GDC 2011: Strategy Games - the next move

[Me,myself,and,I]

Wow engine is even not that good and starts to be severly outdated. SC2 contents are very limited compared to SC1.


I'm not too sure about this, make a crappy Civ5 which sells a lot, you won't be able to sell Civ6. That's just narrow minded thinking, you're betraying customer's trust.

In response to the first quote. Where do you get that idea?

And second. You may lose half the the fan base. but remember you also GAIN a significant number of people who are new to strategy games.

 

[Drawmeus]

Prove this statement.

Don't point to Steam saying it's been a #1 played game for a long time. There are lots of sales for very cheap. Prove that Civ V is financially a 'slam dunk' with actual numbers. Even better, numbers that compare profits to expenses. That would really show whether it was a 'slam dunk' or not.

You can't. No sales numbers have been released. Profits/Loss will probably never be released, I've don't recall ever seeing a game do it anyway. Therefore, you can only guess. Perhaps a correct guess, perhaps not, but it's still a guess. So you shouldn't make a statement like this. Even milestone sales numbers haven't been released, which a publisher is typically happy to announce.

Even VGchartz can only guess, and has no direct dl Steam numbers as far as I know, so you can't even use them as a source. Civ V may or may not have sold a lot of games and/or made a lot of money (the 2 aren't necessarily tied, because pricing can fluctuate based on things like weekend discount sales), but the point is that you DO NOT KNOW. So don't admonish people to 'remember' something that you can't prove.

No. You can feel free to not accept my statement, in which case we have nothing left to talk about. Given the data available, I find it vastly more likely that Civ V made good money than not, to the point that thinking they DIDN'T make good money sounds more like wishful belief in a just universe than anything grounded in reality.

 

[Becephalus]

Also, the AI in civ4 actually got bonuses that helped it compensate for its lack of skill. (If a stack is big enough, then it doesn't matter so much that it has bad unit composition, and bad attacking order). In civ5, more units does not compensate for bad unit positioning. (In fact, AI unit positioning may even get worse with more units.) Such a compensation should have been included in the form of a combat bonus for the AI as difficulty is increased. (Instead of the current bonus to unit supply and unit production.)

Exactly. This couldn't be said enough.

 

[builer680]

Remember that Civ V was financially a slam dunk.

Prove this statement.

No.

That's what I thought.

 

[Smote]

I can easily see that it is not a great AI that attracts players to games like Civ. Its the flash, and the sense of building your own empire with your own style, and making interesting decisions with tons of options. This is what consumers look for, not a "perfect AI."

In fact, people frequently dislike good AIs. Many gamers want to play games as an easy escape from life, and want to win, with enough difficulty that they feel a sense of accomplishment. If a good AI beats 90% of players, people will not like the game, because it has too high of a learning curve, and they need to look up strategies and such.

There are really just two options for making a perfect game.

1. Completely different AIs, based on difficulty, with far more programming and skill being given to higher difficulty AIs. This would perhaps require a 10 man AI team, with a 3 month tournament deciding the top AI. I find tournaments provide participants with enough motivation to work their asses off. Solutions could be integrated at the end, to get even better. Another option would be a large-scale public AI programming tournament with cash prizes or something. A university tried one of these for Starcraft, with pretty good results. This is necessary to create top-level AI.

2. A great multiplayer experience. This is the option where businesses say "Creating an AI to challenge skilled and experienced players is too hard." The alternative, is simply the beautiful option of letting those great players play each other. However, the business must make the game balanced enough so that players feel that they can innovate their strategies in new directions, or else they will lose interest. They also must make the multiplayer work , which Civ has had difficulty doing, though I have recently had good results in my online games.

In order to create balance, the only real option is to let dedicated players play-test the game and nerf dominant strategies and buff strategies that are weaker than intended. The goal is to create as many good and distinct strategies as possible, with no single or couple strategies being universally dominant.

Starcraft 2 did a good job with this, by having professional Starcraft 1 players dedicated to finding every exploit and winning strategy, and getting it all equalized.

It takes a lot of work to create a game with lasting appeal. For challenging play after learning all the rules and experiencing many strategies, the options are:
Great AI (hard!)
Competition between players

 

[Calouste]

Well, Civ 5 not only has bad AI, but is full of bugs and has an awful game engine. You would think that the game would be faster than Civ1, since computer power improved by 1000x, but it's in fact worse than ever.

