The first avenue is building games using an alternative API such as OpenGL. This is the path Valve have taken with their own games, developing OpenGL versions of their Source engine and the games that use that engine such as Half-Life 2 and Team Fortress 2. Much as DirectX has the advantage of being used by the Xbox line of consoles as well as the PC, OpenGL is used on the Playstation 3 and 4, so many developers may already, or in the years to come, produce OpenGL versions of their games. However, since DirectX dominates the Windows-based PC gaming platform, most existing PC versions of games use DirectX, and rewriting them to use OpenGL or any alternative API probably isn’t practical in most cases. This is especially true with older games, where the developers may not even have access to the original source code, or where the developer may not even still exist. If SteamOS simply replaced the requirement for DirectX with a requirement for OpenGL, this would wipe out most of Steam’s extensive game library. The extensive back catalogue is one of the major features of the PC as a gaming platform, so this is obviously undesirable.
The second avenue is for the OS or other software to provide compatibility with DirectX. Essentially, the OS or other software pretends to be DirectX, and allows the game to call on it the same way it would call on DirectX in Windows. This is the path a number of major publishers, such as EA and Activision, have used to produce Mac versions of their games. If Valve wants people to be able to play games written for DirectX on SteamOS, this is probably how they’ll make that possible for most games. The only problem with this sort of compatibility approach is that it’s not perfect. Compatibility software has been getting better and better at replacing DirectX, but there are still problems, and to my knowledge there’s nothing out there that can perfectly substitute for DirectX. The existing solutions also require some messing around to make them work with existing games. So Valve will have to address these issues if they want to make SteamOS compatible with DirectX games.
The third way to get around Windows-exclusive games is probably the most interesting one, and that is to stream games to a PC running SteamOS. This would mean having a game running on another PC, a conventional Windows desktop gaming PC, and a PC running SteamOS would connect to that desktop gaming PC via a network, and then stream video and audio from it, and input commands to it. The desktop PC would run the game in much the same way as any desktop PC would, but instead of outputting video and audio to a monitor and speakers, it would output them to the network. And instead of receiving input from a keyboard and mouse, it would receive that input over the network. Likewise, the SteamOS PC would display video and output audio from the network, and pipe input back down the same network connection. The biggest problem with this approach is that it requires a second PC to run the game, one that runs Windows rather than SteamOS, and one that has all the usual Windows gaming PC hardware requirements. While companies like OnLive already offer streamed gaming over the Internet, Valve’s SteamOS announcement makes it clear that they intend to focus on streaming over a local network. This means users would have to use a desktop gaming PC they already own, or buy one. Streaming, therefore, would restrict SteamOS to complementing the existing Windows PC gaming platform, rather than replacing it.
It’s obvious from Valve’s announcements that they intend SteamOS to use a combination of all three approaches to getting around Windows-exclusivity. Their own games will run on OpenGL, as will many other games. Many more will run using compatibility software. And for games that can’t do either, users can fall back on streaming gameplay from a Windows gaming PC. This three-pronged approach should allow SteamOS to cover every base, and make Steam’s vast library of games accessible on a PC running SteamOS. But each approach has obstacles that will have to be overcome, and there are bound to be games that fall through the cracks. It remains to be seen how well Valve can overcome those obstacles, and catch the exceptions that SteamOS is bound to throw up.
After DirectX, the availability of drivers for video cards and other hardware on Linux is probably the significant challenges that SteamOS will face. While Linux has been catching up on driver availability in leaps and bounds over the last few years, there's still a lot of hardware that isn't covered by the drivers that are available. And even where drivers are available, they're often poorly-optimised in comparison to their Windows counterparts. Video card drivers have probably made the most progress, but my understanding is they're still not on par with Windows equivalents. And drivers for wireless hardware are a perennial stumbling block for Linux, something SteamOS will have to address if streaming games over a local network is going to be a primary feature. That said, at least where video card drivers are concerned, Valve has existing relationships with hardware manufacturers, and they should be able to work out a solution.
Finally there's still no indication of whether, or how, a PC running SteamOS will be able to run games that aren't on Steam. Other digital distribution services such as GOG.com, Desura and EA's Origin service have made significant headway as competitors with Steam. Furthermore, many of the most popular MMOs, such as World of Warcraft (still the juggernaut, despite declining subscribers) and Guild Wars, and other online games such as League of Legends an d World of Tanks, aren't available through Steam. Even those online games that are available through Steam have historically had problems running within Steam, causing players to prefer running them outside of Steam. Losing access to games from these other channels would be a bitter pill for SteamOS users to swallow. How SteamOS will handle games outside of Steam might be the biggest unanswered question of the whole Steam platform reveal.
Still, SteamOS has a huge amount of potential. And if it's successful, SteamOS could have an impact beyond the OS itself, pushing PC gaming beyond Windows in all sorts of ways. That can only be good for PC gaming as a whole.
