Well said Wombat!!! Now back to the first post...



Possible? probably, likely... no.



Possible? probably, likely... no.

Have to wait to see what they do in the next one... if you like that stuff I can suggest galactic imperium one or two though from memory they are real time (with time controls), or better yet wait for Imperium Galactica 3

--------------------------------

Graphics don't exclude great gameplay, intelligent AI's or "real strategy". Beyond invested finances and the design doc the two should be irrelevant to each other. Do game designs get catered to a certain audience... yes, all the time.... no.

Harel Eilam can see where your coming from and I agree, like Wild Wombat just said build it and they will come.

Never say never, the only limits to what you can produce is the limit to your imagination and the depth of you wallet. Might leave it there too much fuel in this thread for me to join it clear headed.

~SDC~

Wombat, I disagree with the fact that the companies pay. It's the customers who end up paying. To develop great graphics, you need the insurance the game will sell because it will cost more. So we can have games with deep gameplay and good graphics, but we won't get many of these, because they must sell.
Also consider nowadays' graphics are much harder to make than before. The first computer games had no artist in the team. It was plain text, ascii pictures. Now, the trend is more graphic artists than programmers. This is because many things programmerese can be reused (3D libraries for isntance, existing packages for ai, 3D engines...) whereas graphics have to be made up everytime. The tools make it simpler, but you cannot reuse graphics, while you can reuse code (your game core like from CallToPower to CTP2, generic libraries, 3rd party engines (3D, ai)). Thus developping nice graphics is actually more expensive, while developping good gameplay requires more creativity. Big companies will sometimes give you both, but more often than not they will offer you solid tested gameplay with better graphics. Smaller companies which want to have their games successful will have to compete on gameplay because it is affordable, but will never have state od the day graphics (though the graphics they release will probably be slightly superior to those the big companies released the year before).

LDiC: it becomes a bit of a chicken and egg argument, but no game ever has "insurance" that it will sell. Companies are in the business of taking that risk; of going out and getting venture capital and trying to make a product that will sell at a profit. It's like film making. I could walk in to Paramount Pictures and say: "Hi, my name is Wild Wombat. I've got a great idea for X-Men 3 and I reckon it'll cost about $500 million. If you could just sign the check, I'll get cracking. Oh...and do you have Hugh Jackson's phone number?" Of course they wouldn't do it - I have no history of film making. I'd need to make a small, well received short film, then a critically acclaimed medium budget film before they'd consider trusting me with a big film (you know what I'm getting at).

Sure, an individual game designer is going to need to show they are worth investing in; the point I'm making is that consumers should be more aware of their power in the marketplace. We should be saying that we wont PAY for games with cruddy graphics, or bright shiny things full of bugs with lousy AI. I would have thought that sales of MOO3 would have been WELL below predictions - that publisher thus learns a lesson and next time SPENDS the time and money to make a product the customers want. Companies are then perfectly entitled to set a price that will recoup their R&D costs - that's how it works in every other sphere of business, isnt it? If the average game sells for $50 and the perfect game with great graphics and AI sells for $70 - do you really think people would NOT buy it?

If you construct a budget for your brilliant game idea that means you'll need $70,000 worth of programming time and $130,000 worth of graphic development...then don't bother only borrowing $100,000 from your venture capitalist and crying to me that I'm being unreasonable by asking for good AI AND good graphics. Your project is UNDERcapitalised and you will pay the price of developing an unmarketable product.

Look, I know this is sounding really harsh, and I don't mean it to be quite this way, I just want consumers not to roll over and accept stuff just because Microsoft/whoever says we should!

Either that, or they think, "Gosh, those strategy games are nothing but trouble - let's stick with the glitzy games, they're safe bets." :notsure:

That's pretty much what I do, but this has little effect. I came to the point where, looking at the games available at the store, I bought civ3 after having decided a while ago it wasnot worth buying because I hope it will show developpers there is still a market for TBS games. And I don't buy that many games either.
I don't think that MOO3 failure is one of time and budget. It's also been a nice piece of bad management. They kept saying that Alan Emrich was jsut a designer among others and that they had no lead designer. A project without a leader is dead from the inception. Pouring more time and money into it won't change the outcoem much, it's better to start again from scratch.
As for getting funds for the project, I am not sure that such things can be done. Look at the difficulty Stardock has of getting paid or getting preorders (which are actually just saying "I'm interested in buying and I commit myself)". Would a venture capitalist spend a lot of money on a developper like stardock? Big publishers probably dismissed Galciv because 'the genre was dead', so they had to use other business models and use less funds than MOO3 would have had. They couldn't use graphics there.

