Same goes for the Source engine that - until now - had been dx only

No kidding, until my dad can run linux without him pestering me to death over the phone it isn't a viable platform for entertainment developers. Let's face it they need to appeal to a broader market. Not to mention the simple fact that DirectX is light years easier in terms of development. The bottom line? OpenGL needs to grow up, and linux needs to move beyond its "I don't pre-can squat" attitude.

Chicken and egg scenario.
There aren't many gamers on Linux because there's no games.
There's no games on Linux because there aren't many gamers.

Not to mention that most people who have Linux also have Windows, so it's a hard sell to sell a separate Linux version of a game (though most games that support Linux generally do it at no extra cost). Not to mention that Wine does a pretty good job on a heck of a lot of games.

Honestly, I think it would benefit the gaming industry in the long run if game developers took the hit now and ported their games to Linux. At the very least, work with Wine (or Cedega's new program) to try to ensure compatibility.

Funny enough, Steam and many of its games already work pretty well under Wine, and Stardock and its games aren't half bad as well. (though not on par with valve's stuff under wine)

Anyhow, Steam is probably coming to Linux for 2 reasons:
A significant enough number of people (probably a few percentage points) were identified as using Wine through valve's hardware profilers and...
They already had to rewrite their engines in OpenGL to port the Orange Box to PS3.

1.  Reading the article it's the Source engine they're looking to port.  Not Steam.  This means we could see a linux compatible version of Half Life 2 at some point.  Not necessarily that we will see a Steam client for Linux.

2.  Games supported under WINE/Transgaming etc. aren't usually done by the game developer themselves.  Specifically in the case of Cedega, they have to patch in game-specific support.

As a part-time linux user myself, I've come to accept the fact that linux is not destined to be a gaming OS.  Until either developers abandon DirectX, or someone figures out a 100% painless DX port for linux, you won't see a big move on linux games.  Why?  Because transitioning from a DX based engine to an OGL one is not in the least bit trivial.  iD can do it because I believe their games are done in OpenGL to begin with, so getting it to run on Linux is a much simpler task for them (by comparison).  UnrealEngine is built for both DX and OGL.

To get developers porting games to linux, there has to be a guarantee on the return on investment.  If it takes 1 full time developer a year to port some game, then that game has to at least sell enough copies to cover the cost.  To make it actually worth the time though it would have to make a lot more money than the cost to develop, otherwise it's a better value to have that developer work on the Windows version which is a better financial bet.

The platform needs a few big-name champions to make it viable, but in a market where a big-name game can cost in the millions of dollars to develop, that's a risk not many companies are generally willing to take. 

 

Agreed, but there is something you are overlooking in the financial planning. This article explains it quite well:
http://blog.wolfire.com/2008/12/why-you-should-support-mac-os-x-and-linux/
Especially the 3rd point, vocal minorities (or: word of mouth) is very important I'd say.

Of cause it is very costly to port an existing dx-only game to ogl. But I don't see any reason why a game developer should start with dx to begin with. SDL and OpenGL are just as powerful and offer easy access to graphics cars as well as the sound system and controls across many platforms. Chosing dx for the engine means that the developer willingly locks himself in and can never again change the system without huge cost (which is basically the whole reason why MS is making money, because most of the market is locked in and it is very costly and in many fields simply not possible to get out).
Is it just me or does that look like a very, very bad long term strategy? If there is any possibility to avoid some lock-in effect a company should take it in my opinion. It may not make a difference today but 10 years from now you might be very happy.

So yes, porting GC2 or Sins to Linux is probably not possible. But you can still release your next game with an OpenGL engine

Windows is a good gaming OS, but it is total crap for doing serious work. It's a toy OS. Just to play a GC2 game, I should reboot to Windows. Now, while in Windows, I cannot alt+tab to the more serious applications that I use under Linux. This makes it a serious annoyance.

And here is the commercial reason to make games compatible with Linux. If there is a Linux version of a game, I am prepared to pay good money for it, since it makes enjoying the game to so much more comfortable.

The problem is that the size of the PC gaming market is known.  With linux, who knows?  Developers and publishers can make a reasonable estimate of the potential customer base on Windows for any given game they're going to release.  And you make the point that most Windows users aren't gamers, so the same would hold true for Linux gamers.

