Let's assume, just for giggles, that the percentage of PC gamers that would use Linux if it was a "real option" for gaming is greater than 0. From a business guy's perspective, having your customers switch from Windows to Linux wouldn't really change a thing, but the "no market for Linux games" thing is kind of self-perpetuating as I see it :p

Beta support... while an interesting point of view and a possible point to start: I think BETA problems of the games itself will be tolerated by users, but not additional deployment & system problems... additional complexity should not be introduced. (I mean, look on the Steam client issues, a entangled mix of game, distro, driver, architecture, deployment, HW support, library version problems & bugs ...https://github.com/ValveSoftware/steam-for-linux/issues?state=open no separation of responsiblity & concerns.)

Infact, if a game is in a BETA stage, game developer prefer a reliable direct update/deployment channel to their BETA customers. Also, developer want to reduce complexity, first they want to fix game bugs and don't want the additional burden of system and deployment issues (typical linux ecosystem issues). Steam for (windows & mac) is offering a direct and reliable developer driven update mechanism, therefore e.g. the Grimrock Beta (http://www.grimrock.net/forum/viewtopic.php?f=3&t=4987) was exclussively on Steam, their own direct bought customers, linux users and the GOG customers were left out. And I don't see that GOG want to introduce (and the customers accept) an steam-client/platform like approach with update enforcement capabilities.

The way to go for linux support IMHO, which fits also the GOG spirit best, is a bundle approach, sharp decoupeling between system and app. Like e.g. http://www.portablelinuxgames.org/ (But for that approach several things still needs to be robustified, unified and stabilized in the linux ecosystem)

You don't have to assume, all you actually have to do is to look at HIB sales and the article linked by Mivas at... some point.

Not really a paradox, if you separate the qualities of the term you was using: DRM-ed / platform
Linux has clearly the potential for offering a DRM-free experience as free and open ecosystem, therefore a good match for GOG. But the Linux ecosystem lacks heavily in the platform department, as it was unable to overcome the old unix disease "fragmentation" and to form a platform. Not a good match for distributor who tries to offer a stable, reliable and therefore excellent user experience.

PS: speaking about this "paradox", was noticed also on the ubuntu launchpad and is the #1 issue there It's easier and safer to install the newest versions of popular open-source software on Windows than on Ubuntu. (Or, why it's high time Ubuntu made upgrading to stable versions of software easier and safer)

Stop calling fragmentation a weakness, disease or whatever. That fragmentation is why you find Linux used for the most specialized tasks, the fragmentation is actually one of the biggest strengths Linux systems have and it is what sets it apart from other operating systems. We don't really need more unified operating systems, we already have MacOS and Windows for the major ones.

nope, I won't :)
Fragmentation is a weakness, flexibiliyt, adaptibility & diversification is a strength.
See this presentation: http://www.youtube.com/watch?v=GT5fUcMUfYg#t=3m09s

Ubuntu thousends < Android hundred thousand < windows millions of applications... which ecosystem provide more diversification, freedom and power by applications to the user?

And I humbly disagree, I want (and think the world need) a free and open source unified operating system, not a continuation of the free and open source mess situation we have now. Also, being a mess and free and open source is not at all linked together as necessarity... we could have since decades a succesful open source OS if some traditional unix crap would have been dropped (or a fresh free and OSS approach like Haiku or ReactOS would be aggresivliy pushed)

The problem with this approach is that you only want to see Linux operating systems as an OS for your average joe, but that's most definitely not its most important use. Of course that when you have plenty of highly specialized distros, there's going to be less applications for the individual distros - on the other hand, I can guarantee you that default installation of backtrack Linux will be better for testing security than your default installation of either Windows or MacOS. And that's where Linux OSes shine, specialization on a particular task - forcing Linux distros to abandon fragmentation (...and the entire idea of Linux distros in general) would pretty much kill this massive advantage over other OSes. Saying that it's not as useful for an average user as Windows or MacOS is quite frankly a pretty narrow-minded point of view.

As for the world needing an open-source OS for the average joe... Well, it might need that. It might not. But it's not going to be Linux, and I most definitely don't want Linux OSes to go down that path forcefully.

Again, I have to disagree, for me the advantage of linux is not coming from the "distro concept" at all. Even for specialized use cases distro hopping is very common, showing that they are not that perfect solutions for special use cases as people like to present them. I would argue, even specialized use cases are far better adressed by stable OS + user selected apps than by distros composed from OS + preselected apps for a assumed use case.

