Natural selection is no mere idle fancy. Every so often the logic behind it stirs a call for eugenics. Last time around evolutionists acted as sponsors and consultants for over 30 states in the USA, promoting surgical interventions to stall our species' descent into oblivion. Now "Unless we embrace genetic engineering, we will become a sickly and frail species" says Johnjoe McFadden, reader in molecular microbiology at the University of Surrey. Is natural selection so established a theory that we must take to the tumbrills and the guilotine once more?

Natural selection is notoriously hard to critique. Logically, it must contribute to evolution to some extent, just as, logically, friction must to some extent contribute to driving an automobile. But just as friction doesn't actually drive the automobile, natural selection may not be what actually drives evolution. How can we think our way through this? We seem to lack an off-the-shelf set of concepts adequate for pinning down the issue. Complicating matters is the endorsement natural selection receives as the effective agent in evolution from sophisticated statistics that has withstood criticism for more than a lifetime, since 1930. It's notoriously hard to argue against statistics, particularly statistics that no one any longer appears to understand. In biology today it functions almost as divine inspiration. I'm referring to Ronald A. Fisher's "The Genetical Theory of Natural Selection."

I've arrived at a context which may make it easier to evaluate natural selection, and to assess Fisher's work. It's a physical situation that I think presents us with a parallel with the action of natural selection.

Compare the genetic blueprints for the superb fitness of existing species of living creatures to a range of steep smooth-sided mountains—perfect cones of rock. Now imagine each one undergoing random damage. For the genetic blueprints this would be random changes to individual genes, for the mountains this would be random avalanches. What happens to the species’ genetic blueprints would be comparable to what happens to the mountains—just as those mountains would steadily degrade and flatten out as avalanches erode their sides, those genetic blueprints would become progressively degraded and code for less fitness. Eventually the mountains flatten out as a plain, and as fitness sinks below some threshold those species go extinct.

Now introduce natural selection. For the mountains this would take the physical form of resistance to the tumbling of the rocks, and small pebbles occasionally being deflected upwards. Very occasionally they may end up higher than where they started out, even atop the mountain they’re a part of. But they’re only top of a much lower mountain. Similarly, natural selection will lead to the culling of some of the worst damage done to fitness, and to individual genes occasionally increasing in the fitness they confer—so-called beneficial genes. But this is only a small addition to a much lower level of overall fitness. In both cases, the overall effect is of a cumulative loss of the original perfection.

This is so obviously how the world works that it’s astonishing it isn’t widely recognized. Why is that? I believe it stems from an error in Fisher's work. Fisher came up with a statistical formula that he applied to only how the frequencies of beneficial genes can compound to predominate at their allele, no matter how small the efficiency of natural selection is, if you allow it to operate over hundreds of thousands of generations. Where he erred, I suppose, is in not applying this same formula also to the frequencies of harmful genes. If he had I think he’d have realized harm would overwhelm fitness in only a few generations as I think is apparent in the physical parallel with the mountains above. If you tracked only how occasional pebbles would be flung upwards in the course of avalanches you would end with the mountains growing taller and taller over millennia, clearly an impossibility.

I believe the entire modern synthesis rests on Fisher’s 1930 book. Discredit that and bang goes the modern synthesis, and any basis for eugenics. One may not be able to say why eugenics isn’t valid, but one can say “Darwinism” doesn’t make it so, which puts the burden of proof in the other court. Just saying Fisher’s wrong isn’t enough, I‘ve found, because no one takes a verbal attack on statistics seriously, that’s why I instead use the physical comparison above. It sums as, contributions to fitness due to occasional damage to genes that’s beneficial are offset by a continuous massive overall loss of fitness due to harmful damage, against which those contributions can make little impression, leading to rapid extinction. I wish some professional working statistician would re-examine Fisher's work along these lines and publish the results. Any volunteers?