Newell describes a trend toward end-users being involved, though, not just as spectators, but as content creators. He describes this in fairly sweeping terms: “Games will becomes nodes in a linked economy, where the majority of digital goods and services are user generated.” That sounds a bit grandiose, perhaps, but it’s grounded in numbers. “The Team Fortress community creates 10 times the amount of content [that developers do],” says Newell. While he says Valve has always been happy to compete with other game studios (“we’re a little bit cocky”), “the one entity we wouldn’t ever want to compete with is our own users; they’ve already outstripped us dramatically. It’s not by a little bit; it’s an order of magnitude already.” Broad-based distributed development like that is what open source has been whipping up in the world of software for decades.
Creating games or games content, though, isn’t for the faint of heart: centralized online app stores (Apple’s in particular) “put an enormous number of roadblocks in front of doing that,” including developer approval as well as vetting individual apps and updates to them. In that context, he says, few users have the stubbornness or wherewithal to get through that. A more streamlined system for taking advantage of eater player/developers is needed.
“Several years ago, we thought ‘OK, if our model is correct, we need to help making Linux a good gaming plaform for users and developers.” To that end, Valve makes for a case study in how Linux has been creeping in: the company shipped the first dedicated games server running Linux in 1999. Now, most games servers run Linux (now several hundred thousand — and “probably a million”).
Those game servers are dishing up prodigious loads of data: “Near as we can tell, we’re generating something like 2 to 3 percent of worldwide mobile and land-based IP traffic, and that tends to startle people who don’t realize what a large sea change is going on. Even ignoring game servers, we’ve delivered over an exabyte of data year to date.” (Internally, he says, there’s approximately 20TB of content in a Linux-based version control system. This, says Newell, is true for companies like Bungie, too.)
Impressive as those data-shoveling numbers are, they don’t exactly shout desktop (or living room) success. But steps that Valve (along with other companies) has taken make it easier to swallow the claim. “Several years ago, we thought ‘OK, if our model is correct, we need to help make Linux a good gaming plaform for users and developers.” The first major move, says Newell, was to get a game — a real, graphics-intensive game — going on Linux. The process, though, revealed a “sweater thread” of issues, revealing flaws in in all parts of the stack: faulty drivers, gaps between Linux distributions’ included software, pitfalls in the user experience, and flaws in the company’s Steam tools.
In the course of resolving problems in each of those layers, “The good thing is that if we get a game like Left for Dead running, we’ve probably worked through issues for lots of developers. We’ve definitely solved problems for the Call of Duty team, or Tour of Duty, or whatever. The games aren’t that different; the key thing is to get changes all the way through for users. In February, we shipped [the Linux] Steam client; today -- at least when I got on the plane -- Valve has 198 games running on Linux.“
The bug-fixing and code-developing isn’t just a sporadic effort; the company has “several guys on SDL,” started by current Valve employee Sam Lantinga, and is co-developing a new Linux debugger, in addition to the work they’ve done on the LLVM debugger.
Making Linux a better platform for games is necessary, but may not be sufficient in itself, though. Platforms tend to cluster not just by operating system, but by context: platform, mobile, and console games don’t always play nicely: “As a user, I shoudn’t have to buy new games, or have new friends, or whatever, just because I’m sitting on a couch.” With Linux certainly a more-than-viable software platform for games, but still in the chicken-and-egg world of low user and revenue numbers that discourage spending developer time on Linux end users, Newell says the next step is necessary work on the hardware side of the equation, to smooth the open-source path between the developer and back-end data handling side of the games business to actual end-users.
“One of the things we had to do, is we're staging out the different pieces we think are necessary for staging to make Linux the future of gaming,” said Newell. “Our next step, having done these other pieces, is on the hardware side. There are thermal issues and sound issues, but also a lot of input issues.” He closed with this tease: “Our next step on this is to release some stuff we’ve done on the hardware side. Next week we’re going to be rolling out more information about how we get there, and what are the hardware opportunities we see for getting Linux into the living room."