Speaking about narrow-mindness: security is excessivly (hysterically?) overemphasized in the linux world, stemming from the history as unix with a server-workstation platform with a role separation between user and admin. This role model is outdated since the invention of teh personal computing and is not fitting well to the PC / platform ecosystems (also used on Mac, IOS, android etc all successful platforms) where the admin IS the user and a new role is existing the ISV, not really recognized in the linux world. Also, how an well known american called it "The price of (user) freedom is eternal vigilance" ... users prefer freedom on application selection before excessive security motivated limitations. I think a good and the "free software" idea well fitting approach, give the choice and freedom to the user.

I'll see your presentation and raise you another:

http://www.youtube.com/watch?v=7XTHdcmjenI&amp;list=PLYjUXmrqOoKvOb8L5TQvnoRp3stQeeVMt

What you call "fragmentation" we call adaptability and ability to specialize. That is why Linux runs the stock exchanges, banking operations, CERN computers, NASA computers, supercomputers, world networking and communications, most of the home routers (mine has tomato on it, light years better than the buggy default fw it came with), smartphones (one of my phones has android on it, the other us an old nokia with symbian) and desktop computers

You can bet your ass that MS and other vendors would go ape like Ballmer if they had the chance to run on supercomputers but no one has the disposition to wait for the likes of MS and Apple to get off their asses and cook features in their operating systems that those people need. Not to mention the potentially prohibitive costs of development for such specialized features and vendor lock-in.

Also, you compare apples to oranges to berries. Want a fair comparison? Compare iOS, Android and WindowsPhone. Compare Windows Server to other server operating systems. Compare Windows to other desktop oriented operating systems.

You talk about the millions of Windows applications and forget a few facts like how most of those millions are noise. I always hear about games(many run on Linux, heck I currently enjoy Dishonored and Borderlands 2), various multimedia programs and MS Office. Less mentioned but important are Photoshop, Illustrator, CorelDraw, various CAD applications, 3D modelling programs like Maya (runs on Linux btw), Matlab (also runs on linux). There are probably a few others I may have missed.
Nobody gives a fig about "millions" of programs. They care about a few that allow them to relax and/or get work done. Using something like the QT SDK allows one to fairly create cross-platform software. There are plenty other options.

There is already a unified, open-source system that does what you say. It's called FreeBSD and it can do amazing things. Check the latest trends in storage and see how something like freeNAS with strong support for ZFS is taking over. See how the ability to use and manage jails makes a lot of sysadmins happy and moderately care-free . Then, look at something like PCBSD built on top of FreeBSD and 100% compatible with it and it's strong server and workstation oriented features like Warden.

Choice is power not weakness. Of course, you need to be informed and make the correct choices. No software can help you with that.

One thing I blame for approach like that of shaddim's (and I'm not saying that this actually is your case in particular shaddim so don't take this remark personally) is that a lot of people want Linux to be open source Windows. I know I did when I started to use the OS. But something people just need to understand is that Linux is entirely different, it has its own rules and its own uses, and that it approaches various issues in a way that's substantially different than that of Windows.

After I have understood that, I found the OS so much easier to use, so much more friendly towards my needs, and just generally more welcoming. And my worries for videogames pretty much went away when a heavily modified Skyrim worked flawlessly. The fact that all I've had to do when I hit a problem was to start IRC channel, ask, and get help within minutes certainly helped.

What about games that require DirectX 10/11? Does Wine support those versions yet? That is going to be a crucial prerequisite of using Linux for big-budget games; DirectX 9 will be gone from that space within a couple of years given the current rate of developer migration (even Call of Duty has made the jump).

It's trying to, but it doesn't - not yet anyway.

I can't agree with that, you can't set a prerequisite on something that doesn't run natively on an OS. OpenGL supports pretty much everything DX10 does, so technical prerequisites are all there - whether the devs opt to use it or not, well, that's another matter entirely.

Sorry, I meant a prerequisite for gamers considering using Linux (not being able to run the latest shiny games). This of course isn't a barrier for developers porting to Linux natively.

How many games requiring only DX10/11 do you know? I know only know about Bioshock Infinite, Metro 2033 and CoD according to you (I don't know, I don't dabble in it). There may be more but I haven't the foggiest.

In the context of GoG support for Linux, DX10/11 is quite irrelevant