If I were to write a list of my ten favourite things, LastFM would probably be in it (the others being: lentil soup, Summer Roberts, Autumnal colours, Saturday night pizza night, Alan Shearer, Flickr, a milky cup of tea, the smell on hot days after it’s rained and in-ear headphones).
Having said that, its main premise, of suggesting music I’d like, really doesn’t work for me. I know it works for other people, so it must just be me. I think the problem may be that my music taste is too mainstream. I do like some stuff that would be considered cool with the indie kids, but I also like a bit of pop.
The typical LastFM user is going to be pretty tech savvy, which for some reason tends to coincide with people who like to listen to as new and as obscure music as possible. Why nerds don’t listen to old or mainstream music in itself would be an interesting sociological experiment.
Just look at a few total plays stats to prove my point. Britney Spears, played: 6,318,579 times. Modest Mouse, played: 23,671,229 times. I’m a big fan of both, but I know if you took a sample of the whole world’s music listening population, Modest Mouse would not be four times more popular than Britney. Death Cab for Cutie, played: 34,656,728, which is three times more than Madonna (12,293,959).
I could spend the next 50 years of my life happily analysing the stats from LastFM. One more interesting thing I think I’ve found. If you go to the top tracks page they show the week’s most popular tracks. They display a normalised bar chart showing the number of listeners against the number of plays for each song (they have to normalise it as you’re always going to get more plays than listeners). Anyway, as a consequence of the normalisation, if the number of plays line is longer than the number of listeners line it tends to be a new track, where as an older track will have the number of listeners as the longest line. It makes sense, people tend to listen to a new track on repeat, so you get many plays per user. Just seems kind of neat that you can work out a track’s age by the proportion of unique listeners to total plays.