Steam's December Numbers Point To A Lower Linux Marketshare But With More Oddities

[Mez']

Can't tell if you're talking about Lutris? Epic is good with Lutris allowing EGS and games to play on Linux. Its really not that much different from Steam except one other application. Seems to be friendly enough and like Steam lots of games have problems.

I would like to play more games on Linux but even with Steam Proton there's too many problems so I don't.

I get the same amount of survey's on Linux and Windows so I don't think that's the problem. If you're just using one OS, you might feel like its broken when a lot of time goes by. They don't need to send a survey that often. Most people don't change OS types all that much.

Yes, I was talking about Lutris. It didn't even cross my mind that you would actually be playing on Windows. :-D

I really don't get why keeping Windows when we have Steam and already so many native games working really well that we can actually support. I'm not a huge gamer but the 70 or so native games I own (another 5-10 comes from Humble bundle so not intentional if they are not supported) work perfectly 99.9% of the time.

[enrico.tagliavini]

I've purchased so many Windows games this holidays thanks to Proton being so great. Possibly their statistics is also based on what kind of game version you download? On Linux when using Proton you still download the Windows version of the game.... I wonder if this information is used in the stats too. Anyway Steam china lunch is definitely a better explanation, also for Windows 7 sudden increase.

When a game has a quality native port, such as the last Tomb Raider series from Feral Interactive (which you should really buy, it's a great game and it's such a good port to Linux), it's amazing. But when this doesn't happen Proton comes to rescue in so many cases. It's sometimes not perfect, required a workaround or two, but most of the time it's nothing hard and the experience is quite good. I'm playing so many old games from the '90 and early 2000 that were mostly PS titles ported later to Windows only. Proton is great!

[Buntolo]

Where the hell did you get that linuxtracker statistic? That has to be 20 years old. For one, the site doesn't exist anymore. And two according to that, the numbers would be about 0.0002% market-share. That's complete nonsense. It's low, but jesus it's nowhere near THAT low. The other statistic you cited shows Linux at essentially 1 percent, nowhere near "one of the lowest linux adoption per capita countries." It's literally right in line with the average. So what are you even talking about? Did you just decide that was a thing? Also, that number is going to skyrocket in the next 2-3 years, considering the Chinese Government is banning foreign-made Operating Systems by 2023 or whatever, and it will obviously be Linux that replaces that (apparently UOS, a new version of Deepin, was made specifically to cater to that law). Windows will be outlawed on government computers and will of course be replaced with Linux, and this being China that will absolutely affect the public percentages too, as the government has an outsized influence there. But that's neither here nor there, because the stats you quoted for right NOW are nonsense, or don't make any sense regarding your point.

Edit: Just saw that linuxtracker map is from 6 years ago. Yeah, super relevant. That's a totally useless source.

1) 6 years old
2) Look at gs.statcounter.com:

android: 52.18%

Windows: 31.97%

iOS: 10.9%

OS X: 3.15%

Unknown: 1.46%

Linux: 0.32%

(December 2019)

Methodology: https://gs.statcounter.com/faq#methodology

It's not an idea of mine, it's the idea I get from tech-related websites:
https://www.extremetech.com/computin...ices-worldwide
https://news.softpedia.com/news/97-P...d-472110.shtml

[gardotd426]

1) 6 years old
2) Look at gs.statcounter.com:

android: 52.18%

Windows: 31.97%

iOS: 10.9%

OS X: 3.15%

Unknown: 1.46%

Linux: 0.32%

(December 2019)

Methodology: https://gs.statcounter.com/faq#methodology

It's not an idea of mine, it's the idea I get from tech-related websites:
https://www.extremetech.com/computin...ices-worldwide
https://news.softpedia.com/news/97-P...d-472110.shtml

It's 6 years old. And you think that's relevant? It's not. Also, those statistics include mobile operating systems, which has nothing to do with the desktop. When you look at desktop operating systems, which is the only way to actually take any of these and view them relevantly, Linux (excluding ChromeOS) makes up about 2 percent of the market. Desktop use for Linux in China is a bit lower, but not by much, it's a little over 1 percent. That's not "extremely rare" or anything like what your original comment claimed, and again, you can't use 6-year old statistics when it comes to technology, nor can you count mobile operating systems when this entire conversation is about desktop computers. By your statistic Windows is only a third of computers. TECHNICALLY, since smartphones are computers you could say that's true, but that has literally nothing to do with anything here, and it's completely irrelevant.

[Buntolo]

Ok, 6 years may be too much (but not that much in this case, since I doubt Linux adoption may have changed a lot). But the data from gs.statcounter should be, according them, pretty new.

TL;DR: according to gs.statcounter China uses +22.04% Windows 7 PCs and -1.33%. This means that in China Windows 7 is relatively ~50% more widespread and Linux is relatively under 50% less widespread.
This corroborates the general idea reported by web tech newspaper.