I don't think you're being harsh. I think that a venture capitalist (or publisher - in MOO3 it was the publisher who provided money - ) would only risk that much, and not invest too much in games tat are not mainstream. You can get a RTS or FPS with huge graphics and great ai, but can you get it for TBS? I doubt it. I am not sure you can convince a banker you'll sell less games but at a higher price. Your point may be valid, I haven't tried to create a game company, so I can't tell.

Gameplay! Gameplay! Gameplay!

imho

There have been way to many games that sacrificed the above to appease those who wanted fancy graphics.

That is not to say that having both, Great Graphics + a superior AI on top of Great Gameplay! wouldn't be great, but if the combination has to comes at the expense of the big one, Gameplay! then that cost is just not acceptable.

GalCiv is a good example afaic. Once you have seen the cut scenes a few times, the Spacebar becomes your friend because it is the Gameplay! underneath that is where it really is at...

If you know of a game that has all three, and is on a store shelf today, please speak up. It would be refreshing to play one.

[Message Edited]

[Message Edited]

Galactic Civilizations Colony Expansion Pack is what I would like to see.

In short it would add a Civ III type environment for each planet so micro managers can have a ball.

Ofcourse this feature could be turned of in the setup screen so us macros still want to play.

~SDC~

From what I understand that will never happen, because it was never designed to be like that in the first place. I think Frogboy actually said something along those lines not too long ago

Ive read the replies, thanks a lot everyone. I completly understand the balance of gameplay and graphics that comes with these types of games. I still think though, that if they could just do a 3d interactive colony view, that it would make the game 10x more immersive and enjoyable. I can honestly say that I would want to play more, if I could see a direct effect of my actions upon a colony. We could get a 3d view of a colony, with flying cars and tall buildings and whatever sci-fi things youd expect to see, and youd see the buildings you had decided to build. If its a propserous colony, you would see tons of lights and activity and general signs of a bustling economy. If it was a bad colony in a bad solar system, or didnt like its leaders or something like that, youd see a dim, shell of a city, with things breaking down and a general gloomyness. If your planet was being invaded, youd see meteors or whatever special attack type the invaders would throw at you, and then youd see the transport ships comming down, while your forces rally in defense, see a shootout, etc. I mean just think about it, it could be so cool!!

You mean like the colony view in MoO 2, except updated to suit the current hardware capabilities? Yeah, I'd like to have that too. But you can't have everything, even from large companies like EA and Activision.

Just briefly-

Games like GalCiv have 2D sprite based graphics because the market for turn based strategy games is somewhat smaller than that of say RTS's. As a result, we have to focus on making sure we work on as many machines as we can.

Now, in a GalCiv 2, we can assume a Geforce 2 level card will be the minimum (by 2005/2006) which means full 3D acceleration, textures, etc.  Though it would still be a 2D "board" so to speak.

Wombat and others,
Perhaps I am merely displaying my ignorance here. But don't you see a tradeoff between graphics and strategy in that a game with great graphics must therefore force the user's computer to devote more time to those graphics, and therefore the AI can't spend as much time thinking about its strategy? Or are the two processes independent?

