With official support for Windows 10 having officially ended a few days ago, let’s take a look and see how its successor, Windows 11, is doing.
Microsoft released the first Patch Tuesday update (KB5066835) for Windows 11 25H2 this past week and it is probably fair to say that it has been a rough start for the new feature update. Despite the announcement of a wide rollout wherein the new version is now available for download for everyone, the company has already confirmed large-scale issues.
First up, Microsoft was forced to issue an emergency workaround as the update broke localhost auth and following that the company today has confirmed another problem where recovery can become impossible if you happen to use a USB keyboard or mouse.
↫ Sayan Sen at Neowin
Yes. This is a real thing. This latest round of patches makes it entirely impossible to navigate the Windows Recovery Environment with USB keyboards and mice. Since it’s 2025, USB is probably the protocol through which most people connect their keyboard and mice (although to be fair, some laptops probably still default to internal PS/2 for their touchpads). This means that if you run into a problem with Windows 11 that requires you to access the Windows Recovery Environment – perhaps OneDrive did too many lines of cocaine again – you can’t actually do anything inside of it.
There’s no fix yet, so you either remove the offending patches, hope your PC still has a PS/2 port and you still have PS/2 peripherals, or hope Windows 11 won’t fall over and die until Microsoft releases a fix for the issue. Of course, people still using Windows 10, people who aren’t installing every single Windows 11 update as they become available, and people using real operating systems have nothing to worry about.
You can’t help but wonder, though – with Microsoft pushing “AI” so hard, how many of these recent faceplants are the result of Microsoft engineers frantically trying to meet code quotas using Copilot?
This is btw why Windows 10 has grown on me: Ever since Microsoft announced Windows 11, they put Windows 10 in maintenance mode (with last feature upgrade being 22H2) which means they stopped randomly breaking things or randomly moving things around. Since then, Windows 10 has been only security patches.
It goes without saying, I hated the “rolling release” model in Desktop Linux distros and avoided such distros (whose idea was to force users have to take feature upgrades with random breakages if they only want security updates?), and if you told me back in 2011 that Windows would get it too, I would tell you to take your meds without skipping any pills. Just imagine a “rolling release” Windows OS: All the breakages of a “rolling release”, plus conflicts with the proprietary drivers Windows relies on (remember when Windows 10 feature upgrades would uninstall any Nvidia GameReady drivers for the stripped-down Windows Update version?)… Microsoft would be crazy to do that.
Basically:
– Windows 7: A beautiful neoclassical hotel that makes walking through it feel like a dream
Windows 10 (since 22H2): A butt-ugly but functional brutalist hotel
Windows 11: A butt-ugly brutalist hotel where the staff is constantly renovating it while you are inside and randomly moving your things around, oh, and you have to sign in to an app to unlock the door so they can track when you unlock the door and also spy on your location
(I think I will hack ESU support in my Windows 10 until 2028)
Probably most if not all of them. Not that humans don’t make mistakes, but “AI” is about as smart as a housefly at this point, able to react to input but unable to actually reason and therefore understand that input. Asking “AI” to code is like talking to a copy machine, it can only string together and output what it’s absorbed from scouring Stack Exchange and Github. Ask it to generate “new” code, then search those sites for that code, and you will find it, with maybe a variable or string changed here and there, or maybe just verbatim even if it doesn’t actually solve the problem you presented it with.
I have read stories of competent coders using “AI” to help them run tests and debugging against their human-generated code to speed up the process, and they are moderately successful with that. I don’t have a problem with that, in fact I feel that’s a legitimate use for “AI” as it stands today. But all of these companies fucking around with replacing real coders with Speak -n- Spell chatbots are about to find out.