https://gs.statcounter.com/os-market-share/desktop/north-america/#monthly-200901-202607
Cloudflare stats also seem to support peaks of 15% usage but around 9-10% on average.
The trendline suggests this is just the start.
https://gs.statcounter.com/os-market-share/desktop/north-america/#monthly-200901-202607
Cloudflare stats also seem to support peaks of 15% usage but around 9-10% on average.
The trendline suggests this is just the start.
There’s no way that’s accurate, right?
None of those graph ever are. Because Linux users are more prone to tinkering, keeping stuff private and blocking a lot of bullshit scripting, there is big dark area of Linux systems listing themselves as something else (usually Windows because of those asshole websites not playing well with Linux by design without an actual valid reason other than “we don’t like Linux”).
(PS: For the same reason a lot of sites believe I’m using Chrome. I don’t but I know they like to fuck with me when they see that Firefox user agent.)
Quite likely. Many Linux distros aren’t visible because they hide their telemetry, and would be marked as ‘unknown’ - or they spoof Windows or Mac for compatibility reasons. It’d be impossible to get an accurate read on how many computers run Linux, but it’ll be higher than the number that report it.
AFAIK they get their data through crowd-sourced browser header data. And I really doubt any significant number of people are manipulating that.
“Looking at browser headers” usually means they’re parsing the User-Agent string, which all kind of people (and bots) spoof for all kinds of reasons. There even used to (and might still) be a Firefox extension that would generate random ones.
https://chromewebstore.google.com/detail/user-agent-switcher-for-c/djflhoibgkdhkhhcedjiklpkjnoahfmg has over a million users on Chrome alone. It’s possibly more common than you think.
There’s also a question of where they’re measuring this.
You can’t intercept and inspect the browser telemetry of every web request on the entire internet, that’s just not possible. So they must only have access to the data from requests from a specific set of websites.
Are those websites a representative sample of all web traffic, or are some of those websites more or less likely to have Linux users than the average website?
And it seems to be counting per request. Do Linux users visit these websites more often or less often than average? Do they spend more time there and view more pages, or less time and less pages?
A truly unbiased OS survey is basically impossible. At least, hopefully, they’re using the same measurement data over and over again through time, so maybe the trends in the data are still meaningful.
The Cloudflare chart would probably be pretty representative given how much traffic goes through it.