As far as actually asking questions, I've been on SO for a few years so I've used it for a lot of different things over time. I used it when I was first getting acclimated with C and Java outside of class projects, and now I use it whenever I'm picking up a new pet language. I've been asking and searching a lot about Haskell lately.
Also I find most questions are simple implementation issues or bugs, rather than around complex design problems. Perhaps I am missing a trick...
Likewise. I started noticing that SO results to software-related queries were of sufficiently better quality that I started to qualify some searches with site:stackoverflow.com. (Similarly, I also add site:ycombinator.com with some other types of searches.)
It's too bad that SO/SX has fragmented their URL structure to make the site: qualifier less effective. E.g., it would have been better to have developer.stackexchange.com, superuser.stackexchange.com, etc. Then you could choose to do just site:stackexchange.com to search the whole corpus vs site:developer.stackexchange.com to be more narrow.
I'm a software developer by day, and I work with a team of around 20 or so other developers. But we don't ship products or create for the web. We're a robotics research laboratory, rooted pretty squarely in academia and non-profit. The closest we ever get to a "product" is a public YouTube video demo or a publication and a conference talk.
I do a little bit of web/app stuff on the side, for fun, but I took the survey from the perspective of my professional programming life, which is my day job.
I know that people like me are in the minority, and that web dev/app dev is where the growth is right now, but it seems a little silly to skew a survey like this away from "people who write software in the general case" because then you aren't really creating a survey of all StackOverflow users. I actually found it a little challenging to answer some of the questions accurately. Seeing how they tied their results so strongly in to their Careers product makes the layout of the survey self-evident in hindsight, but I was a little unsatisfied with the survey itself regardless.