Paleontologists operate in a perpetual Catch-22. We study the most complex events in biology over the broadest ranges of time and space. We uncover various patterns in the history of life, and we wish to measure the multitude of influences that produce them. But we deal in frozen events long past. We can neither experiment nor manipulate; many forms of the past do not even have close analogs among the living. How then can we tease apart and measure the various inputs to provide materials for a theory of macroevolutionary change? One strategy calls for the identification, measurement, and subtraction of general effects in order to identify the more specific causes of a pattern. In allometric studies, for example, we invoke basic geometry and physics to predict the effects of size itself. Larger mammalian bodies carry larger brains. Body size increased in human evolution, but the correlated rate of increase in brain size is vastly greater than the laws of scaling predict. Thus, unsurprisingly, our large brain records the operation of causes specific to our lineage, not merely the general effects of scale (Pilbeam and Gould 1974). What general effect might we try to identify and subtract from the history of life? The patterns produced by random change should be our first consideration. (If we keep at it long enough, simple coin flipping will generate some impressive runs of heads or tails, taken by the uninitiated as a sure sign of causal order. And if the history of life provides anything, it has given us a multitude of flips.) No effect could be more general, for our stochastic models deny the uniqueness of time and taxon, which has served as a cornerstone for causal inference in pa-