Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[sambojin]

Economically it was simple. Essentially compound interest in motion, with quite a few caveats. Hyper-producers, balanced, factoryless and hyper-expanders were all viable economic strategies. Taking habitability, tech and advantages into account all made this even more of a "lots of sizes fit all" type game, even though the basic economy was very simple.

Strategically, it was quite a lot more complicated than that.

Micromanagey as hell, but enough different ways of doing stuff. Slightly imbalanced (CA races were just a bit too good/easy), but not bad/horrible. The AI still isn't a bad example of doing it right, even if a 3:1 or 4:1 advantage was needed to be competitive against a reasonable human player (once you'd tweaked them to be a good race design).

Considering how mathsy the game could become, and the varying strategies available, I'm pretty sure that's the first time I've ever heard Stars! being called "simple". Mind-bogglingly micromanagey, yes. Had an annoying interface, yes. Lots of anti-AI exploits, yes.

But simple? Wow. You must be some kind of genius. Unless you're thinking of a different game. Or I'm really dumb. Was beautiful in multiplayer (although slow to play), and had a vaguely competent AI (more-so than most 4X games), and it didn't even cheat.

[Shadowlord]

I probably am thinking of a different game. :V

Either that or we ignored all the math.

[sambojin]

Maybe that's what I did wrong (although in MP, a bit of quick number crunching was one of the few ways of getting an advantage). Or did right. Heh. I was one of the reasons a patch got made for multiplayer (I learned how to hack/cheat races into a theoretically hard-to-cheat-in game. 20% tri-immune, all the goodies races. But I was a good boy and reported the possible exploit to the devs).

I'm probably overstating the AI a bit. Maybe my tweaked AI races were just very good, or maybe I was very bad at the game. But there's not a great deal of difference in playing Civ or MoO2 against 7 expert AI players as a challenge than when doing it in Stars! Except that in Stars!, 7 expert AI players would regularly beat you down, bit by bit, all without cheating their way to victory. Yes, they specifically ganged up, but it's not as though they don't (eventually) in any 4x game.

[Servant Corps]

some reviewer wrote a great dramatic story about what he thought the AI was doing

If you and I am thinking about that same reviewer...

It's possible that the reviewer is correct. For instance, I don't think it's entirely unlikely for one AI race to vote to "end all wars" so that they can then start a war on an appetizing target. Maybe the code might be less elegant and 'forward-thinking' as the reviewer claimed, and may be simply "I do not want anyone declaring war on this civilization that I want to take over myself. Hey, look, a resolution that can end all wars on said civilization!" But it's still pretty intelligent.

The main problem is that the source code for the AI is not public, so I don't know if that's really the case or if it is "WHEN YOU SEE A RESOLUTION THAT WILL END ALL WARS AND BRING INTERGALACTIC PEACE, FLIP A COIN".

I agree, but all AI is an illusion. We're a few years off a proper AI, so all game AI is just an illusion.

I'd say that, just like highly advanced technology is indistinguishable from magic, highly complex code that can gather data and use it properly is indistinguishable from "proper AI". Or do you define "proper AI" to include code that has "machine learning" capabilities and/or evolutionary algorithms? Because in that case, yeah, we're likely a few years off then. Only a few.

[Sergarr]

Diplomacy in modern games is essentially this:

"WHEN YOU SEE A RESOLUTION THAT WILL END ALL WARS AND BRING INTERGALACTIC PEACE, FLIP A COIN".

To amend that, you'd need to make AI act logically in consistent manner. That would require it to have some sort of memory to remember what it did so that it could act in the same direction, and remember what others did (including player), so it could try to predict their behavior. But implementing that would actually take effort.  And effort is too hard for modern game developers, as evident by GalCiv 3.

[sambojin]

Thus my above off-topic posts.

I'd rather a basic AI be vaguely competent and only be considering war as an option, because the fact of the matter is, every other option only gives me time and room to move, so that I can kick its arse eventually.

A "diplomatic" AI is horribly hard to program. Considering that it has the same "win conditions" on that particular game as me (assuming there's not asymmetric win conditions. Mostly there isn't in 4X games. The first that gets one of those subset of "things", WINS!).

Probably best to make it at least vaguely adequate in scouting, colonization, utilizing tech and advantages and at war/beating you.