Apparently, they didn't have enough money to create a decent code for calculating worker movements (which Shafer probably thinks is the same as AI level), and the difference between the graphics engine and AI computation is not clear, we don't know which of the two make the game so laggy (probably both).

I'm also not buying any argument 'people want shiny things the most, screw the rest'. You just need to see which game company is the most successful atm : Blizzard. Wow and SC2 are no way the most beautiful games, they are just finished games with impeccable mechanics, very few bugs, and challenging content.

Wow engine is even not that good and starts to be severly outdated. SC2 contents are very limited compared to SC1. True, Wow is becoming more and more 'add new buggy and unbalanced features', and I hope they won't follow this line, but they are great games for the moment (as diablo 3 will be). The only great fail of Blizzard is battlenet 2.0, where they wanted to restrict players too much (communism did you say ?).

And I would conclude that people play strategy games to meet strategy challenges, not to watch new shiny graphics, or they would play Sims 3.


I'm not too sure about this, make a crappy Civ5 which sells a lot, you won't be able to sell Civ6. That's just narrow minded thinking, you're betraying customer's trust.

You can't really compare anything Blizzard does to other game studios (except maybe Valve). WoW has a revenue of around $100 million/month in subscriptions, that's probably Firaxis budget for a whole decade. With that kind of money, and no pressure from external investors, you can craft games to perfection.

 

[MadmanAtW]

In fact, people frequently dislike good AIs. Many gamers want to play games as an easy escape from life, and want to win, with enough difficulty that they feel a sense of accomplishment. If a good AI beats 90% of players, people will not like the game, because it has too high of a learning curve, and they need to look up strategies and such.

You are unfortunately combining several meanings for "good" AI. People want good AI in that they want it to be well crafted and do what it is meant to do correctly. People want its _competitiveness_ to be limited. People want an AI that knows the rules of the game, that doesn't shoot its own teammates in the back, that allows the governor AI to make good suggestions about what to do with your city if you ask, that doesn't pathfind off cliffs, etc, to mix genres a bit.

I don't want my AI to be so "good" that it beats me every time I play without sweat unless I'm on Deity level; but I do want the AI to be high quality enough that I'm not going "wait, you _do_ know the rules of the game, right?"

 

[Zogar]

In response to the first quote. Where do you get that idea?

About what ? WoW engine doesn't support crossfire for instance, and I was speaking about new contents of SC2 compared to old one.

And second. You may lose half the the fan base. but remember you also GAIN a significant number of people who are new to strategy games.

Maybe. Still a bugged game with poor engine is not prone to attract so many players. The only good positive thing for Firaxis is that there's not any competition in their field.

You can't really compare anything Blizzard does to other game studios (except maybe Valve). WoW has a revenue of around $100 million/month in subscriptions, that's probably Firaxis budget for a whole decade. With that kind of money, and no pressure from external investors, you can craft games to perfection.

Sure they don't have the same budget. But you don't need a gigantic budget to make a fluid engine and few bugs, it has been proven again and again. Well for bugs it takes time, so money, yes, even if good coding make the possibility of bugs fewer and easier to fix.

Worker management is most relevant anyways, I don't know what they were thinking, but that's plain bad coding from the start. Any decent coder wouldn't have made the game recalculate every possibility for every worker, that's just horrible from an algorithm point of view, not even speaking of the 'line of sight bug'.

 

[bruntfca]

He tries to defend his statement in a post, blaming it on the company:

"To clarify... I didn't disagree with Tom, and it's not that I don't want to see better AI in strategy games. It's just that as long as a company's goal is to make the most money possible, it's very rarely financially sound to pour a LOT of time or money into AI. If I had all the cash in the world and was making games only for fun it would be one of the biggest areas I'd focus on - but it wouldn't be because I thought I'd sell more copies that way."

The guy shows what an idiot he is, and a poor businessman at that. So he kills the franchise and alienates his core audience in one fell swoop. Why bother buying Civ 6 if all the Dev time's gonna be spent in yet more "fractal mountains" or other crap.

There's nothing wrong in looking for more sales, but alienating you core audience is dumb.

The AI is not only poor, it has no idea how to play it's own rules, it's therefore even hard to put the title "game" on Civ5.

