Linux user share on Steam continues the slow climb, SteamOS rises

Quoting: GuestI didn't have a good experience with Ubuntu on desktop (maybe it is just GNOME that makes me hate it), I believe it actually may repel newcomers from Linux, so maybe it is good to let it go away from Steam.

Quoting: CatKillerThe survey just looks at the LSB name, so that part's not going to change.

Yup - I'm on Xubuntu (I don't get on with GNOME, either, and KDE's a bit too heavy for my liking), and finally got pinged for the survey for the very first time recently, but naturally it all goes under the Ubuntu banner.

Quoting: PenglingYup - I'm on Xubuntu (I don't get on with GNOME, either, and KDE's a bit too heavy for my liking), and finally got pinged for the survey for the very first time recently, but naturally it all goes under the Ubuntu banner.

I think it's the right way for them to have done it: it's their branding, and you're using the same software from the same repositories whichever flavour you're using or even if you eschew desktop environments entirely. And it stops the distro stats being diluted over the half-dozen or so official flavours and many community remixes. But it does make Gnome seem more popular than it might actually be.

I think we need to start fitting an exponential rather than a straight line.

Actually, could you have multiple fits on the plot? I don't know what the back end is like.

Quoting: MohandevirAnd it doesn't count the Steam Deck users that don't really use the desktop client (like I do)... These won't get the survey because Gamemode/Gamescope doesn't trigger it, from what I read. So, the Steam Deck market share is probably higher than that.

Edit: Thinking about it, if you are on Windows and use Steam BPM... You won't get the survey either. The results are probably wrong from the start, on both sides of the fence.

This is not true as I received the survey in gamemode

Quoting: PhiladelphusI think we need to start fitting an exponential rather than a straight line.

Actually, could you have multiple fits on the plot? I don't know what the back end is like.

As someone that nagged Liam into putting the trend line on the graph, the simple least-squares line that goes through the mean is the appropriate one to use. It's just a visualisation of the data; it isn't going to show any particular function. Whether the data points squiggle around the trend line, say, or has early data below the trend line and later data above the trend line, helps to visualise what the data are showing.

Quoting: KohlyKohl

Quoting: MohandevirAnd it doesn't count the Steam Deck users that don't really use the desktop client (like I do)... These won't get the survey because Gamemode/Gamescope doesn't trigger it, from what I read. So, the Steam Deck market share is probably higher than that.

Edit: Thinking about it, if you are on Windows and use Steam BPM... You won't get the survey either. The results are probably wrong from the start, on both sides of the fence.

This is not true as I received the survey in gamemode

Happy to hear that. Read on an article that it wasn't so. Nice to know that they were wrong.

Quoting: CatKillerI think it's the right way for them to have done it: it's their branding, and you're using the same software from the same repositories whichever flavour you're using or even if you eschew desktop environments entirely. And it stops the distro stats being diluted over the half-dozen or so official flavours and many community remixes. But it does make Gnome seem more popular than it might actually be.

Yep, totally agreed, though I do sometimes wonder how the desktop-environment usage stacks up. It's purely anecdotal, but in gaming circles I've seen more folks who lean towards KDE first - not that there's any way of knowing this for sure, of course!

Quoting: PenglingYep, totally agreed, though I do sometimes wonder how the desktop-environment usage stacks up. It's purely anecdotal, but in gaming circles I've seen more folks who lean towards KDE first - not that there's any way of knowing this for sure, of course!

Yeah, actual statistics on Linux usage are really hard to come by. The ones that are there tend to line up as roughly a third KDE, a third Gnome, and a third everything else. Studies for school rollouts suggest that students tend to prefer KDE to other options. Red Hat casts a long shadow on workstation use, of course, and a lot of Ubuntu use is headless.

The statistics should always be taken with a grain of salt.
We dont really know what the conditions for the survey selection is, how many devices where surveyed in which month and what the bottom 37.89% as well as the 4.34% Flatpak runtime users are.
Could be that Arch actually dominates with other derivatives like Endevour etc on the other side there is also Linux Mint or POP!_OS which are Ubuntu Based as well.
But I think the trends matter, with so many more Decks shiped this month I thought that we would only see significant increases next month for Deck users looks like there is more to come.
Also about OSX "growing faster" than Linux, I thought that while it got a .76 increase this month thats more or less the amount it regressed last month.

But what bugs me the most is that while SteamOS Holo has 13.69% for Linux, the AMD Custom GPU has more 13.72%. Wouldnt that mean that there is a significant portion of Deck users who installed another Linux Distro on it and even more so if you consider that the Community Holo iso which can be installed anywhere also counts toward SteamDeckOS™.