Lets say for a moment that there are 1,000,000 Linux Desktop Users world-wide. (A low number I know, but it's an even one for the purpose of example)

• Now, we'll chop off 10% straight off the top to account for those who are purists and won't pay for anything that's closed source.  You now have 900,000 potential customers
• Of course, the majority of Linux users will be like Windows users and not be gamers.  From what's left, lets take away 75%.  You're at 225,000 potential customers now.
• Next you have to account for game tastes.  Strategy games tend to appeal to a small chunk of the gaming population.  Lets be generous and say a third of all gamers are strategy fans.  So chop 66% off of what's left.  76,500.
• Not every gamer buys every game that comes out in a genre either.  Again lets assume that a given strategy game is REALLY popular.  Half of the strategy fans pick the title up.  38,250 Customers.

Playing with the number of Linux users in total you get different final numbers (3mil total ends up with 114k at the end).  But my estimates above are probably aggressively positive (I don't have sales figures, I'm just carving up the total based on guesses.  I'm probably assuming more users will buy than actually will).

So we've defined a, by comparison, ridicuously small potential audience for a game on Linux.  But remember it's a guess, no one really knows how these numbers actually pan out, and with the costs of developing games, and later supporting them are skyrocketing it's hard to throw a lot of money at a complete unknown.

And honestly, I feel one of the single largest barriers to getting games onto Linux is support.  Oh my lord supporting software on Linux is a logistical nightmare.  There is no single base configuration you can count on.  One can't assume that anyone will have X version of Y library, or that they haven't done something hacky to circumvent a problem on their system.  The state of Linux video card drivers is also a shakey thing.  They lag behind Windows drivers as a general rule.  Same issue with sound.  Toss in all the different distributions there are out there that behave very differently, and you have a test and support matrix that is crazy.  

The 29 million number does not specifically target desktop users and the number is outdated from 2005, which was before the Linux market size wrote numbers in billions of $$. I think web browser statistics are more realiable, which show the Linux market share to be around 1%. As the total PC desktop market is something in the order of 600 million desktop users, the number of Linux desktop users should be about 6 million.

The issue about supportability is already well tested in the professional world. We see that some distributions are officially supported by most software vendors, mostly SuSE & Red Hat. Most software vendors are still helpfull when users run a different distribution, as long as they have the knowledge to perform specific instructions ordered by support people (i.e. please upgrade package X, please set environment variable X like this).

Linux support is not really much harder from Windows support, but very different. For example, handling drivers issues can be very easy on Linux, because the kernel comes with a set of high quality drivers. A too old kernel can be easily diagnosed by a support engineer and followed by an instruction to upgrade the kernel.

The main problem regarding support is that the infrastructure does not exist. Many publshers have support teams in many countries, which do not have people that have the skills to diagnose typical Linux issues. This makes supporting Linux more expensive. ID Software evades this problem, there is no Linux support on the CD by default, the publisher hasn't sold Linux support, so don't bother the publisher with it. You can download the Linux exe from ID's website and use the forums in case you have problems.

The development of a game is a big "loss" and every sale contributes to turn that "loss" into a profit. Linux sales don't contribute any less to this than Windows sales. The question is how much additional costs have to be made to get these additional sales in.

If middleware like OpenGL, OpenAL, SDL and so on has been used to write the game, the additional development effort can be negligible. In such case, it can become a very good business decision to release Linux and Mac versions, especially because Mac support often comes for free once it runs on Linux.

However, if the coders have programmed themselves into a corner, the effort to port a game can be huge and I fully understand that the Linux gaming market is too small for such efforts.

Well, fixing/improving Wine's DirectX emulation on top of OpenGL by programmers that are unknown to the Wine source code, doesn't sound an easy port, and therefore, it won't be cheap. It would be a different thing if Galciv2 ran close to perfect under Wine, but alas it does not.

But basically the choice you have is reimplement the engine in OpenGL, or link against Wine's Direct3D implementation. The Wine route would probably still be the most realistic one for a game like Galciv2.

Im with you 100% there. I have games on one of my linux machines, but tbh i only have them on there because i can't get them for windows! Its amazing how good free arcade games get when there is nothing 'better' to compete! I do not see linux as a gaming platform, because it simply isnt good at it.The only way for gaming to come across the OS's is for some really clever dev team to get loads of funding to create a better way to implement the 3D graphics in games. Maybe intels new ray graphics will do this? And TBH if i have to pay £30 to intel to put cutting edge games on my linux boxes id be chuffed!

Guys, find someone with a GP2X, take a look at it and think again wether Linux is unsuitable for games. I never had so much fun with a handheld as with my GP2X. And yes, there exist commercial closed source games for the GP2X (which allthough marketed for the GP2X are technically generic Linux games for the ARM processor).