Dave Winer seems de facto to want to tie the concept of a blog to a specific format, or form factor. The truth of it, though, stares him in the face: It's not the software, it's the modality. Lotus didn't triumph by inventing the spreadhseet; it triumphed by modeling it.
The early electronic spreadsheets were imperfect analogs of the paper ones in much the same way that Radio or Blogger weblogs are imperfect analogs of paper notebooks, and BBS systems were imperfect analogs of common cork boards.
To communicate the real relationship, Dave should look at is own fascination with the outliner, and especially with OPML and "instant outlining." Radio's "instant outlining" feature comes closer to being the "platform" that will revolutionize blogging than Blogger/Radio/Manila/MT standard layout templates do. They are merely standard layouts; they don't let people reshape their work dynamically. Sure, if you're good with DHTML you can do all kinds of fun stuff; but where's the capability (just for example) to reverse the chronology so that you don't have to read backward through every weblog?
The real breakthrough in spreadsheets wasn't calculation -- it was presentation. Making the graphs; sorting the data. Back in the day, people would commit enormous amounts of time and effort to proprietary branches like Enable and Trapeze just to get presentation features that those programs could offer. It was Excel's outstanding (by the standard of the time) graphing features that gave it the crucial edge it needed to break in. I'd argue that, given WordPerfect's utter dominance in word processing at the time and Word 1.x's generally poor showing against it, that it was Excel that was responsible for Windows's early inroads on the corporate side.
So the real breakthrough in blogging is yet to come. FWIW, I think that breakthrough will come as some kind of a combination of centralized, network-based storage, and a more full realization of Dave Winer et al.'s idea of linking Instant Outlining to Jabber/Gabber. But they've got to ditch OPML for a more general-purpose and less-fragile format, first.