Blogging via the browser is a potentially powerful model. This post, for example, was entirely written and published using Flock, the content-oriented browser based on the Mozilla codebase.
Of course, the real value isn't in using the browser as a stand-alone blog-post editor. If you're online you may as well use your native blog editor. For instance, WordPress ships with an particularly capable editor, and it is hard to beat the blog software itself in terms of integration. (The client-side spell check sure is nice, though.)
No, the real value of integrating blog-publishing into the browser is to help foster "clip" blogging. I.e. making it easy to grab snippets from around the web and incorporate them into yourown content stream.
Many standalone services, from Google Notebook to eSnips to Net Snippets, etc., already offer clip blogging, but I suspect that mass adoption of the as-you-surf content collection model won't take off until it is built in to the browser or the desktop itself.
And while I instinctively resist clients that are too rich and try to do too much, Flock is pretty damn good and it may just change my opinion about such things.
Speaking of editors, a perhaps more powerful blogging model would be to have exact replica of your site on your desktop/laptop/etc, and have it sync automatically whenever you are connected to the net. (This is the intended mode of operation for Houston; at least for the data backend.)
On a related note, when are people going to start supporting Atom Pub in addition to, or instead of, XML-RPC?