Procedurality and Content: Reactions to Mateas and Stern

Matt Barton's picture

I've been doing lots of research lately for a new book and came across a passage from the designers of Facade, a fascinating "game" that for many shows the way to the future of gaming. I didn't realize just how savvy Michael Mateas and Andrew Stern are until I read their chapter in Second Person--a really good book, by the way, with chapters not just by academics but several key developers (including Steve Meretzky and Chris Crawford). Here's the passage from Mateas and Stern (inspired by Chris Crawford) that really got me thinking:

Where insufficient procedurality is creating a crisis in the authoring of traditional games, it has prevented some long sought-after genres of interactive art and entertainment, such as the high-agency interactive story, from even getting off the ground. Bringing process-intensive, AI-based techniques to the problem of interactive story was one of the fundamental research goals of our interactive drama Facade.

Basically what we're dealing with here is Crawford's concept of "process intensity," or "the ratio of computation to the size of the media assets being manipulated by the system." In other words, how much of the "game" is algorithmically or procedurally generated vs. pre-generated "media files."

I'm still not quite sure that distinction is clear enough, but I think you can see what they're talking about if you compare Rogue with Diablo. Both of those games rely on "procedurally generated," process-intensive techniques to create dungeons and populate them with monsters, items, traps, etc. However, Diablo is a bit more on the right side of the continuum, since it relies more on media assets, such as pre-recorded music, artwork, and so on. Of course, Rogue also has "pre-generated content," if we include text for names and so on (as far as I know, you're not going to meet a randomized creature with a randomized name). There seems to be a need for some type of pre-made structure in any case, unless you really are content to play in a purely abstract realm (oh, there's monster 49c29 or some such).

What's interesting about Mateas and Stern is that they point out a tension between process-intensive stuff and story/drama. It seems that efforts to have the computer react algorithmically yet realistically to story-making, dialog, or the like, is extremely difficult. It seems that most developers result to brute force (providing trees and options that require vast amounts of resources that many players will never see!) or "rails" type linear gameplay (cut scenes, etc.) In short, you need serious constraints if you want to have anything like a focused story. What M&S point out is that all games must, by definition, have other kinds of constraints (rules, borders, possible actions, etc.) and we can think of this focused story as just another such constraint. In fact, stories work like games in so much as they have these constraints. For instance, stories have beginnings, middles, and endings, and characters (if they are to be believable) must have some kind of recognizable consistency. I'd go a bit further and say that the fictional worlds created in stories have an internal consistency as well; even something as crazy as Piers Anthony's Xanth has very predictable rules that you can discover if you spend enough time reading the books (all puns are taken literally, so an "arm chair" is a chair where arms can sit down!). So, a neat trick would be making the constraints inherent in fiction the same (or similar to) those of a game.

Take for instance the Monkey Island games. Although much of Guybrush's decisions are simple, I still recognize him as a character and eventually get a pretty good idea of how he'll react to any given situation he's likely to encounter in the game. Of course, in these games the plot, story, drama, and so on, have all been pre-determined and the player is just left to connect the dots (figure out the trail the developers laid out for you and follow it, navigating or avoiding pre-determined obstacles along the way). Now what would it take to take that trail out of the developers' hands and let the computer generate it, so that you could play Monkey Island forever (again and again) and always encounter fresh, fun content? Would such a thing be possible, or is it only possible to have truly "process-intensive" games that are extremely abstract?

Just imagine an MMO like World of Warcraft and the possibilities...A virtual world that could respond dynamically to the activities of its players, always crafting new realms, stories, plots, characters?

Comments

Nous
Nous's picture
Offline
Joined: 04/07/2007
Static vs Dynamic

Very interesting topic, Matt.

To answer your question briefly, I think that the distinction between process and content is very clear. I've always maintained that what makes videogames a unique, promising medium - unlike any other - is the fact that it's entirely dependent on the "computer" as the enabling mechanism for dynamic behaviour.

This simply means that with other media the "content" is encoded in a static, unchangeable form; other forms of art do not have "behaviours" and what makes them interesting is the way we interpret them. In short, all of the processing takes place in our minds. At the time of conception the content is created by the author, artist, composer, and so on and remains as is for the rest of its life. At the time of reception, the content, static as it is, becomes embedded in the minds of the audience where it is interpreted and "processed" (i.e. experienced in conjunction with the person's make up and life experience).

This need not be the case with videogames!

Certainly, we are a lot more familiar with creating static content and that is of course possible to do with videogames as well. In fact, if taken to the extreme, a videogame can be designed to be little more than a movie (e.g. Dragon's Lair).

But here is the thing: computation puts the emphasis on behaviour, not content. All other media are static by their very nature - they cannot think, therefore they cannot have interesting behaviours, or indeed ANY behaviours whatsoever. But computation is inherently dynamic; it is ALL about behaviours, preferrably interesting ones that can surprise and captivate their audience.

Procedural creation of content, in the context of videogames, should not be thought of as merely techniques for creating "on the fly" objects and environments for our game worlds. The most important thing to remember is that the videogame itself can now be thought of as having a mind of its own - or perhaps many of them, doing several different things together. It is precisely this that puts the "inter" in inter-action, for we can always act on any other item or content, but it won't reply back to our actions in any interesting way other than that which is dictated by the laws of nature (you can damage a painting but that is not to say that it being destroyed constitutes an interesting "processing" behaviour on its part).

