"As we moved forward, if Half-Life was two and a half years and less than $10m, and Half-Life 2 was over five years and $40m, then Half-Life 3 is eight years and $65m? Hell no, it seemed like the spiral was out of control and change was needed." But isn't there a danger that by pursuing episodic gaming, Valve will get left behind in the technology arms race? Johnson emphatically disagrees, arguing that technology development is speedier and more efficient when you're working on more frequent, incremental releases.
"No, we feel like technology-wise we're building it and shipping it way faster than we were when we were building Source during Half-Life 2's development. There's technology that we built along the way while we were making Half-Life 2 that didn't work, it got thrown away. But it was code that since we weren't shipping, the product was pretty early and in flux, kept getting maintained, so there was a lot of loss in the system on that. Or instead of shipping HDR we have to do HDR and physics at the same time, how do those systems interact, well that's really complicated... we had ten systems that we were doing for Half-Life 2, they're causing bugs in each other, it's kind of a nightmare. But if you can do it piecemeal, once you've shipped a piece of code, it has an incredible amount of value, it's real. Before that, it has no value for you."
The only thing I would worry about is how tempting it would be for episodic game developers to throw in features that don't really fit their title, just to say they have it. So lets just hope they can keep an eye on the big picture. We're looking for a fun game, not a tech demo. ;)