We've seen a good example of this experiment with CCP and EVE Online. A few years back, they released versions of their client for Linux and for Mac. A year later, they cancelled the Linux version and kept developing the Mac version.
Why? Because the Mac client brought in new customers, while the Linux version simply resulted in existing Windows customers switching clients; it didn't bring in any new business.
Eve onine's linux client was also based on that terrible cedega game emulator and it rarely worked.
CCP tried to cheat by not makiing a native linux client and it obviously failed.
It was essentially the same wrapper they used for the Mac port. Good as native? Of course not. Good enough to bring in extra customers? Yes for the Mac port, because we're not generally willing to jump through the hoops required to dual boot. No for the Linux port, because jumping through hoops is a way of life for them; they were already playing the game that way.
Doesn't matter how good the port was — they put in the investment they were willing to put in, they got a good return out of it on the Mac client, they didn't on the Linux client.
Until Linux users are willing to stop buying non-native games rather than simply dual-booting or running in WINE — and do so in numbers as significant as Mac users — there simply isn't any economic incentive to make Linux games.