It was more a Q&A than a lecture, but I was seriously impressed with a lot of what Knuth had to say.
Rob Irving got polynomial time alg for stable marriages after reading a paper by Knuth in French posing the problem, so he got a mention & asked a Q. David Manlove asked if P=NP.
Other questions I remember were on functional programming (meh, not for my work, said he [I paraphrase]), literate programming (greatest thing since sliced bread but not picked up like Tex has been), what has been his greatest contribution (alg analysis), should we learn assembly (yes), is computing art or science (art is man made, as in artificial, science is what we know), book recommendation (Princeton Algs book by Bob Segewick), how he communicates/teaches so well (say everything twice [or in two different ways], have jokes that won’t be understood unless the technical point is understood).
He said one of his chapters was called “utility functions” and that then evolved into one on arithmetic. I often have a class in my programs full of static methods called “UtilityFunctions” that have no home; a phd student commented to me that he “liked it, but it’s not very Java-like”, but so what, Knuth agrees with me :).
He’s just published a collection of papers called “Papers on fun and games”. He’s worked hard, and turned out so much. What I noticed was that his brain seems to be always on, and thinking, and he really explores what he is thinking.