But I eventually realized that my writing output had this weird gap. On the one hand, I had 6+ page academic papers, all official and published and whatever and listed on my publications page. On the other hand, I had literally tens of thousands of comments scattered across the internet, one to a few paragraphs each, with no real way of pointing to them or even finding a comment again myself. Writing blog posts fills a nice gap. They can be somewhat longer and digress from the immediate conversation that spawned the ideas, and I have something to point to in the future (and find myself).
The other thing I do now is keep a text file of comments I've posted that seem worth finding again (as opposed to ephemeral discussion), with the URL of my comment and some notes on what else I might want to say on that. Has only sort of worked out so far, because like any "list of things I might want to do someday" they mostly don't get expanded into blog posts, but it feels like a step up in organization.
But you still should blog if no one reads it for the reasons some other commenters have alluded to: to develop your own mind. That's essentially why I blog: http://jseliger.com/2010/09/27/signaling-status-blogging-aca... and read other people's blogs.
That, and so I have a repository of already expressed ideas I can leave in the comments sections of other blogs and HN, so I don't have to reinvent the answer every time.