eah

The spike likely represents Statcounter being too slow to adjust their methods of filtering out bot page views after some brand new web scrapers appeared on the scene hoovering up all the data for AI training. 10% linux market share would be a cosmic event. You would hear about it in mainstream media if it were really happening.

Edit: I forgot about popular youtube channel Linus Tech Tips' recent videos in which they gave linux their highest praise yet. One was published in mid June, another in mid July, and their most recent and most favorable to linux was published on July 22 ("My Experience Using the Steam Machine as a Desktop PC"). There have also been videos on the Framework laptops. Vibes tell me that there's a trend of more people using mobile over desktop, so I would expect a higher proportion of desktop users to be the gamer or technically inclined type, enough to be influenced by LTT and give linux a shot, compared to 10 years ago.

Here's the data for the last 4 months.

    
   Date    Windows     OS X    Linux    Chrome OS    macOS
2026-04      66.39    17.04     2.94         4.42     9.21
2026-05      64.66    19.92     3.56         3.72     8.15
2026-06      59.08    24.06     6.14         2.19     8.53
2026-07      57.63    21.07    10.61         2.07     8.63

The data looks noisy. You would expect the lines to have consistent, regular annual patterns. You can see that with Chrome OS which seems to have a dip in the middle of the year corresponding to when schools are out for summer vacation, and if you scan upward, it appears Windows has a corresponding hump at that time. But there are still unexplained extreme swings. The lines look smooth up until 2012/2013. Maybe Statcounter doesn't have as many members/page views as it once did, leading to higher variance. We could see that if we had absolute numbers, but we only have percentages. That's probably affected by the advent of the LLMs. Nobody's going to the web anymore to find information.

Using reddit's stock price to estimate what fraction of the users are people. Sounds like a plan. The stock price is like a CAPTCHA for social media platforms. Are you authentically human or an empty husk of clankers gabbing at each other?

They're all forms of delay-line memory.

I've heard the reason for that is archive.today uses accounts to get around paywalls and the usernames for those accounts which appear on the page are hidden in order to not reveal to the scraped service which accounts are being used.

Nobody deserves to be bullied.

It's gothub.org. You linked to some dumb AI startup.

It's a sure sign of a bubble that a URL typo coincidentally brings one to a completely unrelated page for a completely unknown, in-beta AI startup with a dumb idea.

I reject your reality and substitute my own.

The correct answer is C: the cube plops out and the object to which the blue portal is affixed flies off in the southwest direction.

That was just an advertisement for the AI companies. I'm surprised the media aired it. Well,... not surprised. It's the media.

Chiming in to say I did the same and had the same result. This is the facebook post. I think we're all learning in real time that unless a photo comes from reputable source like a historical society or something like that, then you can't trust that it has any provenance rooted in the real world. It takes too little effort to generate simulacrum and blast it on the internets.