home.
I did my undergraduate thesis in cultural evolution. Paper can be found (as pdf!) here.


In particular, I looked at the way graffiti topics are distributed across library desks.

The core questions were:

1. Is graffiti of a particular topic (say... race/nationalism) more likely to be found with other graffiti of that topic, or is graffiti in the same topic evenly distributed across desks? The underlying idea here is that people read graffiti that's on the desk and then are more likely to write graffiti in the same topic, which is a kind of positive frequency dependence.

2. Does this clustering (or not) depend on the topic? (aka are some topics more likely to stimulate creation of new graffiti in that same topic?)

What did I find?
Now, of course deciding how to categorize the graffiti is a bit of a subjective process. More details about that in the paper! But the basic outcome was yes, some topics (such as race/nationalism) did seem to cluster, others (love/romance) didn't! Unreported in the paper is the interesting observation that it looked like the most rare topics sometimes clustered the most - on one desk I remember seeing a list of about 10 different entries (in different handwriting and pen color) of types of cars. Someone wrote "Audi 2000" and people took it from there!