James Hague is right:
I read 20 Controversial Programming Opinions, and I found myself nodding "yes, yes get to the interesting stuff." And then, after "less code is better than more," it was over. It was like reading a list of controversial health tips that included "eat your veggies" and "don't be sedentary." In an effort to restore a bit of spark to the once revolutionary software development world, I present some opinions that are hopefully more legitimately controversial.
But James' alternative opinions are (except for the ones about social institutions) hardly more controversial. At best they're offensive to a vocal minority, e.g:
Purely functional programming doesn't work, but if you mix in a small amount of imperative code then it does.
This is taken for granted in the Lisp and ML communities, and even among Haskellers it's not an unusual opinion. It's heresy to the functional puritans, so it's controversial in the sense that it generates argument, but not in the sense that informed people think it's very unlikely to be true.
When I try to generate controversial opinions, I get similar results (and I'm not giving any examples, because they were embarrassingly insipid). Instead of bold heresies, I say things I expect other people to disagree with quantitatively, but not qualitatively (although I'm probably underestimating the degree of disagreement). Apparently it's hard to distinguish one's model of the truth from one's model of other people's opinions.