Google recently launched the beta of Google Reader which is a web-based feed aggregator with the GMail-look you'd expect from Google. Like GMail, feeds are tagged rather than in folders.
If I were still using Bloglines, I might consider switching but I'm pretty attached to NetNewsWire so I'm not sure. The cost of switching is high for me because I flag a lot and I'd like my 'read' list to be consistent between clients. With email, that's achievable because of IMAP.
I took a look at my server logs to see if Google is grabbing feeds with something different from their GoogleBot. Turns out they are. Since 7th Oct I've got regular accesses to my atom feed from a user agent:
FeedFetcher-Google; (+http://www.google.com/feedfetcher.html)
The page linked to in the user agent string makes it clear that FeedFetcher is only for user-initiated feed retrieval (i.e. subscriptions in Google Reader or the personalised Google home page) and that the blog search and regular Google search are crawled for separately.
UPDATE: I started thinking more about IMAP for feed reading. Atom is the obvious contender but something more is necessary because the server needs to indicate what's read/unread and the client needs to be able to mark entries read/unread. A simple extension element would work for the former. What about the latter?