The paradox of the visibly irrelevant

Natural History, Dec-Jan, 1997 by Stephen Jay Gould

These shortest-term studies are elegant and important, but they cannot represent the general mode for building patterns in the history of life. The reason strikes most people as deeply paradoxical, even funny -- but the argument truly cannot be gainsaid. Evolutionary rates of a moment, as measured for guppies and lizards, are vastly too rapid to represent the general modes of change that build life's history through geological ages.

But how can I say such a thing? Isn't my statement ridiculous a priori? How could these tiny, minuscule changes -- a little less leg, a minimally larger size -- represent too much of anything? Doesn't the very beauty of these studies lie in their minimalism? We have always been taught that evolution is wondrously slow and cumulative -- a grain by grain process, a penny a day toward the domain of Bill Gates. Doesn't each of these studies document a grain? Haven't my colleagues and I found the "atom" of evolutionary incrementation?

We have discerned something important, but we have discovered no general atom. These measured changes over years and decades are too fast by several orders of magnitude to build the history of life by simple cumulation. Reznick's guppy rates range from 3,700 to 45,000 darwins (a standard metric for evolution, expressed as change in units of standard deviation -- a measure of variation around the mean value of a trait in a population -- per million years). By contrast, rates for major trends in the fossil record generally range from 0.1 to 1.0 darwins. Reznick himself states that "the estimated rates [for guppies] are . . . four to seven orders of magnitude greater than those observed in the fossil record" (that is, ten thousand to ten million times faster!).

Moreover, and with complete generality -- the "paradox of the visibly irrelevant" in my title -- we may say that any change measurable at all over the few years of an ordinary scientific study must be occurring far too rapidly to represent ordinary rates of evolution in the fossil record. The culprit of this paradox, as so often, is the vastness of time (a concept that we can appreciate "in our heads" but seem quite unable to get into the guts of our intuition). The key principle, however ironic, requires such a visceral understanding of earthly time: if evolution is fast enough to be discerned by our instruments in just a few years -- that is, substantial enough to stand out as a genuine and directional effect above the random fluctuations of nature's stable variation and our inevitable errors of measurement -- then such evolution is far too fast to serve as an atom of steady incrementation in a paleontological trend. Thus, if we can measure it at all (in a few years), it is too powerful to be the stuff of life's history.

If large-scale evolution proceeded by stacking Trinidad guppy rates end to end, any evolutionary trend would be completed in a geological moment, not over the many million years actually observed. "Our face from fish to man," to cite the title of a famous old account of evolution for popular audiences, would run its course within a single geological formation, not over more than 400 million years, as our fossil record demonstrates.