Slashdot Mirror
Can We Create Fun Games Automatically?
togelius writes "What makes games fun? Some (e.g. Raph Koster) claim that fun is learning — fun games are those which are easy to learn, but hard to master, with a long and smooth learning curve. I think we can create fun game rules automatically through measuring their learnability. In a recent experiment, we do this using evolutionary computation, and create some simple Pacman-like new games completely without human intervention! Perhaps this has a future in game design? The academic paper (PDF) is available as well."
You want your creativity to be automated?
You desserve what you'll get, welcome to your dump...
Hope the rest of the world will leave you there, for once.
Smile, don't click...
Can We Create Fun Games Automatically?
Sure we can, depending on your definition of the words 'Fun', 'Game' and 'Automatically'.
:P
Site & blog: http://www.mayaposch.com
The more I play games, both video games, board games or pen and paper RPGs, the more I see the obvious patterns that exist beneath them.
I stopped playing new boardgames as all these become obvious after a few games, and if you tend to like one, old games already implement them perfectly. You basically have 3 (arguably 4) components in any board game : randomness (go play dices if you like it), tactical planning (go play chess), bluffing (go play poker) and, arguably, negotiation that can be seen as a merge between tactics and planning but that often use a whole different range of social skills.
Video games have also some recurring ingredients. I played less of them so I fail to see them more clearly, but some of them are obvious :
- a sentiment of progression. Whether artificial (through leveling in RPG games) or real (from FPS where you get better at shooting, rocket jumping, etc...)
- hidden content of the game, that the player has to find or guess. It is usually some content voluntarily put there by the game developer (quests, levels, maps) some hidden game logic that one must understand (AIs behavior, puzzles, research trees). In the most interesting games (in my humble opinion) there is also content that is almost emergent. The creator only loosely coded some rules and it is the player's actions that create his own problems to solve. It often happens in strategic or development games, where you discover that a design you chose had some vulnerabilities and that by correcting this, you create a whole bunch of new problems.
That one last part is the most difficult to reproduce automatically, in my opinion. But a lot of successful games don't have any such emergent content, so I guess that automated games generation can prove quite fruitful !
The Wise adapts himself to the world. The Fool adapts the world to himself. Therefore, all progress depends on the Fool.
Wasn't there a short story about a computer that created and told stories?
Maybe by Lem?
Not a happy ending. (Ha! My awful memory won't protect you from spoilers!)
Can we get research grant funding automatically ?
I believe the answer is yes.
1. Choose a 25 year old topic (for example, a Pacmangame), reinvent it using lots of buzzwords such as swarm, hive, collective, competitive, but secretly just program a system using some generic rules, and a gradient descent algorithm that will force those generic rules to conform to the behaviour we wanted in the first place. Then publish a PDF (why oh why by the way is PDF proprietary format ANY better than Microsoft's proprietary format ?), and spam it across tech news sites.
2. Make some wild claim that this is the dawning of the age of Aquarius (or similar).
3. ???
4. Profit !
I've been reading too much slashdot lately. I saw the title and immediately my brain said 'automagically.'
ugh.
"You have to make your own fun. If somebody else makes it for you, then it's entertainment."
Fun is learning â" fun games are those which are easy to learn, but hard to master, with a long and smooth learning curve.
That's true, and it's why Go is the most fun of all games.
A random maze generator? That's hardly what I'd call "automatically making a game".
Perhaps this has a future in game design
Uhmm, no ? We need versatility in the games we play, and a complexity that such algorithms can't introduce automatically. Sure, such methods could help in creating some (!) game rules, anything more is beyond speculation.
I am putting myself to the fullest possible use, which is all I can think that any conscious entity can ever hope to do.
Good point. Different levels of "fun" and satisfaction.
Someone wrote about putting Age of Empires 2 on showroom PCs, and all the female customers went ga-ga over this game. They would then build mini cities and so on ... all without fighting. He said they wouldn't give a second look at AoE 3, or The Sims 2 ... they just wanted to play AoE 2.
