... and even more often, historical artifacts which built up for no good reasons, but which aren't worth fixing due to everyone knowing it this way.

Music notation is a mess.

I'm not proposing changing it (that would make no sense -- it's good enough), but it's not worth treating as somehow right. Centuries of unplanned historical evolution don't necessarily result in something rational.

To go with your analogy: it's like the legacy software system written in Fortran which runs on VMS. There's a ton of institutional knowledge and good ideas, but a ton of cruft too.

The nice part is that, in fact, you can ignore most of it unless you are explicitly aiming to reproduce and capture elements of older works. If you are composing new work, you never really have to learn theory, but a few of the most popular frameworks help things along(e.g. composing to familiar meter, motif and chord structure does a ton to fill in gaps) and as a professional getting some familiarity with repertoire, if not necessarily its underlying theory, is expected. But once you have some distinct patterns to play around with, the music is already nearly there, and it's just a case of elaboration and refinement in some direction.

In that sense it's much more forgiving than software. Software becomes indecipherable almost as soon as nobody is looking at it and its structures are in some sense necessary for a purpose. It's not as amenable to pragmatic solutions although often, more of them are possible than we recognize.

Well, I think there are more basic notational issues too, which you can't ignore; they affect everyone. For example, the way that sharps and flats work makes no rational sense, except in the context of history of instruments and scales.

A lot of the notation of timing is cumbersome. Why the large number of wonky-looking symbols? Compare to how a lot of composition software shows timing with how long a position notes take up horizontally (music notation predates the concept of Cartesian coordinates!). It's a lot easier to work with, in all respects.

All this stuff builds up to about an unnecessary 3-month learning curve for kids learning music notation.

Is it worth changing a whole industry -- where everyone already knows the notation and there are countless works in notation -- to save all kids three months? Maybe, but probably not. That's the legacy system problem.

The nice thing -- compared to software systems -- is that none of this stuff is really all that complex.