"The machinery of gaming has run amok... An industry that was once the most innovative and exciting artistic field on the planet has become a morass of drudgery and imitation... It is time for revolution!"
- "Designer X" in the Scratchware Manifesto
When "Designer X" wrote those words back in 2000, the industry, to the degree that it took any note whatsoever, dismissed them as irrelevant ravings. Jessica Mulligan wrote that the Scratchware Manifesto was "naive in the extreme," and obviously written by an industry outsider - and was quite surprised to learn that I was Designer X. Of course, things have only gotten worse since 2000, and the industry - or at least, developers - have started to agree.
Two years ago, speaking at a conference in the UK, Warren Spector said "The publishers have to die, or we are all doomed" - to cheers. And this year, at GDC, I ranted on the problem - and received a standing ovation.
What is the problem? And is there any way to address it?
As recently as 1992, the typical development budget for a PC game was as little as $200,000. Today, if you want a title that will be taken seriously by the retailers - an A-level title - your minimum buy-in is $5m, and $10m for a triple-A title is common. With the next generation of console hardware, the talk is of $20m budgets - not as something that will be unusual, but typical.
On a theoretical basis, the rise in development costs is driven directly by Moore's Law. As hardware becomes capable of displaying better-detailed graphics and higher polygon counts, it becomes mandatory to provide them. If you do not, your competitors will - and your games will look inferior by comparison to directly competitive product. In the accompanying graph, you'll see a huge rise in the '90s; that was driven by the adoption of CD-ROMs. Before CD-ROMs became common, games had to be delivered on floppies - and even if you did a game with several floppies in the box, your application size was still measured in single-digit megabytes. CD-ROMs provide more than 600 MB - and once you had the capability of providing that much data, doing so became mandatory. You had to generate enough assets to fill the disc.
Today, art assets (not programming) are the main cost driver. As machines become capable of rendering more detailed 3D models in real time, the market demands more detailed 3D models - and models are hand-created by artists using tools such as 3D Studio Max and Maya. All things being equal, a doubling in polygon count means a doubling in the amount of time an artist needs to spend generating the model - and a doubling in cost. Faster machines can push more polygons; more polygons means more cost.
That's the theory, but empirical evidence bears it out. Back in the day, a Doom level took one man-day to build. A Doom III level takes two or more man-weeks.
Now one might argue, of course, that the improvement in graphical quality improves the gameplay experience so much that the cost is worthwhile. But if that's so, why was Doom so rapturously received, such a huge hit? And why do the critics basically agree that Doom III - well, it kind of sucks?
It's not the glitz. It's the gameplay.
In principle, increasing processing power also allows better development tools, which helps speed the process. But the reality is that it ameliorates, not solves, the problem. The tools don't get better fast enough. Middleware doesn't solve the problem, either - you might get a product to market faster by licensing the Unreal engine, say, than by building your own 3D renderer from scratch - but you're still faced with building all the content. And the case for buying middleware is rarely open and shut; each engine is designed for specific purposes, and if you want to do a game that differs a lot from what the code was written to support, you find yourself spending a lot of programming time trying to solve the problems. Spector says he's not at all sure that the development of Deus Ex actually benefited from using Unreal - after all, Unreal was built as an FPS engine, and Deus Ex was an FPS/RPG hybrid. They had to rewrite a lot of code.