Someone wrote about his entire family playing mostly older games (including all Mario games), and mostly avoiding newer, copy-protected games.
It amazes me reading these posts.
Just because a fun game has X learning curve doesn't mean games with X learning curve are fun. The learning curve maintains attention, necessary for the game to be fun. The same learning curve in another situation may maintain attention to something droll. And something fun may have no learning curve at all. I suggest you're not looking at fun, you're looking at ability to maintain engagement. I also suggest fun does not have a single definition, or else everyone would play the same game.
"I may be synthetic, but I'm not stupid." -- Bishop 341-B
PDF has been opened. Admittedly the standards body which supports it is ISO, but I don't think anyone bribed them to approve it.
Sounds like the mind test games from Ender - it probes you and comes back with shit to fuck with your head.
"If everybody is thinking alike, somebody isn't thinking" - Gen. George S. Patton
...the FUNGAME MANIFESTO v1.1!! All the Science behind every fun game there ever was, distilled in a nice, just mix-with-water, logic! 101 best practices for every game developer idiot and game developer dummy!...fun games guaran-teeed!!! ...then all the games get fun. Then it all gets boring. ...then one breaks out from the mold. And that's different. And it's fun. Fun is now redefined! Wowee!... ...And so the cycle continues..
It has already been done unless you have some proof that Star Wars Galaxies actually had some thought behind it.
The nice thing about this article is that Raph is out spending his time on worthless stuff instead of sending his time creating a high profile game. He is really good at writing papers on how stuff should work but he cannot implement his own ideas.
Seeing movies produced by following the "formula", do you want automated games? Do you even want a "formula" for "fun" game design?
Maybe its possible, but this starts to sound like automated art.
Think Deeply.
If programming and design could be done automatically, we wouldn't still have programmers. We can't even manage to automate creating simple apps. How could we possibly automate creating entire new games, which means new art, new rules, new everything.
On top of that, everyone finds something different in a game to be 'fun'. Some love challenge, some love adventure, some love collecting things... Attempts to make games that have everything anyone could love are usually pathetic flops.
"If you make people think they're thinking, they'll love you; But if you really make them think, they'll hate you." - DM
I can't think of a graphical OS that won't display a PDF.
Humans will always, or least in my lifetime, have a better grasp of fun. Meta-games like Little Big Planet for the PS3 give the even "newbies" the ability to build their own games with their own rules.
Walking into a computer lab at school, spying a mystified user staring at a screen. Investigating further, it turned out he was confused by the fact that
Make Game
Racing Game
2 tracks
In a programming IDE did not yield anything.
I record my sleeptalking
... will he eventually reach nethack? After all, it's also basically things moving around (you, your pet, monsters, etc.), and actions on getting together (fighting with monsters, picking up things, ...). Ok, it has a third dimension (stair up/down) and a few other degrees of freedom (e.g. additional actions like casting spells or putting on/removing rings) and a few more parameters than just one score, but I'm sure you can generalize the description enough to be able to describe nethack.
But somehow I doubt that he'll create a new nethack this way.
The Tao of math: The numbers you can count are not the real numbers.
Sure, if you define "fun" as "a smooth learning curve", then you can make fun games automatically.
But not all of the fun is in the learning. Some fun is in tweaking humor. Some fun is in triggering a person's likes and dislikes (Nethack, ponies). Some fun is created by changing the venue (is it a space game? a historical shoot-em-up? A politics game?
Yes, there are underlying patterns to a lot of games. But simply limiting our definition of "fun" to "learning" and "follows the pattern" reminds me of the automatic novel generations in Orwell's 1984.
I don't think that this headline defined the problem well. Yes, some parts of fun can be automatically generated. But no, to make a fun game, it has to be interesting to a human, not just to a turing machine. And for that, you really need other humans to make the games, or you don't have the depth required for real "fun".
Correct Horse Battery Staple: 72 bits of entropy. Enter "Correct H" into google. When it generates the phrase, that's
If creation of generated research papers papers can be considered fun, then by all means!