AI is nothing more than strings of code in a file, all the processor has to do is run the code. The only thing that processor really needs to handle is the random rolling of the dice bit. Like if the other civs decide they should tell you about the culture palaces orbiting their planets or if they should say nothing and just lower relations. That and most of the graphics are handled by your video card, if you have a more recent one with 64 megs of RAM then you should do just fine.

vincible: hmmm...I dont know enough about computers to really answer that, and it wasn't in my mind when I was responding to LDiC's comments earlier. Yes, I guess fancier graphics mean that processing time is diverted, but I would have thought that it is not the same as the issue for FPS games, where the graphics are full screen and everything is in constant motion. Sure, game companies need to make sure their software runs on the "average" machine - but as Fulgrymm implies, modern machines with modern video cards can generally handle the load pretty well and divert the graphic handling to the video card, not the CPU. I mean, even with great graphics, strategy games aren't going to need to run like Unreal Tournament, are they? A number of players on this board do say they have problems running gigantic maps, and I think that's more the issue with processing grunt, not the "great graphics" per se. Strat games with gigantic maps, I presume, DO require lots of RAM and high level processors, which is a development issue.

Hmm, I don't think so. Shutdown/reload fixes that for a while, which is more like leaks etc. Also, running 1.3 G HZ processor with 384 mb memory.

I bow to the Suave One's superior computing knowledge! I agree if save/reload fixes the problem, it's in the code not in the hardware. I guess my experience has been with number crunching games - Championship Manager 4 is a good example: it simulates the results of hundreds of soccer games in the background and I reckon there is a direct relationship between the time taken to process and the CPU doing the work. Anyway - I'll stop wombatting on about stuff I dont know about!

Urgh... AI is 'just code' yes, but it can take a LOT of time if you make it well. Brad said that one reason for not computing the destination of transport every turn in accordance to enemy ships etc. was that it would take too much time. Plus the ai is threaded, which means most of it runs during your time. This means this time is not used to display/refresh the screen.
Now, the UI will usually benefit from the video card, but not that much unless you go for 3D.

As for a 64MB vid card, I will not buy a game that requires that before at least 3 years. I've got 32M on my laptop (I hate desktop computers - too big and awkward for a semi nomad). Since many strategy gamers have old computers, a nice UI is in fact a limiting factor. I know I won't buy a game with top of the art graphics now because it may run slower than a snail if I do get it. That's why I like demos: To see if the program can be run on my computer.

So the point is valid. Good ai requires a lot of CPU. Randomly rolling the dice is about 0% of the time spent in ai computing. Pathfinding for instance is perfectly predictable but quite expensive.

I can't help thinking the statement "AI is nothing more than strings of code in a file" is really misleading. You could describe each and every program that way. "Running the code" is what takes time. Rolling the dice is just a very fast function and just another bit of code. There's no reason to single that out.

Just a thought, how AI is GalCiv really? It appears to play the same strategy time & time and doesn't learn from my play (which very predictable!!! )

I was going to say something about that "nothing more than strings of code" comment, but LDiCesare's said it all already.

AI's in most game don't learn from the player. That's a helluva work. Galciv ai learnt from some things that Brad could get through the metaverse, but that's all.
So if by smart you mean 'able to learn', the ai in Galciv is not smart. Neither is the ai in any 4X game that I know. The best intelligence we can currently hope to code is that where the ai notices actions by the player and reacts to them, preferably well (like saying "don't build up", launching preemptive attacks, demanding starbases...).

In order to learn, you would have to be able to recognize strategies/tactics, which in itself is not really easy. Then you'd have to see that a pattern of succession of tactics is often made. This can be done inside one game, but over several games it would need some saving of the tactics to files, which hazs its own problems (see below). And then, the ai should try to react, until it finds a good way to counterattack. Then it could use this way to counterattack until you come up with a flaw in the counterattack and exploit it. Saving previous player strats to disk and looking for the player's moves in the database (or, maybe better, copying its tactics) will lead the ai to make a lot of mistakes because players will adapt much faster than the ai. For instance, the ai may put as much weight on your 3 firt games strategies as on the last game. If so, it will waste time considering inefficient newbye
strategies. If it doesn't, it may miss something actually clever.
I could go on forever (or quite a while anyway)...

Also, the big point is to recognize a threat. There are several cases of false threats reported: You cross a friend's sector with an army on its way to a common enemy and your friend protests, despite your never betraying him before. Is this intelligent or paranoid?
The ai must check for every possible threat in order to defend properly: Are those five speed7 constructors nearby threatening? No? Terror star. BOOM. Even if the ai recognised a terror star for what it is, would it know that constructors are dangerous without turning paranoid?
Note that offense looks easier than defense to me, because defense must take into account cheese tactics while offense will usually not (though checking for undefended targets is something you must think of when coding an ai).

