The OP brought up and specifically disqualified Don Knuth because he had freedom as an academic that working software developers don't. The example given was his ability to take off 3 years to work on TeX.

By those criteria I consider Linus Torvalds likely to be disqualified as well since he made his reputation while using the freedom he had as a student.

Ward Cunningham would be a good addition to the list.

I wonder how Don Knuth would fare writing something like TeX under time constraints. I suspect that most programmers would prefer to have unlimited time to make things "perfect" rather than be under pressure; just because Knuth had the liberty to take that option doesn't mean he wouldn't do as well as (if not better than) anyone else on a restricted schedule.

In my experience, getting anything done when you have unlimited time is a challenge upon itself.

I second that. I also think that a person who quitted using email in 1990 for not loosing time for his studies, wouldn't have unlimited time for writing Tex.