Rickrolling can play a positive role in convincing proponents of internet monitoring to change sides. Anyone can trick anyone else into getting caught in the trawl. Just change the target material from Rick to something less innocent.
She made the willows dance
fun games are those which are easy to learn, but hard to master, with a long and smooth learning curve.
Best example for this is Chess. Easy to learn but takes many years to master.
Abstract concepts generally have the best game rules because they offer the widest set of possible options; game designers are not restricted by the constraints of 'the familiar', but players then face the initial bump in the learning curve in order to absorb the initial concept. The problem here is that algorithms cannot draw upon the vast life experience and everyday cultural references that ordinary people possess. Further, abstract games, no matter how well-designed, are difficult to market and sell (take The Sentinel as an example); they need someone else (reviewer) to say, "This is good, go buy it".
create some simple Pacman-like new games completely without human intervention!
Man creates Pacman. Computer clones pacman. Has a new game really been created?
Linux is a UNIX-based OS, and my favourite distro of it is Macintosh OS X, because it has the full support of a major corporation, unlike other distros, which are all operated by maybe one or two nerds in a basement together. The last thing I want is for a level 5 dwarf (haha) providing me my OS
One factor which I've noticed tends to create addictive puzzlers is quite simplicity of form. The resulting purity of function lends itself perfectly to entrancing, mesmerizing marathon sessions of blocks dropping, diamond spinning or whatnot, always seeking "one more combo!" as the points rack up on top of the screen. Tetris, Lumines, Bejeweled, the list goes on. Keep the concept simple, the list of controls short and the rules easy to learn. If I looked up the amount of time I spent trying to line up that four-block line in a perfect spot for maximum points, I'm pretty sure the number would be terrifyingly high.
Automatically? Most game dev studios can't even make fun games manually!
Novell is one or two nerds in a basement together?
The Tao of math: The numbers you can count are not the real numbers.
Hopefully Raph had fun leaning from all the failings of Star Wars Galaxies. Me, not so much fun playing those failings.
This reminds me of the first 10 minutes of the movie Dead Poet's Society where Robin Williams has them rip out the introduction to their poetry book.
Not very new, since Sudoku became popular when Wayne Gould wrote a program to generate puzzles, graded by difficulty.
I guess most of the paper puzzles can be generated this way (like crosswords...).
The difficulty largely lies into the entertainment's metrics.
The authors seem to have used humans to test their games, so I doubt that creating a game from scratch could be done entirely automatically.
Let me google that for you
Oh, I can't help quoting you because everything that you said rings true
These researchers start off asking the question, can fun games be created automatically? They end saying that they have proved the idea, but it remains to verify that the more fit games are actually more fun. Uh, wasn't that what the research was supposed to show?
Evolutionary programming can be deceptive. Humans define the rules, humans tweak them until what looks like the desired output is produced. One can easily be deceived into thinking that the product is magically as good as it can be, or has magically used an algorithm that humans would have thought of. If we don't understand the nuances of the rules or the fitness function, the product can be quite different to what one would expect.
The real challenge is coming up with rules and "training" detailed enough to iron out all the wrinkles. This research tackled none of that.
They appear to have some sort of formula for spitting out games year after year with little variation. It seems to be working well for them. Although truth be told most gamers hate it.
having played swg, one of koster's grand designs, i would never touch another game he was affiliated with. his concept of fun is grindage / time sink / drag it out as long as possible. the only person i dislike more than him are the ea ppl that killed mco.
koster is a broken tool
-.no
In photography, you set up the boundary conditions, take a TON of pictures, then select the best ones from the ones you have. The best photographers have the best eye for selecting the remarkable ones out of the pack. This would shift game programming from an art like classical sculpture - where you have to plan far, far ahead, and don't get second chances - to an art like photography where it is more about creative curation than creative engineering. Evoluationary development of games wouldn't eliminate the creativity of the process or the product, it would change the creativity of the process and the product.
