This is plain amazing. The practice of adding bugs and silly code to a fully tested and test covered project allows you to spot problems, not needed code, and inconsistencies so easily!
I think the reason for that is because it's not your code, so it's basically evil-mode on. You are basically acting as a super efficient version of a mutation testing library. It would be great to host an event where people could refucktor some famous libraries and at the end of the day find a way to extract useful hints for the library.
Prof Byrd gave a talk yesterday at TwoSigma, in NY.
The talk was inspiring and exactly what I was expecting from one of the authors of The Reasoned Schemer.
He showed a Scheme interpreter in a few lines, and Barliman, which is a tool that generates code starting from test cases.
Time required to complete a chapter is increasing by a lot, and I feel like I am not getting my answer correct most of the time anymore, I should just slow down and understand it, there's no use in rushing / translating exercises from Scheme to Clojure.