2.1-2.2, 2.4

1. Running frequency analysis on English plaintext is good and all, but in the real world, not only is English with punctuation encrypted, but all types of binary data, including compressed data (which has more entropy, I think). If something is entropy encoded, is it no longer susceptible to frequency analysis attacks?

2. I like how he said that everyone "knows" that substitution ciphers can be broken by frequency analysis. It's kind of like the budding cryptographer's mantra: "run some frequency analysis on it". Do all languages exhibit this kind of statistical exploitability? Chinese has a few thousand characters and the semantics and meanings are largely contextual. I suppose digrams and trigrams take that into account to a degree, but I wonder what Chinese codes looked like.