I don't think the Stack Overflow data is fantastic, but I'm not clear that the BLS stats are any better for the fancy credentials of their collectors. In eithe

I don't think the Stack Overflow data is fantastic, but I'm not clear that the BLS stats are any better for the fancy credentials of their collectors. In either case, the problem is coming up with a reasonably accurate reference class. At least in my case, I'm not well represented by the BLS breakdowns (which underestimate what I can expect to get by a factor of about four.) The Stack Overflow approach, using industry-specific features (developer-friendly keywords/job titles/languages) is definitely smarter, though still wildly insufficient. (It's only off by a factor of about three.)