I actually didn’t reject this one for being overly geeky. I actually couldn’t get it geeky enough – time doesn’t really make sense here. I also considered “length of x axis.” The problem is it isn’t REALLY recursive. Carnot cycles, for example have the same essential graph. But they’re cyclical, not recursive. That is, they repeat, but they don’t self reference.
Hm. Looks a bit like the start of the Lorenz attractor. I bet there is a joke there somewhere.
I was thinking of something like this when I saw it on Twitter:
http://i.imgur.com/4keDD.png
But it’s not simple enough to be humorous.
@”G. Rutter” simple or not, I find that funny ;)
I do like how it implies that the recursion increases *and decreases*.
Plus, it requires a time machine.
It would be more recursive if half-way through it entered an identical, but smaller, graph. And so on. Ad infinitum.
A medieval-type map with a spirally swirly area labeled “Stay away. Recursion. Ad infinitum.”
This graph reminded me of another graph comic. http://www.xkcd.com/688/
I’m pretty sure there can’t exist a satisfying joke using either a time or “amount of recursion” axis, as you’ll ultimately want them both to be increasing forever (at LEAST you’ll want the recursion axis to be increasing forever, you could use a time machine if need be, but that defeats the purpose of the axis really).
You could draw an even more Ben Greenmanesque nongraph by having a vertical axis “recursion in this graph” and a horizontal axis “how much I wanted this graph to be recursive” and then you could make a number of jokes from there: a single small loop in the bottom right corner, lines that do loop-de-loops with varying precision as they move around the graph (where the horizontal axis becomes more of a measure of apathy)
I love the idea of a rejected jokes segment, getting into the anatomy and design process.
Also, I very nearly wrecked my monitor with a spit-take from today’s “recursion -> recursion” comic.
Yeah, I tend to agree with Evan I think (in that it isn’t possible). I think the reason is, you’d need to graph your own axis (let’s disregard labels), so then it’ll fork off and perhaps inevitably have two values at some point (i.e. it will no longer be a pure function, it’ll have to have branched) [something like that, anyway].
Maybe as a commisatory graph you could have something like “difficulty of constructing recursive graph” vs “comedic payoff”, and determine some (amusing) maximum.
I guess you can fix it with a 3D graph, but it tends to be a bit overcomplicated. Instead of thinking of the variable t for the time, think of the point e^{ti} or e^{2 t \pi i} in the base plane and put the z coordinate as the recursion. You start drawing a unit circle with z=0 and aproaching the first loop it start to go up, then you get a repeating loop at z=1. (yeah, i know the moment it begins to go up i has to become wiggly). Btw, cool reference to 1729.
oh the reference was somewhere else, I really don’t know what I’m doing these days…