As an interesting sidenote: you may know that Socrates absolutely refused to write down his ideas for others to read because, as he explained, they are dead; they can't actively respond to any questions the reader may have. Since every reader may be different those questions may be very important in the correct, meaningful (or in our case, fun) interpretation of those ideas. He preferred to communicate his ideas to others in person instead, by using a very intense form of interactive dialog. It's a little bit similar to the buddhist idea that you need to be taught the truth by a true master before it actually dawns on you. The truth can't be conveyed only in writing, you have to experience it and for this to happen you need to actively interact with "stuff" in far too large a number of interesting and complicated ways for them to be encapsulated in "static scripts".

Of course Socrates and the Buddhist philosophers were more concerned with the right message being communicated so as to avoid misleading people but you can substitute "true message" for "meaningful experience" or even "fun/interesting experience" and you will see why processing is so central to videogames being a uniquely powerful artform.

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

"Elegance has the disadvantage that hard work is needed to achieve it and a good education to appreciate it."

-- Dijkstra

n/a
adamantyr
adamantyr's picture
Offline
Joined: 01/28/2007
MMO's and Testing

MMO's and dynamic story-telling probably won't happen anytime soon.

You can take the first incarnation of Ultima Online, during the beta and the VERY early starting period as an example. The economy and ecology were designed to react dynamically to outside influences. If you killed, skinned, and butchered deer, the price of leather and venison would drop the more you sold. Items in high demand like tools and weapons would get priced higher. More vendors of the same type meant competitive pricing schemes. Etc. etc.

What happened? Well, the game's resources ran out in no time flat. During beta, it was a race to buy tools (sewing kit, scissors, smithing hammer) then a race to find as many cows/deers/etc as possible to get skins. After about 30 minutes there were no living animals over the whole continent, all the shops were sold out of all tools and items, and players were murdering each other for sewing kits. Seriously, this really happened. It wasn't long afterward that the UO design team chucked the economy into the bin as a "nice idea on paper but unimplementable at this time".

Also, as a software tester, the idea of TESTING a dynamic story-telling crafting engine is scary. Testing software is all about identifying your inputs and your potential outcomes from those inputs. Trees and linear rails may be dull, but at least you know what you're going to get. In a dynamic system, things are so open-ended that you can end up with code and data objects interacting with each other in completely unpredictable ways that could cause the whole system to fail. Without a SOLID design framework that is very robust, it would never work right, and you'd likely end up with combination of events/objects that would be a total surprise.

Matt Barton
Matt Barton's picture
Offline
Joined: 01/16/2006
Total surprises could be a

Total surprises could be a good thing. :)

I do agree though that the challenges are immense. Facade isn't perfect, of course, but it was successful as it was because it was quite limited in scope, with a loose outer shell and many, many micro decisions. IIRC, they originally wanted about 3x the number of situations and responses, but limited them to get it done.

Blizzard's tactic with WoW is to have a giant, ongoing story that is told mainly through in-game sequences and quest cycles. It's actually quite easy to miss, because 99% of the players don't seem to give a shite about the story (or perhaps have seen these non-skippable sequences so many times they don't care anymore). The expansions are all built around this story. My biggest complaint about it is that it IS so easy to miss it. I've been playing for years now and don't give a whit about Arthas or Sylvanas or any of them. It reminds me of the story elements in the House of the Dead games; sure it's there, but I seldom pay any attention to it. Contrast that sort of thing with a game like Twilight Princess where you are given a chance to get to know the characters and care about them.

I like to think of something like Facade but greatly expanded. I think you could use that technology to create much better NPCs. Hell, if Meretzky could make us cry with Floyd back in '83, we ought to be able to do something ten times better in 2010! Actually, Floyd is a good example; as M. has pointed out, the fact that he is a robot lessens our expectations quite a bit for him to "act human." Even animals with dynamic responses would be a huge plus. Imagine just how sophisticated a "simple" character like a dog companion could be if they really put some time into it. The more I think about it, the more I think they should start there. Robots, animals, alien species, perhaps small children ought to be relatively easy platforms for really good advances in AI and interactive drama (or whatever you want to call it).

n/a
adamantyr
adamantyr's picture
Offline
Joined: 01/28/2007
Emergence and WoW
Matt Barton wrote:

Total surprises could be a good thing. :)

I link you to Chris Crawford's short article on Emergence in Software: http://www.erasmatazz.com/library/Game%20Design/Emergence.html

WoW doesn't try to get ambitious with their designs. Frankly, they don't need it. The player base is perfectly happy to have the story if they want it, and ignore it if they don't. It's all about the bottom line...

In fact, Blizzard's introduction of the new looking for group system has shown that someone must have done some serious feedback analysis and realized they were getting too narrow-focused on high-end power gaming activities like raiding and PvP. Other online games have followed this same pattern only to see their player base deteriorate as the game gets too heavy and overloaded for new players to get involved with it. By doing cross-server grouping and making it easy for players to participate, they've opened up a lot of content that was getting passed by or ignored. I'd surmise that character scaling is the next thing to go in, so high-level players can enjoy low-level content without balance issues.

Hammer
Offline
Joined: 03/23/2008
I'm sort of pissed at those guys

I paid money to get the Facade behind the scenes, which supposedly would show how to get the best endings, etc. I received a text file that wasn't complete and said they would hopefully be completing it sometime a few years ago.

Matt Barton
Matt Barton's picture
Offline
Joined: 01/16/2006
Really? Sure that was

Really? Sure that was legitimate?

n/a
Hammer
Offline
Joined: 03/23/2008
Yes I'm sure...
Matt Barton wrote:

Really? Sure that was legitimate?

I ordered it from here:

http://www.interactivestory.net/goodies/behindthefacade.html

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.