| ▲ | jsnell a day ago |
| The statscounter data is not reliable, and it is just embarrassing how often these posts make it to the HN frontpage. You even have a demonstration in this very article, with the surge of classic Mac OS to 7% for several months. The data is obviously nonsense, and when it has errors nobody at the company cares about them. But when they have persistent "data reporting issues", why are we supposed to believe any of these numbers? |
|
| ▲ | zokier a day ago | parent | next [-] |
| Bingo. Cloudflare has also OS stats available and I'd imagine they are far more reliable. Some silver lining of them having such wide dragnet on the web. They report 4.4% Linux desktop marketshare in the US. Tbh I believe the summer vacation season probably influences the numbers here, but there is some real growth too. https://radar.cloudflare.com/explorer?dataSet=http&groupBy=o... |
| |
| ▲ | 827a a day ago | parent [-] | | 1. Radar is also reporting a Linux increase over the past month: 3.3% to 4.4%. 2. Both StatsCounter and Radar break out Linux and ChromeOS; if you combine them, StatsCounter hits 7.7%; Radar hits 6.3%. 3. That being said: Both StatsCounter and Radar experienced an anomalous drop in ChromeOS clients & rise in Linux clients over the past month. StatsCounter took ChromeOS from ~4.4% to 2.7%. Radar took it 2.6% -> 1.9%. This kind of implies that something changed with a major ChromeOS device out that; some model/version maybe changed its UA and started reporting itself as a Linux device instead. | | |
| ▲ | zokier a day ago | parent | next [-] | | ChromeOS drop is pretty easily explained by it being predominantly used in education and schools being closed for summer. And that drop muddies all other numbers, because of course the percentages of others go up when one goes down. In summary, I'd wait until November (or at least October) before making any broad conclusions. | | |
| ▲ | 827a a day ago | parent [-] | | Yeah, schools being out explains the major drop in ChromeOS devices; check out Radar's 12 month view and how ChromeOS data looked August/September 2024. But if it significantly explained the rise in other percentages, we'd see all the other shares go up. However, Windows and MacOS are flat; 64% and 30%, respectively, last and this month. Only Linux went up; so its likely there's some genuine linux desktop adoption going on. But the rise in Linux marketshare is pretty steep; two months ago Radar measured it at 2.6%, now its 4.4%. It could be legit. There's been a significant uptick in tech Youtubers pushing linux content (LTT and Jayz have both done recent videos on it), including the Lenovo Legion Linux vs Windows perf comparisons which found Linux to be faster, Lex Fridman just interviewed DHH and they spoke at length about linux setups (~1M viewers on that likely), and the pushback against Apple in the tech circles is reaching a fever pitch. |
| |
| ▲ | juliusdavies a day ago | parent | prev [-] | | I find this compelling, alongside the fact ChromeBooks are well placed in retail shops and usually the cheapest things you can buy. They are also ubiquitous in elementary schools. This is more about ChromeBooks than linux. Add the fact that all my kids hate their school chromebooks.... maybe this isn't such great news for Linux afterall. | | |
| ▲ | ronsor a day ago | parent [-] | | School Chromebooks are usually locked down (i.e. full of spyware) and poorly maintained. It's no wonder that children dislike them. |
|
|
|
|
| ▲ | danso a day ago | parent | prev | next [-] |
| Mentioned this in another comment [0], but analytics.usa.gov has the % of visitors on Linux operating systems at 5.7% in 2025, up from 4.5% in 2024. Of course "visitors to U.S. government websites" is not fully representative of all U.S. computer users, but it's worth noting. [0] https://news.ycombinator.com/item?id=44582058 |
|
| ▲ | supriyo-biswas a day ago | parent | prev | next [-] |
| Additionally, with the number of people who use ad blockers on Linux and given that statcounter mostly uses 3rd party JS tags, I highly doubt these numbers are correct. There's a discussion in a peer thread about how people never notice its Linux and keep using their refurbished machines as-is. This too, is surprising to me, as my own experience as well as the ones I've heard in person from IT folks and IT-related forums online, people immediately notice that the UI looks different and panic as to how to achieve their current tasks. I'm skeptical of that entire thread too. In general, I just wonder how much of any popular forum is just people LARPing. I do wish that it didn't occur here, though it's undoubtedly difficult to moderate. |
| |
