Subscriber Beware
Most of the readers that I’ve used, anyway, help me collect feeds, but do little to help me sift through or make sense of what I’ve collected. That this caveat emptor policy hews closely to the RSS metaphor of subscriptions is a bit too convenient for my taste. Unless you’re very disciplined and selective in selecting which RSS feeds to subscribe to, more likely than not your feed corpus resembles something like a postal mailbox on the receiving end of dozens of subscriptions a day. Or, put differently, imagine all of the magazines displayed on your local newsstand arriving at your home every month. Or every day. That, to me, is RSS.
I want a more intelligent approach to the problem. Rather than simply dumping posts on me and leaving me to fend for myself, I’d like my RSS reader to help me negotiate that onslaught.
The World Wide Web Won’t Listen
Much as I long ago abandoned the practice of diligently filing my incoming email into discrete mailboxes and directories (I leave every message in my in-box), I want an RSS reader that doesn’t force me to resort to folders in order to impose order on my subscriptions. A single stream of posts, then, that orders posts chronologically, but also employs some intelligence in showing me posts: the feeds I like the most would always appear in the mix, and the feeds I don’t want to read as often appear less frequently.
Granted, this is not easy behavior to encode into what are essentially unthinking applications. Today’s junk drawer-style feed readers don’t learn anything from how we use them. Given any corpus of RSS feeds, a user is constantly demonstrating to the software which ones she favors and which ones she doesn’t, simply by virtue of clicking on posts in her favorite feeds and neglecting posts in her least favorite. The problem is that the readers aren’t doing anything with that input. They’re not watching.
This is a nontrivial expectation, I know. But it’s not like the ability for software to learn from user behavior lies beyond some unreachable threshold for artificial intelligence. Anyone who’s used the launcher applications Quicksilver or LaunchBar has used the technology I’m talking about. Those programs automatically assign scores to your most frequently accessed items, and register the highest scorers as default selections. And as a hedge against the inevitability of automated misinterpretation, they also allow users to assign scores manually and to override those machine-determined defaults. In essence, they’re not just watching but they’re listening, as well. Now if only someone writing an RSS reader would listen, too.
+