Recently, Khovanov and Rozansky have constructed some "homological generalizations" of the HOMFLY polynomial and its specializations (the sl(N) knot polynomials.) I'll discuss some conjectures (joint work with Sergei Gukov and Nathan Dunfield) about the structure of these knot homologies and how they might be related to knot Floer homology and other "gauge theoretic invariants." I'll give some evidence for the conjectures, including a complete computation of the sl(N) homology for two-bridge knots.