quantitative Qwantzology

You guys know Dinosaur Comics, right? Cool.

Let’s read a bit about it from Wikipedia, the Free Encyclopedia:

Every comic contains at least three hidden comments (easter eggs)… The third is found in the RSS feed of the comic and the archive page, being, essentially, the comic’s title.

So I noticed a while ago that the way this “archive title” is used has changed substantially over time. Back in the old days, it was short and sweet, e.g. “lesbians!“. But at some point, it started getting larger and larger until, by the end of 2006, you had such monstrosities as “alternate ending: god says ‘T-REX SOMEONE HAS ALREADY OPENED A STORE CALLED THE RELATIONSHOP’ and t-rex yells to utahraptor that God says the idea is already taken, but that they’re going back in time to prevent the idea from being preemptively stolen, and god’s all ‘I NEVER SAID THAT BUT DAAAMN LET’S GO BACK IN FRIGGIN’ TIME’ and then they all meet marty mcfly“. Hope I didn’t give away any spoilers there?

Anyway, I finally got around to graphing it! Binned by month, with error bars on the means:

Woaaaah that’s actually more interesting than I was expecting. What kind of transition is going on here? Starting towards the end of 2005, and continuing into 2006, the archive-title length undergoes a precipitous rise. It saturates by the arrival of 2007… Why the sudden rise? And more importantly, why the saturation? With continued exponential rise, we could have had 1023-character archive titles. Some mysterious force is at play here.

We may never fully understand the cause for this curve. Or maybe I’ll email Ryan North and ask what’s up.

A completely tangential note on procedure: There’s something interesting going on here philosophically, in that I’m assuming there is some True Perfect Archive-Title Length for any given month — the error bars represent how certain I am of what this Platonic character-count is. As weird as this sounds, assuming the existence of a hidden average is exactly what you do when you imagine that you can describe the growth of the title-text as a nice smooth curve. That’s not what what the raw data looks like at all. It looks like a huge mess of points. But we can imagine that these points are just noisy reflections of some more meaningful underlying moving average. That’s what I’m talking about here.