The three-body problem

In a previous life (somewhere between the punk rock bands and The Economist) I was in training to become a physicist. As the math got progressively gnarlier, we eventually encountered the "three-body problem", which is the challenge of computing the mutual gravitational interaction of three planets or other masses. It's practically impossible to solve, but does prove to be a handy metaphor to use in later life. Such as now...

I have, you might have noticed, failed to post for three weeks (other than adding sidebar shoutouts). The reason is that I've got my own three-body problem, with the three bodies being my job (editing Wired), my book, and my blog. (To say nothing of my family, complete with four little bodies who have their own gravitational attractions.) Two of those three are at least full-time jobs, so the challenge for me is to figure out how to fit the blog into the zero-sum game of time.

In principle, this shouldn't be too hard; the blog is about the book, after all, and my expectation was that the book would spin off plenty of tangents, sidechannel discussions and overmatter to feed a steady flow of posts. And indeed it has so far and probably will for a long time to come. But it turns out that just as Wired has a monthly cycle, of which the last two weeks demand maximum capacity, the book writing also has plant and reap cycles. And the last three weeks have been a lot of planting.

Right now I'm in the heavy research phase. That means lots of interviews and data negotiation, compilation and analysis. The good news is that they're both going brilliantly. Yesterday, in particular, was a landmark day for data. I'm not going to give away too much yet, but we basically compared two very  solid (20,000+ titles) data sets of music sales, one drawing from traditional retail (with shelf-space scarcity and poor searchability) and one online (huge inventory and good filters and recommendations).

My thesis had been a Long Tail business (massive variety plus good filters) would show a much flatter demand curve, with a more even balance between hits and niches. And, mirabile dictu, that's exactly what this data analysis showed. Even over the same inventory, when the two data sets were normalized (hits set to the same levels) the online business showed 2x demand for niches. Overall for the same sales of the top-1,000, the online stats showed 50% higher sales of the next 19,000.

So the  Long Tail doesn't just theoretically or anecdotally shift demand from hits and niches, I can now show it does so statistically, too. Next I hope to extend this to other markets. Oslo awaits!

Comments

[RichK]

I've been having a long-tail problem with Netflix/Blockbuster I thought you might find interesting....I don't have time to watch movies, so I end up keeping my three movies for weeks and feel guilty about not watching them etc...I've done this at both NF and BB...eventually I cancel, that is until I come across a long-tail movie that I MUST see, NOW...the only place to rent it is online, and I'll resubscribe, use it heavily for a short while, then repeat problem ; >

The answer, it seems to me, is a la carte at online movie services...I guess a download service could support this more easily, but can't the DVD rental guys do this too?

Tks.

Rich K

[fling93]

I have, you might have noticed, failed to post for three weeks.... the challenge for me is to figure out how fit the blog into the zero-sum game of time.

Don't sweat it. As you know, that's why us readers use RSS. Work on the book!

[Dan Diephouse]

I suppose I shouldn't ask too many question until you post the data but...

What makes the long tail business differ from the retail business and how do you establish the causality between that and the demand? I see the correllation that present, but could there be a psychological factor telling them "this will never be in my local store" which influences there buying? Or is it because reviews and recommendations push them into the long tail as I think you're hypothesizing?

Is the normalization a fair thing to do? I couldn't compare my local record store with Best Buy for instance because buyer preferences will differ from my local area to nation wide.

Just some questions...