 

[bruntfca]

Are you sure? Remember that Civ V was financially a slam dunk.

Honestly? Shafer is right, in the context he's talking about. Given finite resources and the mandate to make as much money as possible, a strong AI is a waste of money.

Given the mandate to make the best game possible for relatively hardcore players (say, the kind of players who end up on a website talking about the game), it is definitely NOT a waste of money. But, well, we see how important it is financially... Civ V was a financial success.

Basic sales 101, you can only sell a crap product once. Let's see how well Civ 6 sells, if it ever gets made even.

As for making a decent AI costing 10 times more as somone above posted thats utter BS. It's plain to see looking at the game that Firaxis programmers are just not that good. Many games based on Hexes and 1UPT have been developed.

Algorithms to deal with the traveling salesman problem have been around for a long time, as have many min/max decision tree algorithms. On top of this a whole new way of doing AI but using object orientated techniques, by placing AI decisions *inside* the units. This avoids totally the issue of computational intractability of various "top down" algorithmic approaches.

This technique has been used to provide emergent AI for 1000s of units in games like "AI Wars". Brad over at star dock also used this technique combined with multi threading for the 100s of ships in Gal CIv.

I know dozens of decent programmers still in University who could put together a decent AI, and would be happy to work for even $50,000!

As for the lead designer, coming from the "Mod community", does he have any formal training in Computer Science at all? Can he even write a line of C++ code or assembly?, or does he rely on slow bloated script engines, which make the Civ5 game the bloated,slow POS it is?

 

[Aeronomer]

This is all complete crap. I know us "haters" and "flamers" are no longer welcome here, but the release of the new patch caused me to crawl out from under my rock.

The same people who have been defending Civ 5 are now defending this reprehensible philosophy.

Civ 5 sucks as a game. It's even worse than Civ 3. I pre-ordered this game based on my past experience. I will probably never play this game again and I am going to be VERY reluctant to spend my hard-earned cash on a Firaxis title in the future. It's obvious that the decision to release this broken piece of crap was made by some corporate bonehead and not by someone interested in continuing the Civ legacy.

So they have lost my "loyalty". Screw me once, shame on me...etc. Maybe they've attracted enough "new" players to make up for the people like me that they've lost, but how "loyal" are these noobs going to be?

I feel abused and ripped off. I will adjust my expectations and behavior accordingly in the future. What are the financial ramifications of that?

 

[Brian Shanahan]

Are you sure? Remember that Civ V was financially a slam dunk.

And that is a testament to Civ 4 not 5. People bought 5 because of two things, 4 was great, and the hype surrounding 5 coming from all angles, not because of any inherent qualities in the game. The real test of 5's financial prowess is a) sustained sales (which we don't have, but I'm willing to bet aren't as good as 4's by a long shot), b) expansion/GOTY/complete version sales in a couple of years and c) Civ 6 sales. So we won't see the true financial effect on the company for a while yet.

Honestly? Shafer is right, in the context he's talking about. Given finite resources and the mandate to make as much money as possible, a strong AI is a waste of money.

Why? Frankly game companies are going down the route of Hollywood, for graphics think CGI, for "appealing to casuals" think stupid plot lines, and for Ai think writing. Now like films all the money is going into the shiny graphics to the complete detriment of the overall quality of the game (or film). And signs are beyond that like the film industry now thinking in terms of opening weekends for both cinema and DVD releases, the games industry is increasingly going for the early sales and DLC routes. They do not think long term, either with the game or possible sequels, only looking at the bottom line for the next quarterly report where they'll get their fat bonuses, even if their strategies have killed the company long term.

I honestly think that there will be a big contraction in the gaming industry soon unless there is a huge rethink in the way companies think about what they're doing, i.e. go for quality over shiny.

Given the mandate to make the best game possible for relatively hardcore players (say, the kind of players who end up on a website talking about the game), it is definitely NOT a waste of money. But, well, we see how important it is financially... Civ V was a financial success.

Again Civ 5 was a success mostly because of the (deserved) reputation of 4, and had nothing to do with it's own merits (and I'd be saying this even if the game were good), which may I remind you was a game designed far more with the hardcore gamer in mind.

