'And onto Linux' would be super cool, but I see nothing especially bad or wrong with having Windows as an option, too. Healthy competition is a wonderful thing.
I look to how successful Valve was in getting games for the Mac made. That success rate is mixed:
* All the valve games were ported.
* A bunch of casual games get ported.
* Civilization V
* Um, some of the GTA catalog, IIRC.
Frankly, it was kind of a desert. I can't imagine the Linux gaming will pick up unless the GabeN had "bet the farm" on Linux versus Windows 8. Which I don't see happening. He might hate Windows 8, but Steam games play nicely on it, TTVM.
The funny part is that the GTA ports are more playable than their Windows originals (on modern Windows boxes). There are a slew of backwards compatibility issues that now exist that make games like GTA:SA nearly unplayable.
In any case, I wouldn't put Civ5 in the "victory" category either. As with most Aspyr ports the performance is awful, and it doesn't support a bunch of pretty core features like cloud saves, or the very popular built-in mod installers that exist on PC.
Once you've made a game engine with the ability to port, it's easier to port a second time. Steam is (was?) just a store, nothing stopped people from porting to mac before steam went to it, and little more incentive was there after, especially initially. However now Valve has effectively opened up the Linux market in addition to the mac market. Additionally indie games have historically already had high multiplatform support. Now porting to linux will have the same incentive as porting to consoles; While the game engines will have to move first (Unity recently got there), I give it 5 years before making games for all PC platforms is common practice.
At a certain point, it's up to the users to show up.
I think that's what sinks OS X support. [1] And that's the big, recurring question-mark with gaming on Linux.
The vocal subgroup asks and cajoles and begs. But those devs who make the effort don't seem to find a large enough market to justify it.
[1] It didn't help that Apple's Mac App Store arrived so soon after OS X Steam showed up. Nor that mobile computing got so large to the point that PC upgrades basically stalled.
Another thing to consider about Mac gaming: since iMacs and MacBooks are essentially un-upgradable and they're not particularly designed for gaming in mind, they tend to age faster than a similarly-priced self-built tower PC (at least for gaming.)
I don't think that's going to be an issue for Linux enthusiasts, but it will be an issue for this "magic grapefruit" they're talking about.
I seem to recall gabeN posting to the effect that the average Mac running Steam was better-equipped than the average PC running Steam. If that's true, the non-upgradability might not have entered into it.
As to this "piston" thing here - I don't think upgradability will be an anchor. The whole project isn't remotely feasible unless a distro for build-your-own boxes is also launched. In which case the specifics of this box are more a blueprint for builders and a performance baseline for developers.
Sales probably wouldn't even track net adoption until the enthusiasts build out their custom rigs and start convincing their less-technical friends they can have almost all the same awesome without getting any thermal paste on their hands.
