One reason is that I often run the ideas past smart people whose opinions I respect, before running an article. They often find flaws in my own reasoning -- sometimes fatal flaws that shoot down the whole idea so the article never runs -- but sometimes after refining an argument to avoid the flaws they highlight initially, they agree it's an interesting question. If a representative sampling of smart people agree that a question is interesting, but a crowd of Internet commenters keep howling that it's stupid, is it possible that the problem is with the Internet commenters?
One of the most hated articles I ever wrote was questioning the merits of the Fifth Amendment, where I asked: Why is it a good idea that we allow defendants to refuse to answer questions, but that we can force third-party witnesses to answer questions, on pain of going to jail? Before I submitted the article, some people had given me fairly thoughtful answers to this question (e.g. that a third-party witness might only come forward if they know they're allowed to refuse to answer questions about their own illicit activity), but they agreed that the inconsistency was interesting.
But when the article ran on Slashdot, most of the useful discussions in the comments got drowned out by people shouting, "Bennett, you don't get it, the Fifth Amendment protects against SELF-incrimination, that's why you can require third-party witnesses to answer!" Each time, I rolled my eyes, and sighed, and replied, "Yes, I know what it says; that's not the question. The question is: Why is that a good policy?" Nobody in my pre-Slashdot smart-person cadre had ever missed the point so completely, or given that stupid of an answer.