The sentence in the article (written in January 2005) that's supposed to raise your hair: "But 95% of respondents were under age 50."
Okay. So? If that's supposed to be 30 years of solid computing experience, then we're talking about people who started in 1975 (the year before the Apple I came out).
Compare the number of those early adopters to the tremendous increase in the number of people becoming interested in computing in the 1980s -- an increase that just kept growing. Even if we assume a strong bias in talent towards the early adopters, they're going to be quantitatively overwhelmed by the tails of the 1980s, 1990s, and 2000s distributions.
"Compare the number of those early adopters to the tremendous increase in the number of people becoming interested in computing in the 1980s -- an increase that just kept growing."
Not true. There was a bubble in CS enrollment in the mid 80s, and another starting in the late 90s. CS enrollment was down in between, and has been declining recently as well:
If we assume that those 1980s CS grads were 22 when they graduated (let's say in 1985), they'd be 46 years old today -- just below the 95th percentile of the survey results. Effectively, we're saying that the oldest programmers started in the mid-1980s (which was well into the personal computing era). Computer science has been around for a lot longer than that.
In any case, if programming has longevity as a career, we'd expect to see echos of this dual-bubble demographic trend -- a bunch of people who started in the 80s (who are around 50 today), and a bunch who started after the 90s (who are as old as 30-something today). Instead, we see distributions like this one:
Maybe StackOverflow is a biased sample (old programmers don't need to ask questions?), but these results aren't out of line with any other statistics I've seen, or my own experience. By age 30, you're already on the wrong side of the programmer demographic curve. No matter how you look at it, that's very young.
Likewise, the age distribution is going to be slanted towards youth in any nation that is a relative newcomer to the industry in question. It's quite likely that the reason there's few firmware engineers from Pakistan with more than 20 years of experience is simply that there were many fewer jobs in Pakistan engineering firmware 20 years ago.
I agree that programming is a young man's game. But is this true of more traditional branches of engineering, such as Civil, Chemical, Materials Engineering?
In Electrical and Software Engineering, Moore's Law means the technology really does change very quickly, and with it the surface skills required. Oddly, though, it seems that the same concepts are often reinvented, just with incompatible surfaces. For example, OO languages and XML are popular today... and two decades ago, OODBMS and SGML were popular (though in a narrower domain).
My personal solution is to try to learn universal skills that will always be needed: communication skills, mathematics skills, business skills, investment skills, project management skills, general problem-solving skills etc. Within that set, I try to choose those skills that I like and value, and for which I have some aptitude.
But of course, you can't get anything done without surface skills.
Job experience (whether you've been in the field 5 years or 20) isn't really a great proxy for expertise. Gaining expertise takes time. But time doesn't guarantee expertise. Few ever become expert or master in the Dreyfus model sense.
Okay. So? If that's supposed to be 30 years of solid computing experience, then we're talking about people who started in 1975 (the year before the Apple I came out).
Compare the number of those early adopters to the tremendous increase in the number of people becoming interested in computing in the 1980s -- an increase that just kept growing. Even if we assume a strong bias in talent towards the early adopters, they're going to be quantitatively overwhelmed by the tails of the 1980s, 1990s, and 2000s distributions.