This is the function my copy of TAOCP serves. I'm not anywhere near knowledgeable enough to digest all of it, but when I need something and I want to understand the ground theory behind it then to Knuth I go.
My most recent use of it was to convince my team that binary search is in fact not a good interview coding question. It's surprisingly tricky and Knuth explains exactly why!
The part where he explains how a binary tree can be unbalanced, and when, was really surprising to me. Never assume that just because your accesses are random that the tree will remain balanced on average.
