Haystack Blog

I had an interesting discussion with my student Katrina Panovich today.  I’m intrigued by the way people use twitter for “ambient awareness”—watching what goes by, but not worrying about what they miss.   I find this paradoxical—if you don’t care about missing stuff, why watch at all?  Especially given that each arriving tweet provides some degree of distraction from whatever you’re doing?   KP actually remarked that she liked twitter better when fewer people were on it, so there was less information to follow.   Again, the paradox—you can always arrange to follow less on twitter.  The problem is the “insurmountable opportunity”—some of that new content might be really important.   But right now we are trusting to luck to see that content.

I proposed researching some tools that, instead of relying on luck to determine which tweets you see, instead figure out the most valuable ones to show you.  I don’t believe that filtering by person (a big mix of different interests) and hashtag (unreliable, often nonexistent) is the best way to locate the tweets that are most useful to me.  But KP poured some cold water on this idea, arguing that tools that improved your ability to filter tweets would just lead to people following more users, such that they got swamped with tweets again.  More generally, that regardless of what information filtering tools we get, we will always push them to the limit of delivering too much information.

I realized I’ve experienced this myself—I used to visit various web sites to gather information. As I began to find it burdensome to keep up with all these web sites, I ultimately switched to an RSS reader to make it easier for me.  But that has simply allowed me to subscribe to more sources than I was following manually, such that I am again feeling swamped by my information feeds.

Does this mean that any assault on the cliched “information overload” problem is doomed, since whenever we fix it people will load up more?  It seems the only hope is to convince people that they don’t actually need the information they are gathering.

This idea actually relates to another line of our research, on note-taking.   People like to write down all sorts of little scraps of information.  But according to data we’ve logged from our list.it notetaking plugin for firefox (13,000 users—you should give it a try), a lot of those notes are never retrieved.  So why are they written down?  Perhaps it’s because people worry they might need them later, even though they never do.   Something similar seems to be going on with information streams—once they exist, people start to worry they might miss something important, even though they never worried about it before.   If we could somehow convince people that they could find anything that really mattered, they might become less gluttonous followers of information.

And this leads to another project of ours, FeedMe, described in a recent post on this blog.   I’d love to stop following a bunch of my newsfeeds, if only I could be confident that the really good bits would be brought to my attention.  There are collaborative filtering tools like Digg, but I don’t trust them to know what I like.  FeedMe is instead based on having friends forward interesting content to me.   I trust my friends more than any algorithm; if enough of them read a given blog, I can stop on the assumption that they’ll forward interesting content to me.   But right now I don’t have good feedback on how many of my friends are reading what blogs.   That might be an interesting feature to add to FeedMe.  An alternative might be for a group of friends to “divvy up” a blog, each reading a subset of the content and deciding which to forward to which friends.  This would need some supporting interfaces, of course.