Thank you @Kiriko for relaying this. The following will be directed at the original author. As the official KDE fedi account sharply responded to David earlier, I’m assuming there’s a chance they might see it. Charitably assuming this is not chatbot slop (likely a bad assumption),
where the desktop itself can be compiled per user, per device, per moment, from a small portable model of the user — what we are sketching as Kadai: an encrypted, vendor-agnostic personal kernel from which Plasma assembles each Activity.
This is flatly nonsensical. How would this be in any way better than secure synchronization of existing configuration files? I.E. an integration with existing tools like, say, rsync or SyncThing? An interface that changes itself while the user is using it is a recipe for end-user frustration and, ultimately, rejection. I use KDE across a couple of devices, both conventional desktop and touchscreen, and its existing tools for panel configuration and the like are acceptable and worthy of further effort. Continuous “per-moment” automatic reconfiguration of the user interface – what seems to be the proposal here, if this fanciful jargon is at all parsable – is making the assumption that your black-box, poorly-debuggable inference engine knows what the user wants better than the user does.
Do I even need to mention how ridiculous it would be to replace discrete settings files, parsable by both human and machine, with a black-box database of vector weights that may have variable performance across mobile and desktop hardware?
The glaring risk here is taking the alienation users feel at a point-release interface change, say, from Win10 to Win11, and turning it into a continuous-integration experience. The experience being proffered here is very likely to end up being, “Where did that option go that I only need 1 time in 20? Why was it hidden without my express input? What the hell is this stupid fucking computer doing???” and the user taking their friend’s suggestion to try out Xfce or similar.
This brings to mind Steve Jobs’ insistence on using magneto-optical drives for the early NeXT black hardware; the idea being that the user wouldn’t be tied to one specific machine, their files and user identity would live on the MO disk that could be transferred from machine to machine. The problem was that the MO disks were horribly slow and expensive, leading to the impression that the NeXT machines weren’t worth the exorbitant price tag. Frustration was baked into every user encounter with the interface. The proposal here risks being slow on lesser hardware AND continually incoherent, a recipe for an abandoned project that isn’t even as well-remembered as NeXT, with no potential to be adapted into something like OS X or iOS.
Lastly,
Sovereignty, from this angle, is just lovability and AI-ownership scaled up to the organisation:
This sentence seems even more nonsensical. I understand and can agree with current European anxieties surrounding US dependencies in their tech stacks, and I can definitely understand KDE as a European-led project that ameliorates that. But as our community here understands, the word “sovereignty” has also been thoroughly colonized by cryptocurrency bros, most of whom became resolute enthusiasts of the current generation of AI. If “sovereignty” is going to continue to be your watchword for this crap, you may ultimately be operating at cross-purposes to your stated intent.
Really bumping up against this taking a math class. It’s not just the ick feeling of getting slapped with a Google AI rendition that may or may not be accurate. There’s at least 4 or 5 other vendors, Chegg and the like, doing the same thing but keeping their explanations behind a login wall. That’s the entire first page of Google search results now. Of course, simply copying and pasting questions from a textbook is a terrible search strategy. But I still wish I could find a durable math Q&A community that wasn’t slop-adjacent.