Linux hits exactly 2% user share on the October 2024 Steam Survey

Also very interesting that the "Language(Linux)" information seems to be wrong, with English sitting at 97.44% and Russian, German, Simplified Chinese all at >4% alone we have more than >100% of languages used on Linux.
I wonder if that is just an error on Valves side or maybe Valve is now counting all enabled languages for Linux systems (in that case values above 100% would obviously make sense).



Last edited by Lamdarer on 2 November 2024 at 3:05 pm UTC

Quoting: LamdarerAlso very interesting that the "Language(Linux)" information seems to be wrong, with English sitting at 97.44% and Russian, German, Simplified Chinese all at >4% alone we have more than >100% of languages used on Linux.
I wonder if that is just an error on Valves side or maybe Valve is now counting all enabled languages for Linux systems (in that case values above 100% would obviously make sense).

To protect your privacy, external media requires approval to load. Source: i.imgur.com
View cookie preferences.
Accept & Show Accept All & Don't show this again   Direct Link

Does it count people who use more than one language when playing games?

A strange motable thing this month is:

"SteamOS Holo" 64 bit 36.79% -2.41%

AMD AMD Custom GPU 0405 35.72% +9.76%

AMD Radeon Graphics (RADV VANGOGH) 19.40% +6.01%

Custom 0405 is the SD LCD and the other is SD OLED. So SD is actually at 55.12%. Now for the first time more than half of linux users. Not the 36.79% that Holo suggests.

It is not easy to guess why. The difference is too big to be hidden into other.

Actually something is very wrong with the GPUs. They add up to more than 100%.


Last edited by lucinos on 2 November 2024 at 4:08 pm UTC

Quoting: lucinosA strange motable thing this month is:

"SteamOS Holo" 64 bit 36.79% -2.41%

AMD AMD Custom GPU 0405 35.72% +9.76%

AMD Radeon Graphics (RADV VANGOGH) 19.40% +6.01%

Custom 0405 is the SD LCD and the other is SD OLED. So SD is actually at 55.12%. Now for the first time more than half of linux users. Not the 36.79% that Holo suggests.

It is not easy to guess why. The difference is too big to be hidden into other.

Actually something is very wrong with the GPUs. They add up to more than 100%.

Argh. I don't expect statistics based on sampling to be a perfect representation of the territory, but it would be nice if they were internally consistent, like they at least represented a possible version of the territory.

Quoting: lucinosActually something is very wrong with the GPUs. They add up to more than 100%.

They've also all grown this month rather than some growing and some shrinking. Very peculiar.

Quoting: PyrateAnd I only received that prompt a couple months ago, despite being on Linux since last March.

It's really sporadic how often it pops up, and is not reflective of Real World usage... I've been gaming under Linux-based operating systems for close to 20 years, and I could (quite literally!) count on one hand the amount of times I've seen the survey - what's worse is that most of the times I have seen the survey has been on a "fresh" system or after installing Steam for the first time!

I was shocked a few days ago when I got the survey for my Linux desktop and deck on the same day. I've only ever had one survey for my launch day deck and I have not gotten a single survey for my Linux desktop in 4-5 years (even after several reinstalls and hardware upgrades).

Quoting: Naibvery nice... but :( starting to get enough points to fit to a sigmoid... if that inflection point is correct, are we looking at like 3% as the settling point
" />

Out of curiosity, what method did you use to fit that curve?

Quoting: Purple Library Guy

Quoting: lucinosActually something is very wrong with the GPUs. They add up to more than 100%.

Argh. I don't expect statistics based on sampling to be a perfect representation of the territory, but it would be nice if they were internally consistent, like they at least represented a possible version of the territory.

You're both forgetting hardware with dual graphics.

The percentages in that graphics mean "10% of sampled machines have this GPU model" which is related but not exactly the same as saying "10% of GPUs used in sampled machines are of this model"

As a simplistic illustration of this distinction, if 20% of machines have dual-graphics, the sum of each GPU's percentage is going to be 120%.

ps: In the real world there might not match perfectly, because there might be a couple multi-graphics machines with more that 2 GPUs, plus rounding errors.

Quoting: dvd

Quoting: Naibvery nice... but :( starting to get enough points to fit to a sigmoid... if that inflection point is correct, are we looking at like 3% as the settling point
" />

Out of curiosity, what method did you use to fit that curve?

The current perceived deceleration of growth might be noise (just looking at the graph we can clearly see the graph is quite noisy, especially with the quasi-periodic steam China windows growth spurs).

The graph fits well to a bi-linear (steam deck being the turning point between a previous trend and the current trend).

It might also fit well to an exponential... but only stock exchange economists still try to fit exponentials to product/market growth trends because being blind to market saturation, physical limits of the material world and etc is conducent to better stock trades...

And it might be (and this is my personal bet) a sigmoid but with a higher endpoint, distorted by the noise that's dragging stats down at the moment.