Prompted by an interesting essay posted by Jim Stodgill, I’m wondering whether any has tried (and whether it is even possible) to measure the actionable utility of our various ambient information sources (twitter, blogs, status updates). By actionable utility, I’m trying to carve out the obvious entertainment utility (we certainly like hearing from our friends or reading about interesting new stories) and focus on whether the information content has any noticeable impact on us over the long term. In other words, does the fact that we’ve read a particular tweet cause a change in any future action? If we don’t remember the tweet, obviously it can’t. But even if we don’t remember it perfectly, it may have an impact when some future situation leads us to track it down (or something it referenced) and use it for a future decision. On the flip side, even if the tweet did contain some information content that is relevant to a future action, it could still be unimportant if we get that information through some other channel before we need to act on it.
Just to throw out some candidate hypotheses, it seems possible that
- Tweets deliver no information of long-term use
- Tweets deliver of information of long-term use, but are redundant with other, perhaps slower, channels
- Tweets deliver information of long-term use that would not be received any other way.
Of course the true story may be gray—most tweets in Category 1 and a small number in Category 3 for example. But in any case, understanding this could really help us evaluate the proper use of ambient information in our work. We all complain (see Stodgill’s essay) about the overload brought on by all our ambient information tracking. But in the same breath, we justify its use (especially during working hours) by implying that we really need the information it delivers. I find this assumption questionable—see my previous blog post on Information Glut or Information Gluttons—because if we really needed the information it wouldn’t be safe to just hope we happen to be looking at twitter at the right moment. While our ambient information clearly has huge value as entertainment and social grooming, it’s conceivable that it has no functional utility of the sort I tried to define above. There’s a lot of hype now (for example in the same blog that posted Stodgill’s essay) about the “real time web”—this idea that we need to find out about everything as it happens. Unless your in the news business, how much does that timeliness really matter? How often does a tweet lead me to suddenly change to a different task? How often does it change my future actions at all?
Just for comparison, consider traditional newspapers—many of us read them out of a desire (or even a sense of responsibility) to keep up on what is happening in the world around us. But it is unlikely that I will ever have cause to act on our knowledge that the British Labour party may want to ditch Gordon Brown. Therefore, I don’t read the Times at work—it’s entertainment, at home. Does twitter/facebook fall into the same category? How about Lifehacker, given that I’m doing research on personal information management? Can we differentiate useful news from entertainment?
Doing so could have practical consequences. For example, consider the ubiquitous “thumbs up” button, offered by various information-tracking tools to let you say “I like this, give me more stuff similar to it”. Right now that “like” conflates two rather different notions—”fun” and “useful”. The system doesn’t know the difference, and delivers entertaining information when I should be concentrating on useful information at work. If we split this button in two, I could separately train a system on “fun stuff to entertain me” versus “useful stuff that needs to interrupt me while I’m working.”
One problem is that I’m not convinced we could effectively make use of a “useful” button. After all, as I said at the beginning, a true measure of functional utility is based on whether I make use of the information at some distant future date. It would be at that moment of retrieval/use that I could accurately click the “useful” button, if it could somehow be present at that time. But perhaps this approach could be inverted—if I were able to effectively train a classifier on “entertaining” (ie not useful) then I could build a system that set those “entertaining” items aside until I was off work—then, with some self-discipline, I might be able to reduce my at-work information overload.
But coming back to the original question: has there been any serious (non-anecdotal) study of the degree to which blogs and microblogs carry information that impacts my future behavior? Comments/responses are welcome.