Very nice article, as always. However, there are a number of typographical issues with missing parentheses and script letters not being formatted in the text (as well as in the code), which is quite distracting. The use of code with script letters, which I, too, use often, always looks crappy in HTML.
On a more technical note, while I've not yet read Gilpin's paper, it immediately occurred to me that one could use the Lorenz model to achieve the same outcome and a quick search shows a number of papers recently published on this idea (e.g., Google hashing using Lorenz equations). And even simpler, one could use the logistic map, for which there are also many papers. So, I guess it would be helpful to readers to know why Gilpin's proposal is advantageous.