Managing blog overload

This entry from Myke’s Weblog: Too Many Great Blogs: Information Overload (vectored by David St. Lawrence of Ripples) mentions an issue that has also crossed my mind: for a blog that posts content on many different topics, how can we allow a reader to subscribe to only the topics of interest? This is especially relevant to bloggers such as myself, who tend to intermingle family content with the occasional general-interest post.

As I know them, the current solutions leave something to be desired. In Movable Type (the blogware I use), one could set up each category as a separate blog, with all of them controlled from the same console. In addition to requiring the commercial license for MT, this solution presents control issues. Look-and-feel changes would be tedious to propagate across several blogs, for instance. Other commercial blog services are probably even less suitable, though I don’t have experience with them.
UPDATE: MT actually makes it fairly easy to add per-category feeds. See or for examples
So now we’re talking about structural changes on the client side, server side or both. We could alter the blogware so that each category has its own feed, and readers could subscribe to each category’s feed individually. An aggregate feed of all articles would also be available. This would work pretty well, wouldn’t require client changes, but requires every blogger to update to a new software version in order to work.
Additionally or alternately, the RSS aggregator client could filter a feed. This strikes me as a more flexible approach, since it wouldn’t require every blogger everywhere to change software. Ideally the syndication formats would contain some keyword meta-data to facilitate filtering, but that wouldn’t be strictly necessary. The way I imagine this, the client would set up multiple folders, each with a feed URL and some optional keywords. It would need to be smart enough to know that if multiple folders subscribed to the same URL, it should only poll the URL once for updates and then apply the filters multiple times as needed. The major drawback to this approach is that if a reader is interested in only, say, 20% of the content from a particular feed, he still has to download 100% of the content to filter it down.
Last option: change everything. The blog offers a dynamic feed URL that accepts filter keywords as parameters. This would allow the server to return only the desired content, and it holds some interesting possibilities. Want to get the articles from three years ago, or only by a certain author, or only things posted at night? It could be possible with a system like this. But it imposes the biggest burden in terms of standards development and software change. Client and server would have to agree on filtering language (that means somebody has to develop and publish a standard or pseudo-standard protcol) and both would have to implement it. Not something that can happen overnight.
So, all you RSS aggregator hackers out there…you listening?