So let's do it again, using data from last 12 months on gs.statcounter:
Windows adoption in China
https://gs.statcounter.com/os-versio.../desktop/china

Win7      47.88%
Win10     45.14%
WinXP     3.94%
Win8.1    1.72%
Win8      0.68%
Win2003   0.46%

Windows adoption in Europe
https://gs.statcounter.com/os-versio...desktop/europe

Win10     68.14%
Win7      23.1%
Win8.1    5.14%
WinXP     1.62%
Win8      1.48%
WinVista  0.47%

Desktop OS adoption in China
https://gs.statcounter.com/os-market.../desktop/china

Windows    87.59%
OS X       8.64%
Unknown    2.96%
Linux      0.81%
Chrome OS  0%
FreeBSD    0%

Desktop OS adoption in Europe
https://gs.statcounter.com/os-market...desktop/europe

Windows    77.95%
OS X       17.38%
Linux      2.14%
Unknown    1.74%
Chrome OS  0.78%
SunOS      0%

[gardotd426]

Okay, here's the thing... First of all, browser-tracking counters as a method for calculating OS market-share is fundamentally flawed, and that's pretty much universally acknowledged. From the Wikipedia article on OS usage: "All such figures are necessarily estimates because data about operating system share is difficult to obtain; there are few reliable primary sources – and no agreed methodologies for its collection." So, this is all irrelevant. There is no consistent, scientifically rigorous standardized method accepted for determining these numbers. I could just end there, but I'll throw you a bone and still respond to the issues with your point outside of the fact that your statistics are objectively meaningless in any scientific manner (which is an inherent requirement when talking about gathering accurate statistics). First of all, you use Europe as the comparison, instead of Worldwide, which shows a much smaller disparity, as Europe has a higher than average percentage of Linux desktop adoption compared to the World as a whole, per your own source. So you took the numbers from the region with the highest rate of Linux desktop adoption, and used that as the overall average, which is ridiculous for reasons I shouldn't have to explain. Further, if you look at these statistics, you almost certainly HAVE to incorporate "unknown." In China, it's almost 3 percent, while in Europe, it's just over 2 percent, and that essentially makes up for any disparity in the Linux-reported numbers (even more so if you actually use Worldwide average for the Worldwide average, instead of using Europe for some reason). You have to take that into account, because it's obvious that a huge number of the "unknown" operating systems would be Linux, as Linux users are more likely to try and hide their computers from being reported, and on the other end Windows is much more unlikely than Linux to not be recognized by these stat counters. The stat counters are, after all, able to determine specifically which VERSION of Windows people are using, between like 5 different versions. Yet it can't do that with Linux distributions, and it's obvious that at least a large number of the unknowns are linux systems.

In short, there's no evidence whatsoever that China has a rate of Linux adoption that is among the lowest in the world, which is exactly what you stated in your original comment. There is objectively no evidence to support that, and even the evidence you use, when you use it properly, doesn't even state that. At worst, China is slightly below average compared to the worldwide rate and is much higher than many countries, and that's without taking into account the HUGE number of chinese computers that aren't connected to the worldwide internet, but instead are connected to the Chinese intranet. And this is by far a majority of computers in China. China blocks access to numerous places that this data is collected from, and in a country of over 1 BILLION people, that's just too big of a missing chunk of data for the statistics to be relevant even if they WERE otherwise reliable. Also, even if china is at 1 percent, and the rest of the world is at 2 percent, that would mean China has more Linux users than any other country on earth. at 1 percent, China would have 10 million, while the United States, almost certainly the second highest number of Linux users by nation (except for maybe India because again, 1 billion people) would have 6 million even if the US is at 2 percent.

I get why you thought what you claimed in the original comment, but there's literally nothing that has any legitimate statistical value that backs up what you're saying, and actually it's very likely the opposite, but there's no solid statistics that say that, either. But it being the opposite was never my point, my point was that the statistics are meaningless, and I think I've demonstrated that pretty substantively. It's cool, I've thrown out stats before I didn't realize weren't accurate, we all do it. But no, there's nothing whatsoever to back up your claim.

EDIT: The numbers I said for the US and China for overall users are based on total population, not total computer owners, but the ratio still holds up, let's say it's 2 million US users, it would be 3.5-ish million Chinese users. Computers and technology have massively transformed china, and they have an insane number of computers there.

[gamerk2]

Valve is doing a lot for the Linux community, and uses a lot of Linux to make things progress.
But personally as a long time linux gamer and not having any windows install except for when I need to reverse engineer drivers, native builds are a pain for the customer. And a really big ass pain. I can play any windows game (that is not even playable on recent windows) from 2000, but none of the original linux games I had on CD/DVD work.

And this is *exactly* why Linux will never be a gaming platform.

If the kernel devs actually made the API sacrosanct, did away with the mindset of "we changed the Kernel; recompile all your stuff or don't complain if it doesn't work", and actually threw their weight behind a singular distribution, then *maybe* Linux could start to claw into Windows market share over a decades timespan.

