to be expanded on what it evaluates and what are positive solutions.

Same is true on ships. It's easy to say AI should build faster ships. But the AI doesn't script out what technologies it should get. The reason the aI doesn't currently tend to have faster ships is because it doesn't priorities getting faster propulsion techs enough.

If there's any CS people here, I think they can tell you that the fact the AI designs pretty decent ships was a non-trivial thing. Especially given that it's not cheating to do that. Consider that -- even at the highest levels, we don't just let the AI stuff things on its ships.   YOu can build the same ships.

For many people I think what makes GalCiv II fun is that they realize they're playing a game whose opponents are playing the same game they are.

A lot of stuff mentioned here I've already addressed in internal builds -- particularly colony management.

I think the MAIN problem here is that there is a diverse set of people in this thread, and the thread has an inflammatory title. If the thread said "AI can be improved", I think everyone could agree with that. But the title of the thread makes it sound like the AI is nothing special - the same old, same old that you get in any computer game that you buy these days - which I think most people, myself included would disagree with very strongly.

Besides, its always fun to some people to target a game's key strength and start a thread that says "Nope, it sucks!".
Freudian's post is actually quite reasonable, its just the tone of the title that has got some people fired up.


Dano

Brad, good to see you're still with us and not turned away by either destructive comments or petty disputes.

The GalCiv 2 AI certainly doesn't deserve being called "bad" or "broken" at all - I've never seen a better one in a 4X game.
The "problem" is, the better the AI is in general, the more I wonder when seeing things like a wonder on a +300% production tile (good to hear you're aware of the problem) - I can't help but ask "why, oh why has it done THAT?!".

I'm not so sure that research is actually the problem when it comes to ship speed: Admitted, I'm a bit of a speed freak (for good reasons I think), but my ships are almost always faster than AI ships - even when we're on a similar tech level. I don't know how the AI ship design choices are realised, but speed should be valued more I think (compared to the latest official version I'm playing, of course).

Other AI aspects like tech trades, trading away of starbases, starbase design (except for resource ones!), colony management and the like have been mentioned already, I won't repeat them unless specific points come up.

Oh, and thanks again for your dedication to the AI!

There is one cheap algorithm I hope gets fixed, and it's the way the AI trades tech...apparently, they're coded to NEVER trade military techs if you have any kind of strong military.

Big agreement on this one. AI's obviously want to avoid trading military tech; but it's not balanced. An AI was willing to give me 12 non-military techs and 2300 in cash for a peace treaty, but no way would they part with the most basic shield technology for peace. Kinda silly.

Well, I toyed with the idea of starting another thread for constructive AI feedback, but since Brad is still here, here goes. First off, I'm going to say that this feedback is based on what I've observed. If some of this stuff is already coded in the AI, well, all I can say is sorry, but I haven't seen it - maybe the routines could stand another look-see.



Planet utilization:


The AI should place much higher priority on putting the appropriate building on special tiles. I would argue that nothing, no Galactic Wonder/Trade Good/Super Project is better than having a factory on a 700% bonus production tile, for example. The same for a 700% research tile (although I've never seen one). With regard to 300% tiles, the difference is a little softer. Personally, I would never put anything but the appropriate building on one of these tiles either (except for the food bonus tile).


I think the AI should begin to keep track of a parameter which I call the "state of colonization". This refers to how much of the galaxy has been claimed - it should be possible to come up with an AI estimate for this even without cheating based on using the galaxy map to look at influence.


No manufacturing/economic/research capital should ever be built on a poor (low PQ) world. As Brad said the AI isn't scripted, I believe it should be able to project, based on it's current situation, how the worlds under it's control will look like when fully developed. It should place the capitals based on that projection, with the likelyhood of actually doing the construction based on how close the "state of colonization" is to complete. Once a capital is built, the AI should have a bias toward building the appropriate structures (e.g. factories for a manufacturing capital) on that world. A small bias is probably fine, I don't think we need to see a fourteen factory+manufacturing capital world.



Ship construction:


Speed, speed and more speed. Nothing is more important in the early game than speed. While the "state of colonization" is low (early), the AI should be biased heavily towards speed research through Impulse Drive. Increases in engines should be immediately reflected in upgraded colony ships. Furthermore, the AI should never use the original colony ship design. With even primitive Terran Alliance technology, you can design a colony ship with 2 hyperdrive engines and 2 basic life support modules, a speed 3, 3.5 parsec range ship.


An advanced colony ship scramble technique. I always pack the maximum engines on a colony ship and only add life support to fill out the hull. If I reach a situation where the colony ship needs more range, I will create a new design with one less engine and more life support. I will not necessarily build that ship though - instead, I will run a fast colony ship as far as it can go, then upgrade it at the edge of my range. (Actually, this is a place where the AI could do better than I, as it could evaluate the turns lost to upgrade and figure out exactly what the best distance to upgrade is).


Speed should be more of a priority for fighting ships as well. I think it may be counter-productive on tiny hulls, but small hulls should always have an engine I feel. Medium hulls should always have one, and possibly two for fast "raider"-types. While I almost always add two engines to large hulls, I'm not sure I recommend that - but one should always be there. Transports, needless to say, should always have engines, preferably a pair, or even more.



