A Fatter Long Tail? Nope.

Chris Anderson posted a really nice illustration a couple of months ago in Visualizing the Rise of the Long Tail. He shows three photos of mountain ranges that cover the same peaks but viewed from a different distances.

This would be a nice metaphor, but the underlying idea is wrong. On the other hand, the pictures are really pretty, so let’s take a look.

I’ve got a few Canon lens brochures somewhere in the garage with similar series of shots demonstrating the perspective of wide angle, normal, and telephoto lenses. Hmm, looking more closely these aren’t different perspectives at all, they are a cheap PhotoShop hack — they just stretch the aspect ratio and re-crop a single photo. Hmm, sounds more like PowerPoint-mind than real perspective, eh?

Anyway, here they are:

The idea behind this visualization is that the long, thin tail is going to get thicker (but stay long, I guess).

Unfortunately, the tail cannot simultaneously be a Zipf distribution (“long”) and be fat. The basic point of that shape is that the tail is both long and thin. If it gets fat it isn’t Zipf anymore. Since every popularity-based empirical distribution is Zipf, betting on that to change is a good way to lose money. The long tail is not going to rise until human nature changes.

Since that visualization is guaranteed to be wrong for overall web traffic, can it be reused for some other idea? It is a neat visual after all. I hate to waste those.

Instead, think of the top one as aggregate traffic and the bottom one as an individual’s traffic. The individual may spend a lot of time with local interests, things that aren’t every going to be a big hit (the fat head of the curve). The middle illustration shows groups of people. The students at Rice University will have some things in common that are not shared by the general public (or even faculty), so they share some “foothill” interests. This probably represents both local (Montrose-area Houston) and preference (ironic humor) interests. Pity, that.

When you add together traffic from related individuals, you see the group interests. When you add together traffic from a random sample (or lots of groups), you see the mainstream, the now-familiar long tail.

Really, all of these are different Zipf distributions, but I think the mountains are pretty, so I’ll stick with those. Maybe the groups should focus on different peaks.

A website tailored to a group should be mainly thinking about the bottom two illustrations. Your visitors will spend a lot of time in the foothills, and that is the ground you need to cover in detail. Community libraries have known this for a long time — their acquisition policy is tuned to local tastes. Paying attention to your customers is great, but don’t pretend that Zipf will bend to your business model.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s