The entire mindset is wrong. Pretty much every Windows game ever released (aside from 16-bit stuff which explicitly isn't supported) still runs on Windows 10. Meanwhile, trying a run a native Linux game out of the box is damn near impossible. People don't want to bother; they want to install the OS, pop in a disk, and play.

Seriously, it is literally easier to take a native Windows app and run it through WINE then it is to try and keep a Linux app running years down the line. And that's *exactly* what you see happening; it's not worth the headache to support native Linux apps for the market share/sales they offer.

[Ardje]

Seriously, it is literally easier to take a native Windows app and run it through WINE then it is to try and keep a Linux app running years down the line. And that's *exactly* what you see happening; it's not worth the headache to support native Linux apps for the market share/sales they offer.

It is easier to run old windows binaries on linux using wine. It's not at all easy running them on recent windows. Both systems have bitrot and changes.
But wine changes the windows API to a higher layer of middle ware, making it a better windows than windows, especially for older games, as windows does not have a compatibility layer comparable to wine. You have windows on windows (wow32 and wow64) but that does not even come close to wine.
Windows has the same compatibility issues, but not as much as windows, as windows is not really a progressive operating system.

But I am with you on that it's easier to run a windows app on linux than to support a linux app years after, unless the engine is open source. Those that have opensource engines do live a long time... Quake is probably the second most ported game after doom. Quake world scale VR for gear vr comes to mind.

But back to gaming platform: This is exactly why linux with non-native games can be the gaming platform. Especially with containers. Because it will work long after windows becomes obsolete and Office 360 is Microsoft's main platform using web browsers.

[gamerk2]

It is easier to run old windows binaries on linux using wine. It's not at all easy running them on recent windows. Both systems have bitrot and changes.
But wine changes the windows API to a higher layer of middle ware, making it a better windows than windows, especially for older games, as windows does not have a compatibility layer comparable to wine. You have windows on windows (wow32 and wow64) but that does not even come close to wine.
Windows has the same compatibility issues, but not as much as windows, as windows is not really a progressive operating system.

But I am with you on that it's easier to run a windows app on linux than to support a linux app years after, unless the engine is open source. Those that have opensource engines do live a long time... Quake is probably the second most ported game after doom. Quake world scale VR for gear vr comes to mind.

But back to gaming platform: This is exactly why linux with non-native games can be the gaming platform. Especially with containers. Because it will work long after windows becomes obsolete and Office 360 is Microsoft's main platform using web browsers.

All I'm hearing is the same exact excuses I've been hearing for the past 20 years now.

By your own admission, it is easier to run Windows binaries on Linux through a compatibility layer then it is to run native Linux ones. That's the fundamental problem with Linux as a gaming platform.

I also note that for the most part, Windows doesn't need a compatibility layer to run old games. Outside of a few troublesome titles (which can typically work with a few modifications; pcgamingwiki.com is a godsend for those) most stuff still runs with only minor configuration changes necessary. For example, Diablo still runs fine if you disable the compositor, since the old GDI render still works. Another old example is a shitty game I love called Mask of Eternity, that runs exactly as badly as it did 20 years ago on both the Open GL 1.1 and DirectX 5 render (and Glide, via compatibility layer). Meanwhile, Linux can't run a two year old game without modification because someone downstream broke compatibility, and they don't feel responsible for fixing it.

[Ardje]

All I'm hearing is the same exact excuses I've been hearing for the past 20 years now.

By your own admission, it is easier to run Windows binaries on Linux through a compatibility layer then it is to run native Linux ones. That's the fundamental problem with Linux as a gaming platform.

I also note that for the most part, Windows doesn't need a compatibility layer to run old games. Outside of a few troublesome titles (which can typically work with a few modifications; pcgamingwiki.com is a godsend for those) most stuff still runs with only minor configuration changes necessary. For example, Diablo still runs fine if you disable the compositor, since the old GDI render still works. Another old example is a shitty game I love called Mask of Eternity, that runs exactly as badly as it did 20 years ago on both the Open GL 1.1 and DirectX 5 render (and Glide, via compatibility layer). Meanwhile, Linux can't run a two year old game without modification because someone downstream broke compatibility, and they don't feel responsible for fixing it.

I am hearing the same excuses from you. It's not easier to run old games on Windows, in fact it's much harder to run an old games on windows than it is to run it on wine/proton. Not only that. Proton gets easier and easier. I do see the windows API as a middle layer. "Disable compositor since the old GDI render" what? Did you hear yourself talking? When you have to do things like that it's not a viable gaming platform. No more excuses. We need a better proton to be able to game into the next century.
Linux native is not yet the way to go IMHO. But a generic middle layer originating from the windows API.

Windows itself is just a collection of legacy and incompatibility. I've left the windows developer world since it's frustrating place of people with no clue and plates before their eyes.

The only real frustration I have on linux is those systemd developers that have never looked outside their desktop, and hence force break compatibility with everything they've never seen, because non desktop users must be doing it wrong. This is frustration speaking trying to get wayland running using an ordinary distribution. But I guess I am better of trying to run something like tizen.