Ship utilization:


Transports - when running transports up to a world to invade, the AI shouldn't just run the transports up to the world on auto-pilot. The "Rally point" for an invasion should be about 1 turn's move *away* from the target world to minimize the chance of interception. Furthermore, combat fleets should be targeted to this rally point as well in order to take on the possible defenders. The AI should realize that even though a world may be undefended, unless you can move a transport onto it in one turn, any human player can and *will* rush buy at least a cheap defensive ship. So a transport without escorting ships is *never* a useful idea. (This is a different decision from having ships in the transport fleet).


Attacking with fleets. A common complaint that I share is that the AI doesn't attack enough. This has a host of ramifications, not least of which is the occasional "Defense is useless" thread on the forum. If I had to guess, I think the AI is evaluating whether or not to attack based on estimating whether or not it can win the battle. My first suggestion would be that this at minimum needs to be changed - even if the likelihood is that the attacking fleet would be lost, the AI should look at the damage done to the defending fleet - if the damage done to the defenders hurts them worse than losing the attacking fleet hurts the AI, it should attack. It should also evaluate the possible outcome if the AI does not attack and the player attacks the AI - if it is better to attack than to defend, and the AI cannot retreat (for technical or strategic reasons), it should attack. I believe taking those two evaluations into consideration will lead to much more attacks being green-flagged.



Research:


Certain technologies seem to be dramatically underweighted by the AI. Xeno Entertainment, for example. Or Xeno Ethics. Sensors. Whatever routine the AI uses to decide what to research next could stand some adjustment. That said, I'd like to propose something that may or may not be possible: the "Research Alliance". This is a trade agreement, but instead of an agreement to trade current technologies, it's an agreement to trade *future* technologies. Basically, it's the kind of agreement human players would make in MP - "I'll research Laser III, you research Enhanced Deflectors, then we'll trade them." It would then lock your research into researching whatever you agreed to research, and the trade would happen once both projects were completed. Since this would basically double the research speed for the empires concerned, it would only be possible at a high level of friendliness - hence the term "Alliance". Oops, that's a feature suggestion, not AI. Heh


There have been many suggestions about trading - I'm not sure what to do with them, and I don't have anything to add at this point.



Starbases:


As has been said, starbases are an area where the AI could stand some serious beefing up. Many small starbases are much less useful than a few big ones. The AI should give more weight to adding modules over creating additional bases. Defense of starbases should play a role in the AI attack algorithms as well - if an enemy fleet is within it's attack range of a starbase, fleets in the area should evaluate attacking it. Again, win/lose should not be the only determinant of whether an attack is a good idea, it should also look at whether the AI fleet can damage the opponent enough to make them unable to take out the starbase.

When at war, Military starbases become primary targets. The AI should divert/construct more constructors to beef them up, as well as using fleets in the area to defend them. When on the offensive, the AI should evaluate where it's attacking, and give heavy emphasis to attacking enemy military starbases in the area. It should not do this indiscriminately (otherwise players will just spam no-module military starbases), but instead evaluate whether the starbases are contributing to the enemy forces significantly or not. This becomes much easier if the AI builds high speed "raider" type ships.

Also, on the offensive, the AI should bring it's own constructors with the fleet, and in areas of heavy combat, drop its own military starbases near the front lines.

patsy11 said: The games I know of that have the best AI are indy titles made by small development teams on small budgets. Namely Highway to the Reich (Amazing AI) and Dominion Wars 2. Is it a coicidnece that indy titles have great AIs and mainstream titles have poor ones, not if the mainstream is using the AI of RTW as a yardstick.

I believe patsy11 is referring to "Dominions II: The Ascension Wars". I can't strictly say one game has a better AI, as I've seen many AI anomalies in each of them, and the games are different. Dominions is ten times more complex than any other game of which I've heard, and its AI beat me for the first several months, even though I now consider it unchallenging. But then, Go (a trivially simple game) has no AI algorithms that can ever beat an experienced player. So there's not really a direct correlation between complexity and AI capability, either - I've never beat a Go or Chess AI, but I've beaten all other AIs I've ever played, and only been troubled by the AIs for Dominions (for a few months) and Total Annihilation's player-made AI mods (that cheat quite extensively). And in all that, I can't say I've ever seen strategy game with better AI than GalCiv2 (despite its weaknesses). Jagged Alliance 2, Unreal Tournament, and Total Annihilation (all ancient games) had good AI, but none were strategy games (the good TA AI was tactical AI).

Maybe people have too high an expectation? Strategy AI is not a solved problem. It's easy for a human to notice that the AI spams useless, level-0 starbases, but the literature has no algorithms to regulate such things. I learned today that Dijkstra's shortest-path algorithm solved a problem that was open for 6 years, and I was shocked, because it really is a trivial solution to a trivial problem. But that's how math works. You can't just demand that someone find the roots of a quadratic formula - since it's obvious were they are when you draw a parabola - before the quadratic formula is invented.

Personally, I like the fact that GC2 is improving at such a rapid rate, and don't plan to play it again until it's "peaked" - similar to skipping appetizers so that dinner is that much more enjoyable. But no experienced person should expect to be demolished repeatedly by an AI for a complicated turn-based strategy game that plays on equal footing, when the entire human race can't devise a good AI Go opponent.

You know... galciv2 is a multiplayer game. It is everyone-versus-brad. now *thats* what I call a MMO