Though frankly I don't get this hard-core/casual gamer dichotomy (except for the very young and old maybe). Most people if given a good game which is well balanced and propmts them to keep going will keep playing the game no matter whether it is a side scroller as simple in ideas as Super Mario or as complicated and dense as a spreadsheet game like Football Manager. They will not care about the hard-core/casual tags, just like I didn't when I picked up Civ 3 for the first time, they will just want to play the game to get slightly better (every time) and for their own enjoyment.

 

[Becephalus]

Basic sales 101, you can only sell a crap product once. WRONG! Let's see how well Civ 6 sells, if it ever gets made even.

As for making a decent AI costing 10 times more as somone above posted thats utter BS. It's plain to see looking at the game that Firaxis programmers are just not that good. Many games based on Hexes and 1UPT have been developed. And they ALL have crap AI!

Algorithms to deal with the traveling salesman problem have been around for a long time, as have many min/max decision tree algorithms. AND they are poor or computationally very intesiveOn top of this a whole new way of doing AI but using object orientated techniques, by placing AI decisions *inside* the units. This avoids totally the issue of computational intractability of various "top down" algorithmic approaches.

This technique has been used to provide emergent AI for 1000s of units in games like "AI Wars". Brad over at star dock also used this technique combined with multi threading for the 100s of ships in Gal CIv. The AI Wars and GALciv AIs are not really any better, the Galciv AI is easily just as horrible

I know dozens of decent programmers still in University who could put together a decent AI, and would be happy to work for even $50,000! I bet...

As for the lead designer, coming from the "Mod community", does he have any formal training in Computer Science at all? Can he even write a line of C++ code or assembly?, or does he rely on slow bloated script engines, which make the Civ5 game the bloated,slow POS it is? All software grows to the resources availible to run it

Almost everything said in this post is wrong. If you and you friends can really make some AI significantly better than the Civ AI, do so and you WILL be able to sell it. But I am sure once you got down to implementing it, it would be 100 times harder than you think.

I helped make an AI for a game people thought was awesome. All it did was maximize kills. It was super aggressive. Because of this it was very easy to beat once figured out what it was doing, and against experienced players it always needed a huge advantage in material to win. Hell against the best players it couldn't win regardless of how much of an advantage you gave it.

 

[bruntfca]

Almost everything said in this post is wrong. If you and you friends can really make some AI significantly better than the Civ AI, do so and you WILL be able to sell it. But I am sure once you got down to implementing it, it would be 100 times harder than you think.

I helped make an AI for a game people thought was awesome. All it did was maximize kills. It was super aggressive. Because of this it was very easy to beat once figured out what it was doing, and against experienced players it always needed a huge advantage in material to win. Hell against the best players it couldn't win regardless of how much of an advantage you gave it.

I'm not sure if it's worth replying to your post, but here goes...

"And they ALL have crap AI!"

Is "crap AI" a technical term?, Nice value judgment that you just pulled out of your hat. Care to actually qualify your statement?

"Poor and computationally intensive"? Thats why you use algorithms to cut down on the search space as I said or is reading comprehension not one of your strong points? I never saw a computer using an algorithm to cut down search space beat a human at chess and other games....no wait.

"The AI Wars and Galciv AI are not really any better". I don't see the forums of those games full of posts about how totally lame the AI is. Face it the AI in Civ5 does not even appear to know how to play it's own game. Galciv may "not be any better" as you say, but a hell of a lot of people think otherwise.

"All software grows to the resources available to run it". And your point is? You end your comments with a meaningless tautology.

 

[XeroK00L]

I don't need any of the pretty graphics crap! I play CIV5 in tactical mode and tactical mode only. Those 3D graphics only serve to distract me. I don't care if the game looks like CIV1. For a strategy game the focus should be put into gameplay and AI, period!

To me a Diety mode should mean a really, really clever AI playing on the same ground as me, not a really, really dumb AI playing with ridiculous handicap settings. Look at how chess games are implemented--higher difficulty means the AI thinks harder and deeper, not that it gets two queens!

My suggestion to them if they can't afford spending time in developing a good AI would be to at least make finding online opponents easier, and make play-by-email an option. For a game that can take days or even weeks to finish, play-by-email is a must! Or even better yet, allow game states to be saved on a central server so that people can drop off and pick up where they leave off with ease. Take a look at how EA implements Scrabble on Facebook. It's the best way to implement a multiplayer turn-based strategy game I've ever seen.

