I’ve just posted something very similar to this to the NetNewsWire users’ mailing list, but I figure a wider audience might be appropriate. Unfortunately, the only outlet I have is this blog, so you lot will have to do.
Round the table pictured in the previous post, some of us were musing on the problem of keeping up with one’s aggregator, should one happen to drop offline for a few days or (with enough subscriptions) minutes. ‘Mark all as read’ is something of a blunt instrument. I found myself wishing for something analogous to a spam filter, only less judgemental; tools to handle situations like:
- “These posts in Gizmodo and Engadget link to the same destinations and contain similar words. Chances are they’re about the same thing, so I’ll just hide one for you.”
- “You have 5000 unread posts. But 4000 of them are similar to the sorts of posts you usually skip past quickly, and frankly never open in your browser. So I’ll mark all those as read, because there’s a good chance you’re not missing much.”
I realise that this sort of stuff will be (a.) hard, (b.) a fundamental change in the way we use aggregators and (c.) the cause of an awful lot of head-clutching about UI issues. But I do wonder if the next stage of aggregator development shouldn’t involve some sort of heuristic filtering and assistance. That is, if an aggregator shouldn’t take ‘subscribe’ as meaning ‘this looks interesting, add it to the pool from which you draw posts of the type I read.’
Thoughts, anyone? Am I completely mad? Or did I simply read too many of those articles about ‘software agents’ that came out of the Media Lab around 1993?
[update: Quote NetNewsWire author Brent Simmons:
Filtering features of various types are on the to-do list. It’s too soon to provide more details, but I will, as features become more concrete.
Yay! Also see the comments for an interesting link from Allan, and the comments to that link for… basically, lots of people have been having the same idea over the last few months. I’m just late to the party, again.]