[rossdavidh]

Your analysis of the cost-benefits is mostly sound, but I think in the case of programming it breaks down. People's instincts are to think that now that they have become expert, they shouldn't need to be a neophyte again. This works in most fields, but then you see programmers at age 35 realizing that the pool of jobs for their skillset is shrinking, and their labor market leverage is shrinking with it. Yes, there are those last few COBOL gigs, but they only become lucrative after 90%+ of the COBOL programmers have given up and got out.

This is likely always true, but in most fields the skill remains in demand for centuries; wainwright may not pay like it used to (or maybe it does, I don't know) but the decline in demand is slow enough that nobody has to bail out, as long as youngsters aren't continuing to plow into that field it's fine. In programming, the hotness of 10 years ago may already be at its peak, and if you're not willing to move on when you're in your 30's to learn something new, you have sentenced yourself to decades of working in a shrinking job market, which is kind of soul-crushing. Nothing to make your current job intolerable like thinking there's nowhere else you can go.

So, while Lang Lang may not choose to pick up the violin, it is probably fine, because the pianoforte is going to be a good skillset for the rest of his life. The harpsichord players who refused to move on to piano or organ may not have had a crisis in mid-career, because the transition from harpsichord to piano was slow enough to happen over generations. Programming tech changes faster than that (for better or worse).