 

[Calouste]

As for the lead designer, coming from the "Mod community", does he have any formal training in Computer Science at all? Can he even write a line of C++ code or assembly?, or does he rely on slow bloated script engines, which make the Civ5 game the bloated,slow POS it is?

Designers at game studios don't write code, they write feature specs.

 

[Calouste]

Sure they don't have the same budget. But you don't need a gigantic budget to make a fluid engine and few bugs, it has been proven again and again. Well for bugs it takes time, so money, yes, even if good coding make the possibility of bugs fewer and easier to fix.

Worker management is most relevant anyways, I don't know what they were thinking, but that's plain bad coding from the start. Any decent coder wouldn't have made the game recalculate every possibility for every worker, that's just horrible from an algorithm point of view, not even speaking of the 'line of sight bug'.

They couldn't code that in Civ4 either (trade route recalculations).

 

[Becephalus]

I'm not sure if it's worth replying to your post, but here goes...

"And they ALL have crap AI!"

Is "crap AI" a technical term?, Nice value judgment that you just pulled out of your hat. Care to actually qualify your statement? Ok to be technical. I have NEVER seen a 1upt AI, a hexed based AI, or for that matter an RTS AI that didn't need to rely on set scenario AND massive handicaps to be remotely competitive. Never once. Sure for Chess or checkers and super simple games, but even go is beyond an AI's ability. I have played Gal Civ a ton (which people think has a decent AI), I have helped develop a few games, and have worked on AI improvement mods. The complaints I see about civ5 AI are the exact same complaints you see about other games AIs. It is of similar quality. The game as a whole is just worse and it has a rabid fan base so people are less tolerant of the crap AI.

"Poor and computationally intensive"? Thats why you use algorithms to cut down on the search space as I said or is reading comprehension not one of your strong points? I never saw a computer using an algorithm to cut down search space beat a human at chess and other games....no wait. "I am sorry Firaxis isn't perfect for you. It must be nice living and working in a world where everyone has time to devote cutting edge solutions to every problem in every product they design. Back here in the real world people make do, there are budgets, there are deadlines.

"The AI Wars and Galciv AI are not really any better". I don't see the forums of those games full of posts about how totally lame the AI is. Face it the AI in Civ5 does not even appear to know how to play it's own game. Galciv may "not be any better" as you say, but a hell of a lot of people think otherwise. AI Wars had no fan base, so no expectations, it also uses MASSIVE AI handicaps, so it is not really comparable. I played a lot of GalCiv II. The AI is ok, but it is not really that different from the Civ 5 AI and once again uses massive handicaps. At even settings you can beat it making terrible terrible decisions. For instance the AI never even learned how to use the planetary focus mechanism, even after what 2 expansions? (I cannot remember if they fixed that in the last one). Yet that mechanism is at the core of good play. And the AI doesn't even know about it! It would be like if the Civ 5 AI didn't know about changing its cities focus. There were all sorts of things the AI didn't even attempt in GalCiv.

"All software grows to the resources available to run it". And your point is? You end your comments with a meaningless tautology. Its not meaningless, you were whining about the software being fat. The fact that almost all software is fat because humans are lazy is an important point when considering that claim. You criticize Civ 5 for making the same mistakes many AAA titles make. CA/Sega releases games with horrible AI and 5 minute turn waits all the time. What were you expecting? People love paradox games including me, but most of their games are barely playable for the first year after you know a little bit about them.

Don't get me wrong I see Civ 5 as a hastily released and somewhat incomplete failure, but a ton of the criticism of it is misplaced and just nitpicking because people are disappointed the game isn't more fun.

The number of really good games released each year is only a half dozen. SO lets all not pretend we are shocked when Civ 5 isn't on that list. Civ 3 certainly wasn't as released either. Hell I bought a "Play the world" expansion for that game that didn't have functioning multiplayer. You think people are mad now?

People on this forum generally love SMAC, and the AI in that game was way worse than the AI in civ 5.

 

[alexman]

Designers at game studios don't write code, they write feature specs.

While this is true for many studios, it is not true at Firaxis.

 

[MAntoninus]

Think outside the box.

Go after Grad Students in AI. Lure one or two in to do their doctoral thesis with Firaxis and model all this.

Great experience, great dev value, good branding in the thesis and ample public testing of the ideas...

That's what I'd do.