Wednesday, February 23, 2011

Why most bands suck by the time you've heard of them

Let me a describe a situation that long characterised my music listening, and see if it applies to you.

The bands I like always had one great album that was inevitably in the past. Occasionally you'd get to see them live and they'd play their old stuff, but they seemed a bit tired and past it - you never got to hear their good stuff while it was actually current.

I think I figured out a rough model of why this should be the case.

Suppose that each band has a underlying quality distribution - each time they write an album, it's an independent draw from their base quality distribution. Sometimes it's a good one (i.e. The Joshua Tree), sometimes it's a bad one (i.e. How to Dismantle an Atomic Bomb). The distributions will vary across bands in both mean and variance.Some bands have a high mean (e.g. Radiohead), some have a low mean (e.g. Nickelback). Some have a high variance (e.g. Bob Dylan), some have a low variance (Red Hot Chilli Peppers)

Friends don't let friends listen to Nickelback

Now, suppose further that there's a fixed quality threshold required in order to achieve popular success. You've got to have at least one good album to become popular. Additionally, suppose that news about bands only diffuses slowly - not everyone finds out about a band at the same time.

So how will this play out? Well, let's take the set of all bands who cross the quality threshold for the first time. Some of them will be truly talented and have a high distribution mean. Being above the quality cutoff signals a long line of good future albums (e.g. Death Cab For Cutie).

A lot of the time, however, being above the quality cutoff signals the band just got lucky with that particular album. After that, you get reversion to the mean. Think Jason Mraz or Ryan Adams. They have a couple of awesome songs ('I'm yours' and 'Desire' respectively), and whole albums full of complete crap.

This sketch of a model generates a couple of predictions:

-The average band will have their best album as their first major commercial success. After that they'll never have another good album

-The bands that have their best album as one of their later albums (Death Cab for Cutie with 'Plans', The New Pornographers with 'Challengers') will be more likely to to release subsequent good albums. These are the high mean bands, not the bands that were lucky.

-Related to the above, if a band releases a second album that's better than their first breakout album, it's a very strong signal of band quality.

-The chances that you get to hear their good stuff depends on how connected you are. If you hear about new bands quickly, you'll get to see them at their prime. If you're late in the loop, you'll always find out about them too late.

The optimal strategy in this setup is to go to see bands as soon as you can after they become popular - that's the high point, and they'll rarely get any better.

Which is a sad conclusion, but certainly sounds like the world we live in, doesn't it? Dire Straits was probably the best band in the 1980s (bar maybe U2), and yet Mark Knopfler has never lived up to the early promise since.

No comments:

Post a Comment