2.3 percentage points of growth is not all that small when talking about steam's user base I think. and his point about acceleration is interesting, we might still be in the early phases of an s-curve.

personally though, as a 20-year Linux user I'm already gobsmacked by the number of new users and mainstream discussion around linux. if we never crack 5% that will already be more users than I ever thought Linux would get

If Linux adoption was something of a single season, some sort of growth Linux community had in the "early 2020" your argument would be valid: you had a steady growth on Linux's own name:

if in the 2020 Linux were 2 and..
in 2025 were were 20 = you had a 900% growth

but this is not what is happening, Linux isn't growing on its own number, but on the number of the global PC gaming growth. New desktop/gaming PC are sold by default with Windows: it mean people don't "choose" Windows, they simply come with the stuff they bought. Windows 11 "growth" is mostly like that: it's not about a growth of users that willingly are choosing Windows. The very slow pace of decline of Windows 10 tell also that people is unwilling to buy into Microsoft experience... even if they are basically forced to: they also cannot chose Windows 10.

On the other side, every newcomers Linux userbase is an active and willing-fully choice: the fact that "new Windows 11" (aka: default new PC) is not restricting the Linux userbase which, on the contrary, is keeping up with the pace (no, it's not "thanks" to steam deck also: the SD's gpu stopped it's growth as you can see in the Steam HW survey). These are the key elements:

-- PC gaming is growing,
-- PC prebuilt market is slowing down (thanks to the ugly Windows 11)
-- Windows 10 decline very slow (looks like used market and DIY rigs still attract the old "not ugly/AI" Windows 11)
-- Linux is keeping the pace even tho the "pushing" of SteamDeck came to end.

https://xkcd.com/1102/

I hear what you're saying and you got a point because doing only the kind of analysis I did can sometimes be misleading; however, I think looking at relative growth instead of absolute is more informative in this case because it better illustrates a growth trend. The number of Linux PCs running Steam more than tripled in 5 years. I think that that is worth highlighting and I wanted to point it out to balance what the article had to say about Linux growth, which I really thought was minimizing how remarkable that growth is.