Because it kind of sucks to hear "Computer player X just won by Artsy/Colony/Found a planet/Influence/Tech", especially right when you were kicking arse in your chosen field.

Which is why they're not developed for them.

But why they're developed so poorly on a war/conquest victory, I can't say. Most games don't even put a speed bump in the way whilst the other AI players get ahead, a way of challenge, even if bodies and empires fall on your way there. Nope. They just pretend happy until the time comes. As you devote time to slaughtering some little piece of AI code, they're programmed to half-arse an artsy victory. But you're not half-arsing it, you want to win. They can't trust you. They're meant to be going for "insert victory type here". You're going to do your best to fuck them up. But every victory is laughable, because they were never trying to beat you. Barely even each other. At least THEY can trust EACH OTHER'S personalities and outlook. They're not that random. Then there's this huge, monkey-brick-shit-house in the way of those goals. Just to mess them up. And you SHOULD be a speed-bump to them. They should WANT to do their thing. Even if it means war, and calling in every ally they can. You're not even half dependable. You don't care about their agenda. Whether it effects yours or their's. There's no reason why they shouldn't take out the most belligerent, warlike, back-stabbing faction in the game.

You.

Because you want to win. The AIs in most games honestly seem fine with losing. Which is why I don't mind the "cut the crap" of Stars! AI. It doesn't do diplomacy. Why would it? It's not your friend. Your goal is to kick the shit out of it.

Why make a peace treaty? Is not even being vaguely competent at war going to help an eventual outcome? For an AI, or you?

It certainly makes people wonder at certain AI decisions and focuses that were made on a 4X game. Even without the basic 4X's being coded in, relatively hardcore, as the basic win condition.

[sambojin]

As a stranger form of explanation:

I like a messaging based system for treaties and trading and diplomacy between human players. No hard or fast rules, no "you must or must not do that". Agreements made, with all that can be done on them. Very humanly. Even ignoring said agreements. Or making them more expansive.

But in a space 4X, the whole idea is that every other race is alien to you. Especially the AI. No RP needed, no exput from them, especially not canned responses ("Yeah. Have some more time. I'll do other stuff"). There hasn't ever been a deal that worked out to your advantage. Not to help your goals. You couldn't even tell them what they were. They simply never understood your language.

Diplomacy works in MP. But as an all vs you AI thing, it's just trying to get some free time or resources. That's not diplomacy, or AI.

We're way off either if that's what people wanted from GalCiv3.

[Sergarr]

So what you're proposing is to turn off diplomacy and just make it war-oriented. A la most RTS/TBS strategies.

I'm not saying that it wouldn't work, but you would then need some interesting combat mechanics to accompany that, unlike what GalCiv 3 has.

[Henny]

4. All the other factions are very angry at me because "I'm pursuing a tech victory". But I disabled tech victories in the game.

I believe it's because tech victory can be turned off midgame.

[Shadowlord]

4. All the other factions are very angry at me because "I'm pursuing a tech victory". But I disabled tech victories in the game.

Ahaha.

[sambojin]

So what you're proposing is to turn off diplomacy and just make it war-oriented. A la most RTS/TBS strategies.

I'm not saying that it wouldn't work, but you would then need some interesting combat mechanics to accompany that, unlike what GalCiv 3 has.

Unless you can propose an interesting system where you can set up borders, trade exchanges and being able to partake in battle/campaigns that are to their advantage and yours, Yes. Very much so. A basic messaging system between human players gives many forms of diplomacy, treaties, and even subterfuge. Every AI diplomacy option seems to be "let's trade tech", or "let's not war for now", or "let's just be friends. But you're stronger. So I'm your bitch".

WTF?

Can ANY AI's that you know of, in any game, attempt to enforce war upon you, against a third faction? No. They don't even get shitty about your zero troop levels involved.

Can ANY AI's you know of complete your own war goals on a campaign? There's never even a way of telling them what you're trying to do.

Can an AI help you, apart from giving you time or resources, other than helping you kick itself to death? Can they literally assist you on a cultural victory? Whilst they're going for a science win?

No.

So make them at least vaguely competent at war. Then, once you've done that, you can add diplomatic options. Because they're then scary enough to count. Until that point, no human player can consider any agreement even vaguely binding. Even if the last resort is war, you'd better be vaguely competent at it. So I know it's a waste of resources to fight you, even if we're doing entirely different things.

