Jolly Roger

coding

Why I Chose a Mac After Struggling to Use Linux as My Home Desktop

Linux has always looked like the operating system I ought to be using. I work with websites, code, servers and publishing tools, and I like being able to look under the hood. An open system that I can control, without having to accept every decision made by Microsoft or Apple, sounds right for me. I tried to make Linux my home desktop. In the end, though, I chose a Mac. The reason was disappointingly ordinary: I needed my computer to run the things I use, with the hardware I have, without turning routine work into an ongoing repair job.

Drivers were part of the problem. Linux runs beautifully on plenty of hardware, but "plenty" is not a reassuring answer when the device you need is one of the exceptions. Driver support can depend on the exact hardware revision, kernel version, firmware and willingness of a manufacturer to help. Finding the right combination, checking forums and trying fixes is interesting when you're experimenting. It is far less interesting when you have something to finish. I don't want to choose a printer, peripheral or computer component by first researching whether someone has persuaded it to work with my distribution.

Then there is setup. Linux enthusiasts are right that today's distributions can be surprisingly polished. Some installations are easier than Windows. But once something goes wrong, the experience can change fast. One package wants another version of a dependency, a driver behaves differently after an update, or instructions written for one distribution do not quite apply to another. I can use a terminal. That's not the point. Just because I can troubleshoot a home computer doesn't mean I want to spend my evenings doing it.

The bigger obstacle was software. I need access to Adobe applications and Microsoft Office workflows. For photography and design work, the Adobe ecosystem is hard to dismiss with a casual suggestion to use an open-source alternative. GIMP is useful, and there are other strong Linux tools, but a replacement program is not automatically a replacement for the files, habits and features one already relies on. The same goes for Office. LibreOffice can do a lot, and Microsoft has browser-based applications, yet neither completely removes the compatibility concerns of exchanging documents, spreadsheets and presentations with people who use the desktop Microsoft applications. I wanted to work on my files, not negotiate with them.

A Mac offered the compromise I could actually live with. It gives me a Unix-like environment for development and command-line work while running the mainstream Adobe and Microsoft applications I care about. Driver support and system integration are generally handled as part of a much more tightly controlled hardware and software package. There is a price for that convenience. Apple hardware is expensive, the platform is proprietary, and I give up some freedom that Linux would offer. I know all that. But the desktop is where I need the least friction, not where I want to prove my independence from the big vendors.

That personal history is why I found the AppCoding discussion of David Heinemeier Hansson and AI-powered Linux development so interesting. The idea is that AI agents let small teams maintain and improve software ecosystems that once demanded a huge number of developers. Hansson's Omarchy project puts that argument into a Linux desktop. It made me think about my own failed attempt to make Linux the everyday answer. My problems weren't philosophical. They were practical, and practical problems are exactly the sort that better automation might reduce.

I explored the market consequences in a separate MarketAnalysis.com article on how AI coding could shift desktop market share from Windows to Linux. The three obstacles are familiar to anyone who has tried to switch: drivers, setup and missing applications. AI agents could help developers adapt and maintain more drivers, while assisting ordinary users with configuration, diagnostics and upgrades. They could make Linux versions of existing applications cheaper to build and help improve open-source alternatives. Over time, an AI assistant might even carry out a task without depending on the exact Windows or Mac application that traditionally performed it.

There are limits, of course. An agent cannot conjure a working driver out of a proprietary firmware specification it cannot see, and generating a lookalike interface is a long way from reproducing a mature Adobe application. Business documents still need reliable compatibility, not hopeful approximations. Giving an agent permission to modify a machine also introduces security risks. These are engineering problems, and some involve commercial decisions beyond any programmer's control. I would rather see Linux get steadily better at solving them than read another declaration that the year of the Linux desktop has finally arrived.

Would I try Linux again as my main desktop? Yes, if it stopped asking me to make the same compromises. Give me dependable drivers, a painless setup and the applications I need, or genuinely compatible ways to do the same work, and the attraction is still there. For now, the Mac won by being boring in the ways that matter. I could get on with my work. AI may eventually give Linux that same advantage, with more freedom and less dependence on a single vendor. If it does, I would be happy to reconsider.