Umm, Redhat?
"Can We Create Fun Games Automatically?"
Sure if you want a boring game. So lets see what are the common aspects that help a game along and can an AI game writing system create the environment that usually aids in creating a good game?
Here is what is required based on the one truth mentioned in the story regarding 'easy to play hard to master' I will use online FPS games as the example because that is mostly where my head is at. But my comments also apply to most other games and for single player with a little tweaking.
The basics of play should be simple, allowing some success even with the basic weapons and basic movement skills. This is very important. To win requires to master both weapons and movements. For other games it is less acquired but allows advancing to a degree and not a dead stop in the game. Avoid at all costs where it requires going back significantly in order to acquire what is needed. Fortunately fps games rarely require this even in single player unless it is a long level and it must be started from the beginning. This is greatly assisted and avoided if saved games has a variable rollback feature.
Along with this, is the ability to replay a segment where you see your mistakes and even have them pointed out to you by the game visually or spoken. ie: You missed an item(s). You should have used this weapon for that distance or situation etc. Live tip manual system of sorts.
The environment must supply a few rudimentary visual and emotional rewards. In other words. There must be a certain level of eye candy that attracts the attention of the player. Actions of the player generate some reward such as a visual or sound and object reward beyond simple item reward. This should be removable for the skilled player. IE: Player ability to remove fancy explosions that affect frame rates and dial down sound effects that can be annoying over time.
Direct player to player interactions. The more a player can interact with friends the better. This includes such things as a text or voice channel. It is important in a game to be able to express your and hear the emotions of others.
Puzzles should be scalable and non linear. Basically to be able to set level of difficulty and with multiple outcomes and solutions. Do I need to explain why? Replay value is one reason.
Optional: A system that matches up equally skilled players and or a handicap system.
Interesting immersive world. This is the creative component that no machine can generate and why we still require real coders and artists bums in seats. So many games these days do not offer enough variation from game version to version. I am convinced this is due to a creative decision to keep a familiarity and not simply lack of creativity. There is room to maintain a series and still vary the content for the hardcore audience. After all, the only reason you do this is because of the repeat customer and certainly not for the new blood. For example: BF2142 did not fail, but it was also not a huge hit, simply because it was significantly buggy as to limit it's appeal, especially for the repeat customer. So a conclusion can be drawn that familiarity is not the be all, end all. The internet factor killed off the game after initial sales as word spread of major unfixed bugs after each patch. Call it the frustration factor.
The frustration factor. This is a big issue for any player. The cause is varied. I mentioned bugs that are game killers. From crashes to unexpected problems that prevent proper completion of some task or level segment. It goes well into the required skill to complete an action or solve a puzzle as well. Testers with various skill sets must be used before the game is released to avoid these issues. Beta testers tend to be in house and are to familiar or just as bad, drawn from the trusted experts in the gaming communities. The cause and effect is to many knowledge based problems ocurr that turn off the beginner. It is like writing a document and not including the acronym definitions properly or not at
Yes... Kids do it all the time...
Oh.. you meant computer games...
David
Redhat has 3 nerds.
And they're lvl8 blood elves.
Grandparent's point was that common Linux distributions includes a PDF viewer on the disc. This means you don't have to Google it. And if you live in the country, it also means you don't have to keep the phone line occupied for an hour for a 16 MB download or drive an hour round trip to an Internet cafe that has broadband.
Fun for me in games usually means there's always something to do or press. The old 2D games were more like this, but a large reason why I hate modern 3D games, is that there's often lots of sprawling around without really doing much (partially related to the 3D world, but it can be solved with difficulty).
I like the idea of how the article mentions that the algorithm biases towards games which can't just be won randomly. The board game is Go is the ultimate example of this I guess, where there are many *levels* of mastery.
But one has to be careful with this approach. If in a 3D game there's a small opening in (say) a castle wall, and miles around of plain grass, it's pretty easy to solve for a human player, despite the huge searchscape and 'narrow' solution that a computer would find tricky (which would apparently potentially rate as a good 'puzzle').
