Mark Pasc has some comments about what he thinks a super aggregator would do. For a few days now I've been tossing around the idea of writing something that logs when people update their blogs. Personally, I tend to post late at night between the 11 pm and 2am hours. Other people who have faster connections at work might post more often during the day. In itself this is not very useful, merely a neat statistic about the blogosphere and when it's most active. My theory would be that blogging would be pretty steady throughout all hours as the active time (defined by me as the time that any given timezone would be active) moves around the world.
This gets more interesting though when you look at Pasc's comments. He would like his aggregator to check new feeds based on their likelyhood of being updated. I think this would be best implemented as a service: I would have a bot that collects info from pages and tracks their update pattern. The aggregator at start up hits my service, to find out whose sites should be updated for a block of time (maybe 5 hours into the future), stores that info, and rechecks feeds accordingly. This saves it the trouble of having to do all that work on its own as well as having to store the data. I'm trying to work through Dive into Python by Mark Pilgrim to learn Python. Maybe I'll try doing this with Python... I need a few more questions answered first, though.