FTW!

eXplore, eXpand, eXploit, eXterminate. 4X.

If an AI can't do the basics, then what other victory options can it do?

[Shadowlord]

Can ANY AI's that you know of, in any game, attempt to enforce war upon you, against a third faction? No. They don't even get shitty about your zero troop levels involved.

The AI in Sword of the Stars will sometimes insist that you attack, or help defend, a particular star. If you don't, they may get upset. They may also tell you that they want a particular star system. Of course, this is only if they have researched the language techs to speak your language (if they're a different species), and they think you may do it (In practice, playing Liir, I tended to get impossible requests - to send someone to defend something in x turns, when I had no way to get a fleet there that fast, but they were probably sending them because I had a strong fleet and we were allied). They send the defend requests when they spot an enemy fleet coming towards a system of theirs that they think they can't hold, I expect.

(IIRC, I've had an alliance cancelled because I wasn't sending help.)

Can ANY AI's you know of complete your own war goals on a campaign? There's never even a way of telling them what you're trying to do.

Same game. You have access to the same system, which is basically a system by which you tell someone what you want. I don't know if anyone actually used it. I don't think it made it into SotS 2, but I'm not sure.

Can an AI help you, apart from giving you time or resources, other than helping you kick itself to death? Can they literally assist you on a cultural victory? Whilst they're going for a science win?

They cannot assist you in a thing that does not exist, no. There's only a military victory. IIRC, if you ally with someone, you all win once you have crushed your enemies beneath your boot-heels (or w/e).

Can an AI help you, apart from giving you time or resources, other than helping you kick itself to death? No.

Wut.

[Mephansteras]

Just popped up at 50% off on Steam today. What's the prognosis on this? Worth it for that price, or should I wait until they get more DLC out and look for a bundle sale at some point?

[Retropunch]

Just popped up at 50% off on Steam today. What's the prognosis on this? Worth it for that price, or should I wait until they get more DLC out and look for a bundle sale at some point?

100% wait. It's really not worth it until a few DLCs/expansions down the road (if they get that far), and even then it'd have to be with a pretty great offer.

[Niveras]

I played a free version for a few hours, but it many updates behind. I bought it to try the most recent version, wondering if the problems I was having with it may have been addressed. They weren't.

I don't mean crashes or broken AI - I can't even speak to the AI. I just can't get a handle on the game itself. I don't know if the map gen is broken still or if it's working as designed, but maps are small, habitable planets (extreme or not) are ridiculously rare (meanwhile duranthem is ridiculously absurdly common compared to everything else), are low quality, and even if you find a planet with a decent PQ rating, it may be nigh worthless because all the tiles are in lines or outright scattered, making it very difficult to really work with the new adjacency benefits. The new trade goods often also take up valuable space, even on your homeworld! Compare a Drengin homeworld versus a Krynn one, for example. The Drengin get 8 tiles in a ball, while most other races the planet tiles are pretty much randomly scattered about... and then you might even get a random worthless trade good taking up a tile (but I guess you could just destroy those if you want).

I don't know if that improves with terraforming later in the game, since I never managed to stick through a game long enough to research to the last of the terraforming techs. I was definitely disappointed to see that the techs unlock only one tile each, at least of the ones I did manage to get through (I've seen hints that the last tech allows you to upgrade all the potential tiles the first tech unlocks). In GC2, the techs often unlock many tiles on planets with low base PQ.

On top of that, they really didn't do a good job at improving the monotony or the interface in general (compared to GC2). Ironically, the non-updated free version I tried handle the constructor game better than the updated version. Handling multiple colonies is still a hassle, perhaps even moreso because of how selections "flow" after issuing a command; having to constantly re-select the list cancelling queued buildings, for example. Or having to destroy a building manually before changing the building that will be placed there. And if SD never had a clue to fix or improve these things in the transition from GC2 to GC3, they probably never will. I doubt the interface is moddable, either, not to the extent required.

I don't know, maybe I just expected a prettier GC2 but got something else instead. I do like some of the new mechanics: the way shipyards work, and the adjacency rules themselves (it is just that, with planets as they are, it is difficult to make use of them in a way that feels viable). GC2 was by no means perfect, but GC3 doesn't feel like it is better yet.

Anyway, I'll be asking for a refund.