I keep reading comments that Linux gaming is pretty much stagnating and not worth investing in because it is still at around 1% in the Steam Hardware Survey. So I decided to try and find some numbers. Unfortunately, there isn't all that much data publicly available, if you have additional or better data, I'll be happy to add them. Also, if you find any mistakes please let me know.
In order to quantify the state of Linux gaming, I tried to find answers to the following:
How many Linux gamers are there, how much do they buy, what kind of growth do we see in Linux gamers and games?
Steam
Steam has the most data available and also the largest number of games and gamers, so I'll look at this in most detail. The Steam survey gives a good starting point for how many Linux gamers there are.
How reliable is the Steam survey?
Valve hasn't published their methodology, so it's hard to tell. I often read Linux users claim the survey is less likely to appear on Linux, because it only ever shows up when they boot into their Windows partition. I've read similar complaints unrelated to Linux. I suspect that you are more likely to get the survey if you log in on a system that hasn't been connected to Steam in a while (or never). I haven't had a survey on my Ubuntu PC in some time, but a while ago I logged in from my Fedora laptop which I don't usually use for gaming and the survey popped up. Other users here have reported the same behaviour. This might also explain the initially larger Linux shares around March 2013 when many Linux users used Steam for the first time on their Linux boxes and were thus more likely to receive the survey.
What we do know though is that the hardware survey doesn't show up in Big Picture Mode so SteamOS is currently excluded. It is hard to estimate how much of a dent that makes since there is no data on how many people run Steam Machines in BPM.
Assuming that the 1% Linux share is roughly correct, the total number of Linux gamers has grown quite significantly since 2012 simply because the total number of Steam users has grown significantly as shown in the figure below (Note: Some of the sites talk about registered accounts, but as far as I know Valve reports active accounts.).
You can find the links I used at the end of the article. Additional older data can be found here.
The first time Linux appeared in a survey was in January 2013. At the time, Steam had about 54 million users, meaning there were about 540,000 Linux gamers. The most recent number of Steam accounts I could find was for February 2015, at which point there were 125 million active accounts, thus 1.25 million Linux users. I did some fitting and extrapolation of the data, and come up with 160 - 190 million accounts as of March 2016. Take this with a huge grain of salt though, this kind of growth can't continue indefinitely. Whatever the real number, the number of Linux users on Steam has kept up with the overall growth, and should at this point be larger than 1.5 million unless Steam stopped growing at all.
Also shown in the above graph is the number of Linux compatible games on Steam. This number is growing at a comparable rate to the number of users, even slightly faster. Overall it seems Linux on Steam is in a fairly decent state; no mas migration from Windows, but a steady growth in games and gamers.
GOG
Unfortunately, GOG.com has, to my knowledge, never released any data on user or OS numbers. Looking at the GoL survey retailer statistics, about 90% of participants buy games on Steam. That means there are about 10% who only buy on GOG, Humble or other stores.
Total
Combining the above information, I would estimate the total number of Linux gamers at 1.6 - 2 million not counting SteamOS in BPM.
Probably more important for developers is how many games Linux users buy.
Humble Bundle sales
Cheese has a very nice collection of Humble Bundle sales data from May 2010 to February 2016. Looking at the combined results, Linux accounts for 4.7% of purchases and 6.9% of payments. Note that this data goes back to before Steam or GOG supported Linux, and Humble helped bring many games to Linux.
Sales data from developers
A number of developers have shared their sales data. I used the data I could find here on GOL, and combined it in one plot:
( Some notes: For Democracy 3 the revenue is plotted rather than sales; for Trine 2 the revenue was given as 4.2% while the sales are at 1.9%; the value for Awesomenauts is over 4 months; Defenders Quests gives 7% for lifetime sales from their website.)
Based on these data, the average share of Linux sales is (3.2 +/- 0.4)%. The median is 2.7%.
Plotting the same data as a histogram
shows that the maximum is between 1 and 2%. One thing to keep in mind is that for many games the Windows version has a head-start, for example Dust: An Elysian Tail was 9 months late on Linux.
Both the Humble data and the data from developers shows that on average, Linux users buy more games than the 1% that the Steam survey would suggest. This does not necessarily mean that the Steam survey is wrong though, it could simply be that since fewer games are available on Linux, Linux gamers are more likely to buy the games that are available. In addition, I suspect that Linux users are relatively more likely to buy from DRM-free stores than Windows gamers, meaning that the Linux share on GOG, Humble, and developers' websites would be higher - as shown by the Humble Bundle statistics and seems to be the case for Defenders Quest.
All in all, looking at these numbers, Linux gaming has undergone an impressive growth over the last years, it's anything but stagnating.