Quoting: LamdarerCould be that Arch actually dominates with other derivatives like Endevour etc on the other side there is also Linux Mint or POP!_OS which are Ubuntu Based as well.

Arch derivates have their own /etc/os-release (or whatever) and don't appear as Arch.

Quoting: LamdarerBut what bugs me the most is that while SteamOS Holo has 13.69% for Linux, the AMD Custom GPU has more 13.72%. Wouldnt that mean that there is a significant portion of Deck users who installed another Linux Distro on it

People got bored and started doing other way around😂 Anyways, there a few different use cases for using non SteamOS but it doesn't matter.

Quoting: LamdarerBut what bugs me the most is that while SteamOS Holo has 13.69% for Linux, the AMD Custom GPU has more 13.72%. Wouldnt that mean that there is a significant portion of Deck users who installed another Linux Distro on it and even more so if you consider that the Community Holo iso which can be installed anywhere also counts toward SteamDeckOS™.

Desktop use aside, 13.69 : 13.72 = 0.9978 => 0.22 % of Steam Desk users would be installing another Linux distribution. One in 450.

Quoting: mr-victory

Quoting: LamdarerCould be that Arch actually dominates with other derivatives like Endevour etc on the other side there is also Linux Mint or POP!_OS which are Ubuntu Based as well.

Arch derivates have their own /etc/os-release (or whatever) and don't appear as Arch.

Exactly, even though one could argue that e.g. Endevour is closer to Arch than Ubuntu LTS to non LTS etc (not to mention Mint), and with ~40% other I would be very carefully with statements about which Linux Distro is "the most widely used"(even more so when you consider the other factors I mentioned), still it is a good indicator in general and I would love to see developers adding "Most modern Distros", "most Ubuntu 22.x. and Arch based Distros" or "ubuntu or SteamOS or Simular" to the System requirements section instead of just Ubuntu.

I was just thinking. I have my main gaming PC details on here, but I own a Steam Deck too. Would be nice to have an option to show you have both. As it is, the Steam Deck for me is a good way to show my friends what Linux gaming is capable of.. But I still prefer to game on my main rig most of the time.

Quoting: LamdarerBut what bugs me the most is that while SteamOS Holo has 13.69% for Linux, the AMD Custom GPU has more 13.72%. Wouldnt that mean that there is a significant portion of Deck users who installed another Linux Distro on it and even more so if you consider that the Community Holo iso which can be installed anywhere also counts toward SteamDeckOS™.

Another interesting thing...
Resolution 1280 x 800  12.45%
AMD Custom GPU 0405    13.72%

So ~9% of Deck Users had a monitor attached?

Quoting: CatKiller

Quoting: PhiladelphusI think we need to start fitting an exponential rather than a straight line.

Actually, could you have multiple fits on the plot? I don't know what the back end is like.

As someone that nagged Liam into putting the trend line on the graph, the simple least-squares line that goes through the mean is the appropriate one to use. It's just a visualisation of the data; it isn't going to show any particular function. Whether the data points squiggle around the trend line, say, or has early data below the trend line and later data above the trend line, helps to visualise what the data are showing.

My post was mostly a tongue-in-cheek way of conveying hype. Though just for fun I did my own least-square fit of a linear and an exponential function to the data, and – for this date range – the sum of the squares of the residuals was ever-so-slightly lower for the exponential, indicating that the data, on average, more closely resemble an exponential function than they do a linear one (though not by much).

Quoting: Philadelphus

Quoting: CatKiller

Quoting: PhiladelphusI think we need to start fitting an exponential rather than a straight line.

Actually, could you have multiple fits on the plot? I don't know what the back end is like.

As someone that nagged Liam into putting the trend line on the graph, the simple least-squares line that goes through the mean is the appropriate one to use. It's just a visualisation of the data; it isn't going to show any particular function. Whether the data points squiggle around the trend line, say, or has early data below the trend line and later data above the trend line, helps to visualise what the data are showing.

My post was mostly a tongue-in-cheek way of conveying hype. Though just for fun I did my own least-square fit of a linear and an exponential function to the data, and – for this date range – the sum of the squares of the residuals was ever-so-slightly lower for the exponential, indicating that the data, on average, more closely resemble an exponential function than they do a linear one (though not by much).

I'm a nerd. I'm not a nerd for this particular kind of stuff, but I find I appreciate it when I see nerds nerding out about stuff even when it's not my particular stuff, you know? One of life's simple pleasures.