Unlock: Dvoretzky's Theorem and Almost Euclidean Sections

Every n-dimensional normed space has (1+epsilon)-Euclidean subspaces of dimension of order log n. Milman's concentration-of-measure proof, the critical dimension n(M/b)^2 for random subspaces and its matching upper bound, the epsilon dependence after Gordon and Schechtman, and what Artstein, Milman and Szarek proved about duality of metric entropy. Not the Dvoretzky-Kiefer-Wolfowitz inequality.