Old bookstore

I wish I could preserve the smell of the old bookstore.1

This is my last chance to browse in a joint that doesn’t ply my confirmation bias with its hideous machine learning. The bookstore is a random walk with simulated annealing, having honed in on interesting sections. It’s a good way to shake oneself out of local maxima in a way that, say, Amazon can’t quite.


Figure 1: The labyrinthine Cliff’s Books: vector of random walks

Local maxima can be social: “this is what people are reading;” relational: “people who read \(x\) also read \(y\).” What’s missing, however, is the long tail; the forest of parentless roots discoverable by the random walk.2

Old bookstores are optimized for the random walk (peregrination); and the random walk is missing from the bubbles of confirmation bias which shield us from the stochastic, the chaotic.

The internet has become so fucking safe.



It’s a pleasure to write with the pen, incidentally, and propagate it to machine.