One of the best presentations on what can happen if someone takes a naive approach to network data. It also highlights what is, to my mind, the greatly underappreciated t-distribution, which is typically only used in connection with frequentist Student t-tests, but serves as a generator someplace between the Gaussian and the crazy Cauchy distribution. Also relevant is the Lévy flight which has significance in biology. (See also.)
The message is that a combination of multiple paths, sampling rate changes, and a glitch on one of the paths can make an event appear to occur where there is none.