Correlated errors are a problem in all sorts of places. Most statistics assume everything is independent; super important to verify that before drawing conclusions.