SCIY.Org Archives

Originally posted on sciy.org by Rich Carlson on Sun 06 Apr 2008 08:00 PM PDT  

If complexity metaphors can be problematic when applied to social phenomena because they often reduce historic inequalities of socio-economic status to mere patterns of self-organization, metaphors of increasing complexity can be misused when combined with socially constructed ideas of progress. Stephen Jay Gould at least thought as much and this review of The Game of Life outlines some of the problems with equating increasing complexity with directional progress

R. McNeill Alexander

Full House: The Spread of Excellence from Plato to Darwin by Stephen Jay Gould, Harmony Books, $25, ISBN 0 517 70394 7. Published on 7 November in Britain by Jonathan Cape as Life's Grandeur.

DIAGNOSED as suffering from abdominal mesothelioma, Stephen Gould went to the library to read about the disease. There he found the statistic that the median time from diagnosis to death was eight months. His case sounded hopeless until he wondered why the median was given, rather than the mean. The reason, he surmised, was that a few long-term survivors were skewing the probability of dying to the right. Plainly, the distribution could not extend far to the left (no one survives for less than zero time after diagnosis), but it might extend to the right over many decades. Happily, Gould's place was on the right tail of a strongly skewed distribution, and one improved by recent advances in treatment.

His new book Full House is about strongly skewed distributions and their implications for the concept of progress, in baseball and evolution. In baseball, a batting average for the season of 0.400 is exceedingly good. The best professional players before 1930 frequently achieved 0.400 averages, but no one has done so since 1941. Gould tells us that we should not conclude that his countrymen are not the men they used to be (indeed, it would be very odd if standards in baseball were falling in contrast to the manifest improvement of performances in athletics). He finds that the authorities have fiddled with the rules from time to time to keep the mean of the distribution fairly constant at 0.260, so the mean tells us nothing about standards. The standard deviation has fallen steadily over the past century, however, a trend that he attributes to improved standards. If the mean is held constant artificially while the standard deviation falls, the scores of the very best players in successive years can be expected to fall.

The discussion of baseball may be read for its own sake, but many readers are likely to value it more for the light it throws by analogy on the largest theme of Full House, the concept of progress in evolution. Gould attacks the perversity of the traditional view of progress through an Age of Invertebrates and an Age of Reptiles to an Age of Mammals in which, despite the name, most individual organisms are bacteria and most species are beetles. He prefers not to focus on the mammals at the right tail of the skewed distribution of complexity, but to look at the whole range.

The modern horse is often thought of as a pinnacle of perfection, the culmination of evolution of grazing, running mammals. Gould is less complimentary. He sees Equus as a remnant of a remnant, the sole surviving genus of a formerly successful family within an order, the Perissodactyla, that has fallen on hard times. The history of horse evolution is not a study of progress, but a tale of a promising start leading to ultimate failure. It starts with Hyracotherium, a terrier-sized animal with many toes on each foot and low-crowned teeth. A path leads from it through a complex evolutionary tree of many, diverse species to the sole survivors: large modern horses with a single toe on each foot and high-crowned grazing teeth.

This may seem like inexorable progress if we ignore fossil genera that are off the main line of evolution to Equus, but we think of that line as the main one only because Equus is the genus that happens to have survived. If the survivor were Nanippus (small and three-toed, it became extinct only 2 million years ago) our impression of the progress of horse evolution would be very different.

Gould's new book is an expanded version of the presidential address that he gave to the Paleontological Society in 1987. He showed then that apparent trends in the sizes of foraminiferans and the brains of mammals are better viewed as expanding variance with a fixed lower limit of size, rather than as trends of increasing size. Since then others have joined him in refusing to see evolution as steady progress. Gould tells us how Bruce McFadden in 1988 analysed the history of the horses and found a complex pattern of branching with many reversals of direction. He describes how Dan McShea, in a series of papers since 1992, has found no consistent trend to complexity in the evolution of backbones. He also describes what is perhaps the most telling of these examples— G. Boyajian and T. Lutz's analysis in 1992 of the evolution of ammonite shells. These have sutures that make simple curves in the earliest fossils, but became on average increasingly complex as evolution progressed. Boyajian and Lutz used fractal dimension as an objective measure of complexity, and compared ancestors with their identified descendants. Rather than a general trend to increased complexity, they found a tangled web of lines in which complexity decreased as often as it rose, with simple-sutured shells present throughout. If the starting point is the simplest possible structure, evolution will result in increasing mean complexity even if lines of descent are random walks.

The same sort of thing happened on a larger scale, in the evolution of the whole range of living things. The starting point was inevitably simple, probably the simplest structure consistent with life. Starting from there, diversification could result only in the appearance of more complex forms. If increased complexity is seen as advance (which it by no means always is), the course of evolution must seem like progress. But we should remember that simple bacteria persist in immense numbers and formidable biomass. The message is important for professional biologists, but Gould's new book is as accessible to intelligent general readers as its predecessors, in all respects but one; a rudimentary knowledge of baseball will be an advantage.

Gould's message is that it is commonly misleading to focus on the most "advanced" or complex organisms at the extreme right of the frequency distribution. Rather, we should look at the whole distribution, the spread of excellence. It is an important message with great potential for redirecting our thoughts, but I hope we will not heed it all the time. We should often think in the way he advocates, of diversification rather than directional change, but we should not abandon the adaptationist view that is presented so strongly in Richard Dawkins's Climbing Mount Improbable (Review, 27 April, p 48). There are contexts in which we should focus on the extremes and reflect on the modern horse as the ultimate (so far) running mechanism. Gould himself focuses shamelessly on the tail of the frequency distribution when he enthuses about baseball players.

Like other important statements on evolution, Gould's message has potential for misunderstanding. He emphasises that, if there is a limit at one edge of the range of possibilities, diversification can be expected to shift the mean. A drunkard staggering on the pavement must eventually fall into the gutter because the houses on the other side prevent him from falling that way. Creationists will probably tell us that Gould sees evolution as a drunkard's random walk, and has presented a more extreme view than ever before of perfection being generated by chance. But Gould is not telling us that the diversity of life is the product of undirected chance, merely that the distribution of complexity that we see is no evidence of a persistent, directed trend.

Attachment: