Apparently, the Operating System Is on Wednesday
There is now a Wednesday morning slot on my calendar for working toward designing an operating system. AI suggested the experiment. I said, why the fuck not? I'm already building a file manager and a window manager. Apparently, Wednesday needed something to do.
Then I remembered Herb.
Fifteen or twenty years ago, my mentor asked me why I wasn't writing my own operating system. I looked at him and asked whether he was out of his goddamn mind. I wasn't that smart. Writing an operating system was an insane fucking thing to do. Absolutely not.
Here we are in 2026, and I'm considering how I might actually attempt it.
I could use the Linux kernel. I've already been building things that deal with devices and the desktop. AI changes how much I can try, and how quickly I can get from an idea to something I can actually examine. The possibility has become interesting enough to fuck around with.
The thing I would want is freedom to use whatever device, shell, and software I want. Run software from the macOS, Linux, and Windows worlds. Pick the shell I like. Have the operating system accommodate the combination.
I don't know how much of that is possible. That's part of the experiment. I haven't built it, I haven't solved compatibility, and mentioning the Linux kernel doesn't magically make every application run everywhere. What I have is something I would enjoy investigating.
Naturally, the conversation acquired a little disclaimer about what could be promised. Who the fuck asked for a promise? This is an experiment. I can try a thing without guaranteeing the final result before I've started.
If I start it and put it on GitHub, other people might find it. Somebody might have the same complaint and an idea I haven't considered. Somebody might contribute a piece. With the tools available now, there are things we can actually attempt together that would have been a very different proposition when Herb asked me that question.
I'm not claiming I'm going to solve every operating-system complaint in the history of computing. I'm interested in building something and seeing how far it goes. It would be hella fun.
I will probably fucking attempt it, because I'm a maniac and the idea amuses me. For now, the concrete commitment is a recurring architecture or prototype experiment, grounded in the file manager and window manager work. One bounded thing to investigate at a time.
Herb asked me that question fifteen or twenty years ago, and I told him absolutely fucking not.
Apparently, the answer is Wednesday.