The mood of the crowd

The Freakonomics guys are excited about a project that tracks the moods on LiveJournal blogs by hour to see reaction to such events as the London bombings. If the timeframe were greater than two days and one saw longer term vectors, I wonder whether this could be a leading indicator: an upswing of “happy” means greater consumer confidence; an upswing of “bored” means higher TV ratings.

  • Pingback: Blogspotting()

  • I played around over at the site they reference – it’s pretty interesting, but it’s still just a proof of concept. Really interesting things (crosstabs, overlays of different moods, inflection points for changes of mood, etc) will probably be coming at some point.

    The most interesting thing currently is that the data is based on automated content analysis, and they’re just using LiveJournal data as a control to refine their process. The eventual portability of this process to any subset of blogs gets even more interesting. Or consider applying it to the story pitches Steve Baker was talking about today.

  • Yes, I agree it would be a very useful input . I’ve called it– along with many of the techniques that James Surowiecki talks about in The Wisdom of Crowds, “aggregatable declarations”, or “peg” for short (in the tradition of blogging, tagging… pegging). And I make the case that this is the future, and will in fact be more useful to decision-makers than blogging as we know it.

  • Hmmm, looks like similar ideas are flying all around but everyone is just describing them in different ways. I also thought about utilizing a tag to aggregate single word moods that could potentially show the daily “mood of the world” (see second update down). I also wondered what other small bits of information you could join together to bring awareness to things. Of course for this tag idea to work though, a “shared mental model” would have to exist for everyone to agree upon which mood words to use to make the data aggregation feasible. If we are going to be collaborating more and more on a larger global scale in the future, then we’ll need to start agreeing more and more on these rules and standards to use for this “shared mental model”.

    Jon, in your Gatekeeper post, I wouldn’t equate linking to something as making it more “popular”, I equate it with increasing the awareness of it. I mean everyone may go to a website to look at something that a company is doing that is causing a lot of people to get upset but it doesn’t make the company or the website “popular”. If anything the awareness of what they are doing (as indicated on their own website) could cause the company to be very unpopular with people and cause them financial ruin. Sure Google’s Pagerank for them may shoot through the roof but within a week that website may be gone because the company may have gone out of business. After that, their PageRank won’t even exist. :)

  • Nollind:

    That’s the definition of popular I was using “Widely known” and not “widely favored.” The fact that popularity is value-neutral causes some consternation to some search-engine watchers, particularly those who prefer authoritativeness as a value. In that part of the Gatekeepers series, I point out that even Google has hinted that it is going in this direction by registering the name “TrustRank” as a trademark.

  • Nollind, I suspect that one of the advantages of using the LiveJournal reference data is that it contains a restricted vocabulary. Hence, calibrating your algorithm against that data set means you already have a restricted vocabulary. It may or may not be the best one, but what the heck – it’s there.

    Jon, I’d forgotten “pegging” from that excellent series of yours. I’d argue that it might be more valuable for decision makers dealing with groups and statistical norms for behavior, but less so for people needing to deal with infrequent or singular events.

  • Greg– yes, I remember your making a comment on the site then– and thanks again for the compliment! Yes, reading the details, the Freako- authors were most interested in decision-making aspects. Nonetheless, with Internet communities, there aren’t many “singular” events.