References
https://en.wikipedia.org/wiki/Steam_(software)
http://store.steampowered.com/news/3390/
http://store.steampowered.com/news/4502/
http://www.pcgamer.com/steam-is-doing-quite-well-100-sales-increase-40-million-registered-accounts/
http://www.pcgamesn.com/steam-has-50-million-users-5-million-playing-concurrently
http://www.engadget.com/2013/10/30/steam-surpasses-65-million-users/
http://www.engadget.com/2014/01/15/steam-has-75-million-active-users-valve-announces-at-dev-days/
http://www.gamespot.com/articles/steam-reaches-100-million-users-and-3-700-games/1100-6422489/
http://kotaku.com/there-are-over-125-million-steam-accounts-1687820875
Linux games on Steam
http://www.geek.com/games/steam-has-15-linux-compatible-games-so-far-1520713/
http://www.pcgamer.com/steam-begins-listing-linux-system-requirements-on-some-game-pages/
http://www.pcworld.com/article/2098972/steams-linux-game-count-explodes-in-one-year-big-publishers-still-absent.html
http://www.cupoflinux.com/SBB/index.php?topic=714.0
http://fullcirclemagazine.org/2014/06/09/steam-hits-the-big-500-for-linux-games/
http://www.pcworld.com/article/2824526/steam-for-linux-tops-700-games-as-big-name-games-increasingly-call-it-home.html
http://www.phoronix.com/scan.php?page=news_item&px=SteamOS-Linux-Games-1K
http://news.softpedia.com/news/steam-for-linux-now-has-more-than-1300-games-487629.shtml
http://gameranx.com/updates/id/29889/article/steam-has-1-400-linux-ready-games-two-months-before-steam-machines-launch/
http://news.softpedia.com/news/1-in-4-games-on-steam-now-have-linux-support-490087.shtml
https://www.phoronix.com/scan.php?page=article&item=steam-1500-linux&num=1
http://www.phoronix.com/scan.php?page=news_item&px=Steam-Linux-Beta-Three
https://www.phoronix.com/scan.php?page=news_item&px=Steam-Linux-1700
http://www.linuxtoday.com/high_performance/steam-on-linux-hits-1800-games-available.html
https://www.gamingonlinux.com/articles/no-linux-is-not-at-1900-games-on-steam-we-didnt-get-100-games-in-nine-days.6573
Personally, I think the share is more important than the absolute number, because there is this wrongful assumption a lot of Linux gamers that is there are a million and a half active Linux games on Steam, that means that somehow they will all magically buy each new game and that will somehow make developers prone to porting their game. Xpander's suggestion above makes much more sense, though I think that 3% of sales is still too generous. The most basic and probably accurate assumption is that sales are more or less relative to market share. So that 3% is probably what a Mac port get from the overall game sales. So if we assume it's actually 1% of 200k - you can see why a lot of developers wouldn't bother.
The solution to the problem is Valve giving developers incentive to port for Linux. And help, lots and lots of help.
There are plenty of aspects of the Steam Hardware & Software Survey results which definitely aren't statistically insignificant, but it does draw attention to some things that are. I'm not interested in engaging in an argument via proxy, but I am happy to stand by the calculations under the Insignificant Fractions and Margins of Error headings in the full article and enjoy chatting with my readers - especially if it reveals something that I've missed.
Whether the sampling is reliable/not buggy and whether the aggregation is appropriate/not flawed is a separate issue IMO.
Don't get me wrong - I'm very glad that Humble helped provide incentive and desire for developers to pursue and investigate Linux in their earlier days (putting aside for a moment that plenty of those developers were pursuing Linux support regardless of bundle inclusion), but if they are indeed misrepresenting the hard work of developers outside of Humble as work of their own, that is a separate issue.
I did have a typo in that post - I'd meant to write "on that front" (regarding the shortcomings which I believe undermine the survey results) rather than "on the front".
The change in the ratio of Linux users to Linux titles on Steam is what I was alluding to when I mentioned that the number of available titles had increased dramatically. I think that kind of adjustment you're exploring there is only fair when considering the potential market size (which is all the survey could ever indicate).
I'd always expected that we're quickly arrive at a transitional period where Linux gamers would no longer be purchasing every title and instead (hopefully) only purchasing ones they like. I've found it hard to keep track of whether this has had a real impact on developers' experiences. I like to pretend that it's an indication that the market size has grown enough to compensate ^_^
I feel like it's also worth taking into consideration the kind of work culture and attitude towards rigorous quality control. There's a strong likelihood that the survey is one person's passion project and that nobody beyond them really checks to see if it's working (the sampling bug I tried to report in 2011 and had presumably been around since July 2010 went over a year before it was noticed and corrected).
A friend of mine who worked at Valve at the time went on a hunt to find the person responsible for the survey (I don't know their identity) and passed on my full article. Other than calling me "wordy," the only thing they had to say was something along the lines of, "I’m pretty sure we’ve mentioned some details publicly about our methodology (number of samples, what users and when we survey them). I can’t find a link in 60 seconds of googling."
In spite of the acknowledged difficulty of finding details of methodology, nothing further was published from them :(
Never asked you to join an argument, was just pointing out someone who believes strongly on the other side.
Sorry, I wasn't meaning to suggest that you had. It was more to ensure that there wasn't misunderstanding about my intentions with responding.
My point of view on this is that most of the AAA developers doesnt let the Linux community grow.
The people who want to start or at least try to use Linux with every AAA title they just say "fuck this" and go back to Windows just to play the NEW AAA Title.
How many NEW AAA Games did Linux have in 2016 ? as far as I know Only XCOM 2 ?
Most of the PC Gaming sites have around 20-25 Anticipated games for 2016, from all that games maybe only 2-3 will get a Linux release and maybe on day 1.
Gamers doesnt care about OS
They care about two things --> Performance and Availability
That is what Valve is trying to do with Vulkan.
Valve already knows that and when they are ready , they have many ways to advertise how Linux is better than Windows but they are not ready yet.
Imagine in a Dota 2 sponsored by Valve Final tournament the winner says
" I am using SteamOS because it gives me better performance, stability and no lag issues ,is incredible how fast it is comparing to Windows"
if Vulkan Delivers the performance we need and developers start supporting Linux from day 1 then the Linux Gaming community will grow in a snap.
Last edited by wolfyrion on 14 March 2016 at 3:09 pm UTC
Iv got it once or twice in windows. But I got that survey alot more in linux.
I dont know the logic behinde the apperience of the survey