Linux market share on Steam now at a 16 month high after a rounding error was fixed

With these rates we will have total world domination in no time. ;)

Imagine, in a year we'll be at 16.8%, and in two years we'll have a market share of over 470%!

Or maybe my math is a little off, that's a possibility too.

https://www.xkcd.com/605/

Nice avatar, by the way. Haven't you changed it, like two times in a couple days?

I've updated our dedicated page tracking it

Wow. I hadn't realised how much of a difference this revision made until I saw that. Truch be told, last month's “old” increase wasn't really all that much bigger than August's, although that was a pretty big jump too. But this really stretches it out.

What I don't think has been mentioned here, although it's clear from the chart, is that among English speakers the uptick over the last two months has been huge, almost doubling the Linux share. We're now comfortably over 2% in that (pretty large and important*) demographic. I don't know what the actual figures are these days, but that's getting close to what I think of as “Mac territory”.

*Apologies to speakers of other languages, but it's true.

Yeah, well, but I have Steam's language set to English (easier for bug reports and co), so it might be a little biased.


What I would like to see is Valve improving these surveys. But maybe the statistical significance of Linux users isn't good enough?

Network speed, more filesystems, break down distros a bit more (maybe Ubuntu 18.04, ubuntu-based, Arch-based, etc). Filesystem. Kernel version. Proprietary driver vs OSS, mesa version, AMD vs nVidia vs Intel. Wayland Vs X.
You know, stats that could help the developers make choices, and that we could point at when wondering about the Linux Gaming landscape (other than this awesome site, of course).

Didn't they reveal that os share is calculated automatically and is independent of the surveys?

Maybe, but not that I remember hearing. Source?


Last edited by Purple Library Guy on 4 Oct 2018 at 7:02 pm UTC

Yeah, well, but I have Steam's language set to English (easier for bug reports and co), so it might be a little biased.

That kind of proves my point, actually. I didn't mean to suggest that we native speakers are more important than everyone else, but really just that: “Steam clients set to English” are a very big market.

With these rates we will have total world domination in no time. ;)

Imagine, in a year we'll be at 16.8%, and in two years we'll have a market share of over 470%!

Or maybe my math is a little off, that's a possibility too.


I mean, if all those ex-PUBG players just washed right back out again as they stopped playing PUBG, most of them presumably never bought another game.

Why do you think that ? What does make the PUBG players so special ? I have no idea and do not know much about PUBG. :)

To make things very clear, i am talking about their alleged characteristic of non-other-game buyers only.

PUBG was a [fad](https://www.urbandictionary.com/define.php?term=fad) which attracted a lot of people including folks that might not normally be gaming much (read: not spending much money on games).


Last edited by Brisse on 5 Oct 2018 at 12:11 pm UTC

I mean, if all those ex-PUBG players just washed right back out again as they stopped playing PUBG, most of them presumably never bought another game.

Why do you think that ? What does make the PUBG players so special ? I have no idea and do not know much about PUBG. :)

Apparently a whole lot of the influx of new Chinese users to Steam was people who came to play PUBG, which was a huge fad. Theoretically, they might have bought other games on Steam once they had arrived, but if they did that presumably they would have stayed on Steam to play those other games after they were finished playing PUBG. But Liam said,

I think one of the big reasons is also the major decline in people playing PUBG. Look at the "Players every day" graph on SteamDB and see how far it has fallen, which is big considering it was a game previously pushing the "simplified chinese" use up high. 3.2 million in January to around 1 million now.

So it seems a whole lot of the people who came to play PUBG in fact left again once they were finished playing it (and since they were all Windows users, Linux share rose as a result). But if they left Steam once they were finished playing PUBG, that means they didn't buy other games and stick around to play them. QED.

Yeah, well, but I have Steam's language set to English (easier for bug reports and co), so it might be a little biased.

That kind of proves my point, actually. I didn't mean to suggest that we native speakers are more important than everyone else, but really just that: “Steam clients set to English” are a very big market.

If Linux users tend to set their language to English (and Windows users way less so), it actually makes the English-only statistics unrepresentative.

Is there a reason Linux users would be more likely than Windows users to use Steam in English rather than their own language?


Last edited by Salvatos on 5 Oct 2018 at 6:51 pm UTC

Didn't they reveal that os share is calculated automatically and is independent of the surveys?

Maybe, but not that I remember hearing. Source?

I can't seem to find the gol article quoting someone about this. @liam, do you remember this?

Didn't they reveal that os share is calculated automatically and is independent of the surveys?

Maybe, but not that I remember hearing. Source?

I can't seem to find the gol article quoting someone about this. @liam, do you remember this?

Just to clarify, I don't find it hard to believe. I've seen plenty of plausible speculation that Valve must have the accurate data somewhere, which they could release instead of the survey info, but I don't remember anything about them actually doing it.

If only they'd fix the damn survey bug... You know, the one that doesn't exist.