At the very least, developing models for other human factors such as reaction time, subtlety of graphic elements, and the challenge of pressing certain key combinations, would also be needed before final game automation could be achieved.
Why OpalCalc is the best Windows calc
90% of games that came out wouldn't suck
juenger1701
Because it's not proprietary?
Oh come on. Its an article about a novel nested evolutionary algorithm, not about game design...
"Easy to learn, hard to master."
This falls under that umbrella. You're basically rewording this mantra in a more complex form.
"All great wisdom is contained in .signature files"
that gets the impression that the innovative fitness function of the GA was the point of this article, and not the pac man games?
If the pac games were teh point, surely the author would have made "the best" of them available to play online? Really, programming this GA setup was almost certainly a lot more complicated than programming a simplified pac man clone.
Okay you threw politics in the beginning and made a good point. Then you severely detracted from your argument by throwing in the stimulus package and your opinion that it will suck and that it's not imaginative.
I'm not going to argue that point with you, because this is off topic, but in terms of making your point, in the future, try not to apply so much politics to a post that has nothing to do with it, and don't end a perfectly good post about creativity with a political opinion.
"All great wisdom is contained in .signature files"
K. C. Munchkin!, that's a good game:
http://www.gamespy.com/articles/488/488692p1.html
"MIT betrayed all of its basic principles."
I propose an interactive website that posts a bunch of text, then people can create an account and add other bits of text to it. Some people will find those new bits so interesting or so ugly they will add more bits of text. Then people who posts lots of text will get credits that they can spend to hide other people's text that they don't like.
Then we'll pretend it's all factual and news and call it Slashdot!
Evolution SHMUP: http://www.kloonigames.com/blog/games/evolution-shmup
This was an experimental game, where the theory was to see how long a player would remain in a given game - as people continued playing, the system adapted gradually in order to maximize the fun value (in this case, the amount of time spent on a single game.)
This experiment has a smaller search space than the article, but isn't generating any "successful" games. This may be caused by the environment(i.e. the evolution scope is too narrow and thus isn't generating a variety of enemies), but the same problem can easily apply to the article in question.
People haven't even played the games that could be generated with this theory and they're already harshing on the technique.
I hope you guys hate fractals and Amazon.com recommendation lists as well. Nothing cool is ever generated procedurally and there is certainly /NO/ creativity involved in initially making these procedures. At all. Really.
And this long long speach comes to one point... That-- OOOO! QUARTER!
I.e. not really likely, in my opinion.
At best I think it could be interesting to use automated tools to verify that a ruleset can cover corner cases (by randomly or semi-sistematically try a large number of strategies) and or to help balance things, i.e., again, "playing" an enormous number of games to see if some resource/piece/characteristics is too costly or too cheap for the effect it has on the rest of the game.
In other words, automated playtesting. Which may still be near to impossible, but would benefit the gaming authors without devaluating their creativity and skill.
Nuf Sed
Table-ized A.I.
Don't let Raph Koster have anything to do with the design. Ref: Ultima Online and Star Wars Galaxies
I have to laugh every time I see something put the words 'fun', 'game', and 'Raph Koster' into a sentence together. After Starwars Galaxies I have to wonder about him. If that game was seriously his idea of 'fun' I don't think I'd ever go to his house for lunch or anything lest I end up on the menu...
I think Popcap has being doing this for years. I mean, nearly all their games are just variations on, "make a group of three or more objects of the same color", endlessly iterated with some peripheral rules into the most addictive configurations. Throw in a random aesthetic theme and some cheerful sound effects and watch the money roll in.
This troll always amuses me when I see it. The idea that an OS evangelist WOULDN'T EVEN WANT their OS to be popular. I guess it's like the people who were into bands "before they were famous".
With all due respect I think this is a lame idea. The input of human creativity is what will make a game stand out. A machine will spit out random garbage as it does in article spinners.