Legal code is more like the Windows kernel than Project Gutenberg

Dan Kaminsky has produced slides showing the "information density" of several different blocks of text, including the corpus of Project Gutenberg, the Windows kernel, and the US legal code. The conclusion? The law has more structural similarities to software code than to the prose in Gutenberg's 17,000 books: "Legalese is a massively structured dialect. Symbols appear in very distinct patterns that are more reminiscent of machine code than text."