Yeah, that’s a good point. The steam survey for OSes has always been relatively poor quality and untrackable. Ideally, they’d either put the resources into making it transparent and clear or stop it to not spread false data.
GamingOnLinux does a good job of tracking the data separated to just the English language, which shows how much effect China has on the survey as well as a decent approximation of the rest of the audience’s Linux usage, which is closer to 4% these days, post-Steam-Deck.
Steam Decks would still be at least 42% of Linux users though, it just depends on scale. I don’t understand why the Linux community wants to cut out Chinese users just because they are from China.
They’re good outliers to cut out from a sample set. If you want to use the survey as a developer to determine what your audience looks like, how much merit is there in factoring in a player base that mostly only plays Dota 2 and PUBG for major patches or something? And don’t forget that the other way to look at that sample set is that for every Steam Deck customer out in the wild, you can double it to find how many Linux users there are, which is a number in the millions.
I don’t feel like it’s reasonable to say Chinese players only play dota 2 or pubg. My games have been bought in China without localization. The real way to look at it is that Linux users are 2.4 million users and users in China are 30 million users. Both you can likely equally tap into but localization typically costs a bit more than Linux porting. But you really should do both for maximum reach.
Overall just because there are 2.4 million users who can buy your game doesn’t mean there will be even 1% of those actually buying your game. Marketing funnel and etc. the bigger number at the start of the funnel the better but you also have to keep that conversation percentage up.
So overall as a game developer it’s more useful for me to look at the unedited stats for hardware choices. Editing them to make 2.4 million seem like a larger percentage doesn’t help and isn’t meaningful when you break the percentages to raw numbers.
I’m not saying that Chinese players only play Dota 2 and PUBG. I’m saying that measurably, the biggest swings in the survey come from Chinese players who only play a couple of the biggest games and nothing else, and they play on homogenized hardware at gaming centers running the same graphics card and operating system. In these same ebbs and flows, people write articles saying “Linux usage surges” and “Windows 11 users leave for Windows 10 en masse”, but neither is true. All that happened is that those Chinese players came back for one specific game this month compared to last month, and you can see that by the increase in Simplified Chinese users.
There are plenty of people, regardless of location, that use Simplified Chinese and wouldn’t be outliers, but you’re better off collecting that number during a down month, and you can get a better representation of actual Linux usage over time by selecting one language that isn’t Simplified Chinese, like English.
You can make those generalizations about Americans too. Most play CS2 and Dota 2 on steam. Also since COVID the home computer market in China has been on a rise.
That said, China has a 20% decrease, I think that’s the most down month you’ll see in a while.
That said percentage doesn’t matter since we extracted the numbers from the percentages. 2.4 million. If 2.4 million is 4% or 2% doesn’t matter.
Yes, it’s true of all demographics that most people just play those few big games. None are so massive in absolute numbers on homogenized PC setups like China that they visibly swing percentages on their own.
The percentages only matter as far as observing trends, which is why this article and its Windows equivalent need to be presented in the context of how much China moves the needle in either direction, since Valve only releases numbers on total monthly active users at irregular intervals. The last time we got a number on that was March 2022, as far as I know. Home PC usage in China may be on the rise, but 12% of Windows 10 users didn’t switch to Windows 11 and Linux in the past month; Simplified Chinese dropped by 19%. That’s not a trend in user behavior, the thing that interests us about the percentages. It’s just a large part of the survey not participating in it this month.
If we had absolute numbers for monthly active users to go along with the percentages, you’re right; the percentages would matter a whole lot less. But since we don’t, we can observe trends, and those trends make a lot more sense when you get rid of outliers.
Any surge also has more to do what China did or did not do that month than how many Steam Decks they sold.
Yeah, that’s a good point. The steam survey for OSes has always been relatively poor quality and untrackable. Ideally, they’d either put the resources into making it transparent and clear or stop it to not spread false data.
GamingOnLinux does a good job of tracking the data separated to just the English language, which shows how much effect China has on the survey as well as a decent approximation of the rest of the audience’s Linux usage, which is closer to 4% these days, post-Steam-Deck.
Steam Decks would still be at least 42% of Linux users though, it just depends on scale. I don’t understand why the Linux community wants to cut out Chinese users just because they are from China.
They’re good outliers to cut out from a sample set. If you want to use the survey as a developer to determine what your audience looks like, how much merit is there in factoring in a player base that mostly only plays Dota 2 and PUBG for major patches or something? And don’t forget that the other way to look at that sample set is that for every Steam Deck customer out in the wild, you can double it to find how many Linux users there are, which is a number in the millions.
I don’t feel like it’s reasonable to say Chinese players only play dota 2 or pubg. My games have been bought in China without localization. The real way to look at it is that Linux users are 2.4 million users and users in China are 30 million users. Both you can likely equally tap into but localization typically costs a bit more than Linux porting. But you really should do both for maximum reach.
Overall just because there are 2.4 million users who can buy your game doesn’t mean there will be even 1% of those actually buying your game. Marketing funnel and etc. the bigger number at the start of the funnel the better but you also have to keep that conversation percentage up.
So overall as a game developer it’s more useful for me to look at the unedited stats for hardware choices. Editing them to make 2.4 million seem like a larger percentage doesn’t help and isn’t meaningful when you break the percentages to raw numbers.
I’m not saying that Chinese players only play Dota 2 and PUBG. I’m saying that measurably, the biggest swings in the survey come from Chinese players who only play a couple of the biggest games and nothing else, and they play on homogenized hardware at gaming centers running the same graphics card and operating system. In these same ebbs and flows, people write articles saying “Linux usage surges” and “Windows 11 users leave for Windows 10 en masse”, but neither is true. All that happened is that those Chinese players came back for one specific game this month compared to last month, and you can see that by the increase in Simplified Chinese users.
There are plenty of people, regardless of location, that use Simplified Chinese and wouldn’t be outliers, but you’re better off collecting that number during a down month, and you can get a better representation of actual Linux usage over time by selecting one language that isn’t Simplified Chinese, like English.
You can make those generalizations about Americans too. Most play CS2 and Dota 2 on steam. Also since COVID the home computer market in China has been on a rise.
That said, China has a 20% decrease, I think that’s the most down month you’ll see in a while.
That said percentage doesn’t matter since we extracted the numbers from the percentages. 2.4 million. If 2.4 million is 4% or 2% doesn’t matter.
Yes, it’s true of all demographics that most people just play those few big games. None are so massive in absolute numbers on homogenized PC setups like China that they visibly swing percentages on their own.
The percentages only matter as far as observing trends, which is why this article and its Windows equivalent need to be presented in the context of how much China moves the needle in either direction, since Valve only releases numbers on total monthly active users at irregular intervals. The last time we got a number on that was March 2022, as far as I know. Home PC usage in China may be on the rise, but 12% of Windows 10 users didn’t switch to Windows 11 and Linux in the past month; Simplified Chinese dropped by 19%. That’s not a trend in user behavior, the thing that interests us about the percentages. It’s just a large part of the survey not participating in it this month.
If we had absolute numbers for monthly active users to go along with the percentages, you’re right; the percentages would matter a whole lot less. But since we don’t, we can observe trends, and those trends make a lot more sense when you get rid of outliers.