Process Intensity

At one of the earliest (C)GDCs I listened to Chris Crawford lecture about what he called "process intensity." In an article for the Journal of Computer Game Design, he later revisited the subject, and wrote:

"Process intensity is the degree to which a program emphasizes processes instead of data. All programs use a mix of process and data. Process is reflected in algorithms equations, and branches. Data is reflected in data tables, images, sounds, and text. A process-intensive program spends a lot of time crunching numbers; a data-intensive program spends a lot of time moving bytes around."

In his lecture, one of Crawford's points was that computers (then) had far better ability to crunch numbers than to store data. As an example, consider the game-map for a board wargame: a typical such game has over 2000 hexes, each of which can contain a variety of terrain types; figure a byte per hex. Each of the hex sides can also contain terrain--roads, bridges, rivers, and so on. Each hex-side is shared by two hexes, so there are 6000 or so in total, and figure you need a byte for each of those too--so to represent the game-map for a board wargame, you need 8K.. If you're working on, say, a 48K Apple II, you've already consumed a big part of your memory budget. You're better off trying to create a game that exploits the strength of computers--the ability to crunch numbers--instead of trying to work around its weakness--the ability to store data.

Of course, this element of Crawford's argument was made moot by the progress of technology; today, you typically have a gig or more of RAM to work with, and (except for downloadable games), application size is almost never an issue.

Since the early days of digital games, the focus of the whole industry has been not on process, but on data, on improving our ability to display media--graphics, sound, and video. And it hardly needs to be said that enormous strides have been made, from the square blits of the Atari 2600 to the near-cinematic quality of modern games. Those who have followed this path have prospered, too; having a better looking game than your competitors has been the path to success.

But success has come at a cost; once technology permits better graphics, better graphics become mandatory (what Raph Koster calls Moore's Wall). An RTS game that does not look as good as the most recent Ensemble title, or an FPS as good as the most recent one from id, has little chance at conventional retail. While the tools used to create graphics have improved over time, they have not improved nearly as quickly as the capabilities of our game-playing machines--and this, in fact, has been the main driver of the industry's ever-spiralling development costs.

When we began, the bulk of the effort in developing a game was in the coding, and perhaps an artist was brought in toward the end of the project to turn "programmer art" into something better. Today, 80+% of the man-hours (and cost) for a game is in the creation of art assets.

In other words, we've spent the last three decades focusing on data intensity instead of process intensity. I believe we've reached and indeed gone beyond the point of diminishing returns; budgets are now so high that a million or more unit sales are needed for profitability for A-level games. And as Will Wright says, a game with 22,000 animations is not twice as good as a game with 11,000.

Moreover, while Moore's Law may not have slowed, the perceptible improvements in media display are increasingly minor. That is, the difference between the colored blocks of the Atari 2600 and the 8-bit sprites of the NES had a huge impact; the difference between the sprites of the SNES and Genesis and the 3D of Playstation was revealing; the improvement in poly count between the PS 1 and 2 was important. But while the demos that show off the capabilities of the Xbox 360 and PS 3 are impressive, they're still incremental improvements over the last generation, and nothing like as earth-shattering as prior leaps.

I would argue, therefore, that to expect to gain a competitive advantage over others by continuing to push the graphic envelope is today foolish. Graphics today are good enough. Indeed, it is worth noting that World of Warcraft, today's most commercially successful title (in terms of overall dollar gross) does not have high-end graphics--Blizzard made a conscious decision to keep their costs down, and broaden their audience, by sticking with much lower poly counts than most modern games.

With Spore, Will Wright is pushing the notion of "procedural content"--that is, models and animations that are not created by artists and stored on a disk, but generated algorithmically. He holds this up as one way to improve the variety of experience provided by a game while also reducing the effective asset creation cost. Which indeed it is, and some have suggested that procedural content is the future of games.

Perhaps so. However, the idea of "procedural content" is, in essence, a subset of the idea of "process intensity." Wright is using a computer's ability to process data to reduce the need to provide lots of data. He is, in other words, following Crawford's advice. But using processes to generate "content" is only one path to more variable and interesting games.

Crawford's original argument was based on the inherent characteristics of computers, as they then were, and today it is often viewed as moot because those characteristics have changed. Let us consider therefore not the inherent nature of computers, but instead the inherent nature of "the game" as a medium.

All other artistic media are either static and unchanging, like a painting--or change over time, like film or music. What's different about games is that they change over time, but not in a fashion planned by the creator; instead, they change in reaction to the actions of the players. Games are interactive at their core, and this is as true of Chess as it is of anything you play on a computer. Indeed, the reason games took to computers like ducks to water is because computers, being interactive by nature, are naturally suited to providing games, which are likewise interactive.

How does a game change state in response to players' actions? Through the use of algorithms. This is obviously so in the case of digital games, but equally so for non-digital ones; "rules" are in fact algorithms.

The very "gameness" of a game, the very thing that distinguishes games from other forms, is their use of process. Indeed, designing a game means specifying the processes that will be used in play--determining the rules, in other words.

The graphics, pieces, sounds, et al. are used to provide a representation of the game state to the players, but in a sense, they are the mere surface of the game. The heart of the game, and its meat, is in its processes.

Now, I do not wish to suggest that the nature of this representation is irrelevant to a game's appeal; the aesthetics of a game's imagery, components, etc. greatly affect the experience and the enjoyment of play. I suspect you could build a client for Quake that used 2D sprites, and play against someone using the original client, and have pretty much the same experience of play--but it wouldn't be anywhere near as interesting, or as much fun.

Nonetheless, what game developers have spent most of their effort on to date, and where most of the money is spent today, is on improving the surface expression of games--not on exploring the possibilities of new gameplay through the exploration of new processes, and combinations of processes.

To put it another way, we've been doing precious little actual game design; we've mostly been implementing new versions of existing game styles, but with prettier graphics.

But as I've argued, we've come to the point of diminishing returns along that road. In future, we have to do something else. And the something else we have to do is to start thinking about process--about how different algorithms can be combined to create challenges and gameplay that people haven't seen before; about experimenting with different approaches to gameplay; about how rules shape player behavior.

For two reasons, the indie game movement has a big role to play in this, I think. First, the financial pressures on conventional game development are so enormous that publishers are forced to do whatever they can to reduce the risk of development, and by and large that means sticking with the tried-and-true, with incremental changes to existing game styles and with licensing or creating different IP, and implementing an established game style within the framework of that IP. Yes, there is Spore--but don't expect a lot of similar innovation from the big publishers. Few people other than Will Wright are given that much rope.

And if we're talking about experimenting with process, rather than visuals, then you fundamentally don't need a big budget, anyway. You can play around with different approaches to design, releasing them at whatever graphic level your budget allows--and if and when you hit on an approach to strikes a chord with your audience, you can then gear up to deliver the high-end version of your design.

That's probably necessary, anyway; historically, major innovations in design take two or three iterations before the developer really nails the ideas down and figures out how to make them work. Wolfenstein 3d predated Doom, and there were two version of Grand Theft Auto before everything came together with the third... In the conventional market today, your first attempt would be viewed as a sales failure, and you wouldn't get the opportunity to try again. In indie games, its possible.

I don’t mean to suggest that there is any dishonor in taking an existing game style, and implementing it well; but if the field is to retain its vitality, we need to devise new gameplay, not just new games. This is, perhaps, the real promise of independent games--that, like independent music and film, it can provide an avenue for creativity and exploration, at lower risk than the larger market, and thereby serve to reinvigorate the mainstream.