[microtonal]

I don't believe this for a moment. I have seen many more Chromebook users than desktop Linux users. Not individuals, but a lot of schools use Chromebooks where I am. They are probably not well-tallied in the results, because they are often used to do some work in Google Docs/Sheets, plus some education-specific sites and less browsing on the general web.

I guess my point is that it is nearly impossible to get reliable counts from user agent statistics, etc. (lots of crawlers, that identify as a regular desktop as well etc.). The best way would probably to do real-world surveys among random population samples. But that costs money/time.