Let's get this clear from the start. I want a Tesla and have seriously considering buying one. I've also stated publicly and repeatedly in this column that Tesla will completely disrupt the automobile industry, and that the established car makers are incapable, culturally, of making a comparable electric car. And I generally write very positively about Elon Musk, although I don't always agree 100 percent with him.

Despite all that, however, I have decided NOT to buy a Tesla, not because I don't think it's a great product but because I believe that what makes it great today will turn it into a headache tomorrow. Allow me to explain.

Traditional car manufacturers have been adding software to their vehicles for a couple of decades, and, without exception, that software--and the interfaces they created for it--is crappy and awkward. Why? Because car companies aren't software companies. Their execs don't understand software. Their cultures don't appreciate programmers. They're basically clueless.

Tesla, on the other hand, could be characterized as a software company that's manufacturing cars to run the software. This is a hugely different proposition than traditional automobile making. Because Tesla draws on decades of Silicon Valley experience in creating and updating software, a Tesla can be field upgraded and field debugged. As venture capitalist Lou Steinberg explains in Inverse magazine:

Aside from navigation maps, all of my cars had features that were largely fixed on the day they left the factory. Not my Tesla. Every month, it gets software updates that make it better. It learned how to park. Then it learned how to do it better. It opens my garage door when I come home. It improved its self-driving. It improved the stereo. It added anti-theft features. After one year, my car is safer and better to drive than the day I bought it. My Tesla driving experience keeps improving through patches and updates.

Pretty cool, eh? No argument there. It's revolutionary.

Only one problem: Tesla software is, well, software. And one of the distinctive characteristics of software is that, as it accumulates more capability and more internal complexity, it becomes increasingly brittle, less stable, and prone to unexpectedly stop running. In software lingo, such catastrophic events are called "crashes." In the case of an automobile, the term might not be a metaphor.

Consider, just for a second, your experience with PCs. As operating systems have gotten more complex, they're showing absolutely no sign of becoming more stable or secure. Windows 10, for example, still ends up occassionally trapping to the infamous "blue screen of death."

Again, in an automobile, that's maybe not a metaphor.

To be fair, the software that the traditional car companies develop to compete with Tesla will probably reach that point (a.k.a. the "tar pit") much sooner and probably with much more dramatic results.

Even so, Silicon Valley software culture, while vastly more competent than Detroit software culture, continues to cling to crazy-ass ideas like the infamous 80-plus-hour workweek. Look, all programmers create code with bugs in it, but tired, burned-out programmers create really buggy code. Anyone who has worked in a software development organization knows this, but this simple fact doesn't seem to have penetrated into ranks of high-tech management.

Which leads me to my decision to NOT buy a Tesla. Sure, I'd enjoy it for a few years, but I know from experience that the car that's cool and self-improving today will eventually become a headache and perhaps even a nightmare. And that's doubly true for non-Tesla cars that crutch on software.

So, for the time being, I'm sticking with my 2000 Honda CRV, which has 265,000 miles on it, seldom breaks down, and costs almost nothing to service. And when it finally dies, I'll probably buy another just like it for a few thousand dollars. Or maybe two and use one for spare parts.

Of course, I'll have to actually drive the car, parallel park it, and walk across the parking lot. But, you know, I never really saw any of that as much of a bother. In fact, when I think about it, the real reason I wanted to buy a Tesla is because I'm a gadget-head. This is just one case, though, where I'm going to ignore the craving for a shiny new toy.