| ▲ | ryukoposting a day ago | parent | next [-] | | > people immediately notice that the UI looks different and immediately panic as to how to achieve their current tasks This was probably a bigger problem 10 years ago than it is now. Plenty of people never do anything at all with their computer besides opening a browser. No matter what OS you use, "click the Chrome logo" still applies. I've watched my grandparents use a computer. I guarantee I could swap out Windows for KDE or Cinnamon and, as long as I make the wallpaper the same and I put the Chrome icon in the same place, they wouldn't notice anything had changed. I'm not actually going to do that, because then I become the only person in the family who can tame their computer if it starts acting out, but still. Also, Microsoft's own UI isn't a steady target. Windows 11 is, dare I say it, more akin to Plasma 6 than it is to Windows 7. | |
| ▲ | extraduder_ire a day ago | parent | prev [-] | | Unless they're using JS for something specific, getting the user to load anything at all would give them the OS from the useragent string in almost all cases. I'd believe their URLs being included in default filter lists though. |
|
|
| ▲ | oefrha a day ago | parent | prev | next [-] |
| Pretty sure OS X and macOS should be combined, not doing that feels like amateur hour, very puzzling. But even with that in mind, you see wild ups and downs as large as 3.5% a month from 10/24 to 11/24 to 12/24 to 01/25 and there’s no way in hell actual deployments are fluctuating like that. Error bars like that make a number of 5% pretty meaningless, however feel-good it is. Also, for people unfamiliar with the Apple ecosystem: the OS X => macOS rebranding happened back in 2016, IIRC the Safari user agent never ever included macOS (Safari on M4 Macs running latest macOS 15.5 reports itself as “Intel Mac OS X 10_15_7” in its UA), so absolutely no idea where they’re getting this new “macOS” category. Maybe they publish technical details of their methodology somewhere? I can’t bother to check. |
| |
| ▲ | necovek a day ago | parent [-] | | Don't confuse percentage points with percents. 25%+-3.5% means it's 5%+-0.7% for proportional error bars. They don't have to be linear, true, but they are certainly not 5% +- 3.5% either. |
|
|
| ▲ | elsjaako a day ago | parent | prev | next [-] |
| Note that OS X goes down for the same period. I believe Apple is calling it MacOS now. So that looks like it might be some change in how Apple computers are reporting their OS. |
| |
| ▲ | jsnell a day ago | parent | next [-] | | Indeed, OS X goes down, and obviously none of us actually believe that. But not only does Statcounter report that clearly faulty number, but they have yet to fix the problem. This happens all the time. When their numbers are clearly wrong, they don't care about the numbers enough to fix even the glaring problems, their sample is unsound, and their methodology is unpublished, why exactly are we supposed to give any of their numbers any credence? What you've written is the first I've heard of a recent change to the Safari on OS X user-agent string, and I see no indication of it in my access logs. What's it supposed to be now? It seems a bit unlikely, and given Safari never ran on classic Mac OS, it seems like a company that's supposed to specialize in analytics should be able to handle it... | |
| ▲ | IshKebab a day ago | parent | prev | next [-] | | I don't understand why Statcounter reports them separately though. They're just two different versions of the same OS, and those are grouped for other OSes in this chart. Makes no sense. | |
| ▲ | zozbot234 a day ago | parent | prev [-] | | "mac OS" not "MacOS". MacOS is for the older pre-OS X versions. | | |
|
|
| ▲ | arp242 a day ago | parent | prev | next [-] |
| > the surge of classic Mac OS to 7% for several months I'm not sure what's up with listing both "OS X" and "macOS", but I'm quite confident it's not classic Mac OS. |
| |
| ▲ | bichiliad a day ago | parent | next [-] | | This could be related to the re-versioning of macOS from 10.x to the year of the release. | |
| ▲ | speedgoose a day ago | parent | prev [-] | | Can you even have a successful TLS handshake with Mac OS 9 ? | | |
|
|
| ▲ | weberer a day ago | parent | prev | next [-] |
| You can also look at the Steam survey as another data point. Linux use among English speakers is just above 5%, but the data is biased toward power users/ gamers. https://www.gamingonlinux.com/steam-tracker/ |
|
| ▲ | phendrenad2 a day ago | parent | prev | next [-] |
| I wonder what a more reliable measure would be. Maybe something like the "Crane Index" where we count the number of new software packages for Linux. Particularly, it makes sense to focus on paid software, because there's actually some bar to entry there (setting up an LLC, accepting payments, etc.) I haven't actually looked into this, but I think the initial data for this figure is zero, and we're projected to reach zero by next year. |
|
| ▲ | necovek a day ago | parent | prev | next [-] |
| The article claims this is due to Apple rebranding OS X back to MacOS with newer releases. Are you disputing that? Or did you miss that in the article? |
|
| ▲ | jama211 a day ago | parent | prev [-] |
| That’s not classic macOS… that’s modern macOS, as in post OS X dude |
