Valve is intending to develop for Linux, as an insurance policy against Windows 8. I think that will lead to a self-fulfilling prophecy. If more games are available for Linux, demand for Linux will increase, along with market share.
There’s historical precedence for this.
In 1984, the IBM PC and compatibles sold about 2 million units. The Commodore 64 all alone sold about 3.5 million units–my first computer was among those–and the second-best selling 8-bit computer–don’t tell Steve Wozniak–was the Apple II series, which sold around a million that year.
In 1987, the IBM PC and compatibles sold 6 million units, and the market share doubled again in 1988 and 1989 while the 64 held steady at around 2 million.
It was around 1987 that companies like EA made a real effort to release games for IBM PCs and compatibles. Games existed before then, but not in the same quantity, and with a few notable exceptions like Flight Simulator and King’s Quest, frequently the games that were available were developed on 8-bit machines and ported to the IBM PC. By 1989, you were seeing games like Railroad Tycoon that were designed from the beginning to utilize the IBM PC’s 640K of available memory and faster CPU. The C-64 had better graphics and sound at first, but since people were buying PCs to play games on them, developers started supporting VGA graphics. And then Ad Lib released a sound card that let PCs play music, and games started supporting that.
Momentum built, and PC clones went from being something you bought to run Lotus 1-2-3 to a major entertainment platform. Meanwhile, challengers like Commodore’s Amiga and Atari’s ST struggled. They were better in many significant ways, but there weren’t as many games available for them, and there weren’t a lot of productivity apps available either. Amigas survived by doing video production and STs survived by doing music production, but that ultimately was a losing battle too. Eventually Apple’s Macintosh took over those niches, in addition to its desktop publishing/graphic design niche while PCs running MS-DOS and, later, Windows, rode the generic productivity and entertainment software tidal wave to maintain the lion’s share of the market.
If a couple of game studios start developing for Linux and it proves moderately profitable, more will follow, and momentum will start flowing in Linux’s direction. Linux has all the basic applications it needs, if not more, but that’s not enough for mass adoption.
If Linux is going to reach (let alone exceed) 10% market share on regular PCs, it’s going to take games to get it there. If a couple of hot titles land on Linux and they happen to run faster on Linux than on Windows, the sky would be the limit. Especially if recent patterns hold and Windows 8 turns out to be Vista remixed.