I discovered #4 through him and I agree both with his summary and the paper itself --- canonical Huffman is astoundingly simple if you look at it in the right way, yet almost everyone doesn't.<p>ryg is also a member of the Farbrausch demogroup, which has released some amazing demos. This one is probably the most memorable: <a href="https://www.youtube.com/watch?v=Y3n3c_8Nn2Y" rel="nofollow">https://www.youtube.com/watch?v=Y3n3c_8Nn2Y</a>