i202 Fall 2009 School of Information, UC Berkeley

Wednesday's lecture about mobile and multimedia IR got me thinking about the great This American Life episode about people who try to quantify the unquantifiable — and specifically about the two artists who tried to distill what elements would combine to make the songs that would be the most and least pleasing to the greatest number of people.

In brief: In the '90s, Alex Melamid and Vitaly Komar conducted a survey and found that people liked songs that were relatively short, about love, and featuring male and female voices and dramatic key changes. On the other hand, people hated hearing childrens' voices, advertising jingles, bagpipes, country, rap, and songs about the holidays. So the most wanted song sounds like some sort of love duet between Meat Loaf and Celine Dion, while the least wanted song ... well, you should really just hear it.

I actually used to own this album, and the liner notes were pretty hysterical, full of charts and graphs that illustrated how people responded to the 1996 survey that led to the creation of these songs. While I don't know where that's ended up these days, it's still worth checking out the notes from the composer, who claims that fewer than 200 people in the world would like the least-wanted song. (By contrast, about 72% of people should like the most-wanted song.)

It gets me thinking all over again: Can sound ever really be quantified with descriptive text? For the record, these artists also launched a similar endeavor to quantify art, with some pretty interesting (if probably not scientifically valid) results.