Overall, galciv ai is quite clever, but too hard coded (tech tree) and thus predictable, and a bit poor on the tactical scale, as it doesn't know which unit is best to build (AMMs), and doesn't adapt to mods done to existing ships. For instance, it can do nothing with a constructor + a -9 range colony ship. It would require lots of thinking in order to do:
1)My colony ship needs a yellow star to be useful.
2)There are none in range. Solution?
3)Increase range.
4)How? Tech? Won't increase enough. Constructors can be used to increase range too.
5)Send constructor to target sector with uninhabited yellow star...

Not an easy task when there are many such situations in the game to consider...
[Message Edited]

Interesting comments LDiCesare. I appreciate, at least as a novice AI person, the difficulty/complexity of the things that you mention.

(Puts on flame-proff suit, fire-retardant gloves, crouches behind blast shelter.....)

For me though, calling something advanced AI, implies changes/learning each time it is in action. I haven't seen that with GalCiv. Yes it has very good balancing mechanisms, a decent evaluation model & some knowledge of threats etc etc etc..
Yes I appreciate the $$/time/graphics/complexity trade-off so I don't think Stardock can actually be all things to all people.

Maybe the game engine could have more alternative plans for situations and employ some form of a goal oriented approach with alternative plans tried when some approach fails.

For example, at the moment if you build a culture threatening starbase then the game, when it recognises the threat and decides to react thorugh some simple thresholds, (basically) says "don't or we will be at war", if you refuse then war ensures (always!!!) Maybe the game could then try something different when the first strategy fails, I don't know exactly, but maybe trade for the base to remove threat?


Similar approaches have been used for robots and agent-oriented AI approaches, with certainly some impressive results.

Anyway just a thought.

I'd like to see a game that does this, too, but don't think there is any.
Robots and agent-ai usually tackle tasks far easier than strategy. The big problem in any strategy is that you have to be able to think of something if you want to be able to thwart it or do it. Sid Meier obviously never thought that One City Challenge was possible in civ, thus the game didn't try to fight it, when the ai could have realised that the player was building a spaceship.
The ai can work with goals, but even if they model it like this internally, it doesn't mean there will be the goal that you want to see. The point is that any good ai will be able to adapt itself up to a point, but then when you find a hole in its defense, it menas it's unable to learn from that particular hole, and needs rewriting in order to correct the flaw. And humans will always find holes if there are enough rules/elements to play with (complexity).

Wouldn't it be possible to implement different "goals" for the AI's - Say one path for each victory condition.

If each AI more or less randomly chooses their "goal" by the beginning of the game, it would make it pretty hard for a human player to follow the same strategy each time.

I do realize that this would probably take a lot of work to code.

Okay, assuming deciding between strategies is too difficult, there are some things the AI could do to make it better decisions.

1. There is rarely a good reason to give a resource starbase. Unless the resource starbase is in a sector with an opponent's planet - they shouldn't be traded.
2. The should prioritize developing starbases, particularly resource starbases.
3. The most popular stategy is culture bombing with no military. Stardock should realize this.


As a fairly easy AI strategy, the AI should check for three things:
1. Is the the player running away influence?
2. Is the player building starbases with cultural modules?
3. Is the player significantly weaker in military?
If answer to all three is yes, then I'm being cultural bombed -- respond by attacking opponent's starbases.
If the answer to first two is yes, and the third question is no - respond by building cultural starbases in inhabited systems.

This additional check (along with the AI putting value on resource development) would make the game significantly more competitive - and shouldn't be two difficult programatically. I'm sure there is more than enough metaverse data to determine balanced criteria for all three questions.

-Cypher
~SDC~
[Message Edited]

One more thought -- the AIs should build AMMs. I can build a paper tiger military in the game by building a few AMMS - while they're building Rangers and Avatars and I look stronger... adding AMMS in the mix would help a lot and would be useful when the AI decided it was under culture attack...

~SDC~