*Rolls eyes*

I actually just got the survey today for the first time in over 2 years.

I actually just got the survey today for the first time in over 2 years.

The last time I had it was approximately six years ago - back when I ran Microsoft Windows 8... Of course, if you open the survey file and check the date, it thinks I complete one at the start of this year; the same is true for the rest of the Linux computers in our house (but different dates).

Yet according to Valve, it's all a conspiracy theory, because "the survey works as designed".


Last edited by Cyba.Cowboy on 6 Oct 2018 at 5:18 am UTC

The last time I had it was approximately six years ago - back when I ran Microsoft Windows 8... Of course, if you open the survey file and check the date, it thinks I complete one at the start of this year; the same is true for the rest of the Linux computers in our house (but different dates).

Yet according to Valve, it's all a conspiracy theory, because "the survey works as designed".

How often have you been asked which party you're going to vote for lately? Still, there's quite accurate data about voting estimations.

Is there a reason Linux users would be more likely than Windows users to use Steam in English rather than their own language?

I don't know why people are not regularly using their language in the first place, so... yeah, I don't know.

I set my OS to French without thinking a lot about it but i sometimes wish i had not. (Well, it sets your keyboard automatically).

Can't you just change it now?

Now why would someone set the system and every application to English instead of their native language? For me it’s because the translations always suck and it’s easier to understand things in English (it’s also true under Windows btw).

I did mean to ask about Steam specifically since that is the part that would influence the survey :) I can see plenty of reasons to set the OS to English. The few times I saw Steam's French interface, it seemed very well translated. Which is surprising, in retrospect, since I think I remember them crowdsourcing their translations to volunteers.

Probably some game with a bad French translation insisted on using the same language as Steam?

Ah, that issue sounds familiar, yes. I think Hand of Fate did that at launch, but with the system language. Maybe it was a different game, but I distinctly remember something giving me no reasonable way to set the language.

Though again, I don't think any of that would differentiate Linux from Windows use cases and skew the survey.

How often have you been asked which party you're going to vote for lately? Still, there's quite accurate data about voting estimations.

I beg to differ, because the voting polls often vary wildly from company to company (that conducts the polls)(in Australia at least)... They are a rough guide at best - but they also do not claim that I have participated when I have not.


Last edited by Cyba.Cowboy on 6 Oct 2018 at 10:58 pm UTC

How often have you been asked which party you're going to vote for lately? Still, there's quite accurate data about voting estimations.

I beg to differ, because the voting polls often vary wildly from company to company (that conducts the polls)(in Australia at least)... They are a rough guide at best - but they also do not claim that I have participated when I have not.

Polling seems to have gotten a lot worse over the last decade or two. Seems to have been a combination of overuse and the internet. Overuse, doing lots and lots of polls for more and more trivial reasons, or on elections trying for more and more frequent polls, "rolling" polls and so on, all give strong incentives to do it on the cheap, with less attention to randomization for instance. And the internet furnishes a way to do bad, poorly randomized polls very cheaply. So they've rushed to save money by doing crap polls on the internet; result, crappy predictions.
Then there's push polls; they sometimes report the results of those as if they're the real thing, hoping to influence people in the direction the push poll is designed to point. Of course, those don't track reality very well because that's not what they're for.

I beg to differ, because the voting polls often vary wildly from company to company (that conducts the polls)(in Australia at least)... They are a rough guide at best - but they also do not claim that I have participated when I have not.

Do you have data about these wildly differing results? These are for the next state elections here (Bavaria)...
http://www.wahlrecht.de/umfragen/landtage/bayern.htm

I don't understand your last phrase... Did anybody claim Steam polls are querying every user?!?

Do you have data about these wildly differing results? These are for the next state elections here (Bavaria)...
http://www.wahlrecht.de/umfragen/landtage/bayern.htm

I have no idea how the political system works over there and to be honest, I don't really understand too much about our (Australian) political system (only the American system makes sense to me, for unimportant reasons)... But I do know that in Australia, when they publish their polls in the news every week or every time a politician does something "newsworthy", you'll see wildly different results depending on which company the news is using for their source.

Maybe it's different in your country, but that's how it is here in Australia and that was my point - the Steam survey is similar in the sense that one should take its findings with a grain of salt, for a laundry list of reasons (one of which being the well-publicised Linux bug)

I don't understand your last phrase... Did anybody claim Steam polls are querying every user?!?

The "survey" file (which indicates the date the survey was last supposedly run) seems to think that three of the different computers running various versions of Ubuntu in our house and used by different people have all had the survey in the last twelve months, when in reality, none of us have... That would be Valve Software claiming to query me, when it has not in fact done so.

And no it's not "user error" (e.g. accidentally closing the survey) - we have a limited number of computers in the house running an Apple mac OS or Microsoft Windows-based operating system, all of which have been able to complete the survey just fine...


Last edited by Cyba.Cowboy on 7 Oct 2018 at 8:42 am UTC