So true. Sanford’s model comes across as, “see, we have mathematically disproved evolution”, and his audience not appreciate the math can be sound, but fail to properly represent nature. It may be possible that working scientists too become overly attached to their models.
I would not want to press the analogy too hard, but when I see these curves ending in genetic catastrophe, I am somewhat reminded of the black body radiation ultraviolet catastrophe, which was an entirely solid mathematical formulation based on reliable physics, except that the catastrophic outcome was never observed to materialize. Planck saw that it was the smooth curve that was the issue.
Back in public school, everybody has learned from their ill informed classmates that it has been proven that bumblebees cannot fly. A less banal and more recent update to this sort of discussion took the form of a recent Scientific American article titled No One Can Explain Why Planes Stay in the Air, which elicited a flurry of letters to the editor from aeronautical engineers in fact offering explanations. The truth is that while we know much, there remains more to learn. Models are abstractions which at best approximate nature.
Models become more useful the closer they track the actual messiness of nature. Having followed some of the vigorous disagreement here between very qualified individuals, it is apparent that the existence of drift, selection, and the spectrum of mutation is not in dispute, but there is much room for refinement in the weighting of contribution and compensation. This is the useful role of models; it is the divergence between prediction and observation which tells of the trail to follow. They are not meant to eclipse the data itself.
Planes do fly and living things still populate the earth, so models which conclude to the contrary cannot be complete or correct. @glipsnort has identified three premises of Sanford’s model which GE presumes but none of which has been demonstrated to be justified. In my mind, all such models, whether from Sanford or in the conventional literature, to be realistic have to eventually account for effects of classes of mutation, be they in coding, regulatory, structural, or DNA with no discernable function. If all mutations are held to be deleterious, whether slightly, ever so slightly, or infinitesimally slightly, what exactly is deleterious about mutation in pseudogenes, already degraded ERV segments, and so forth, must be identified and physiological impact detailed, or else the reasonable conclusion is that such mutations are of zero impact, individually or in aggregate. There may be considerable curve discontinuities in the zone of deleterious but not selectable mutation. The model must account for the relationship of genome conservation vs fitness distribution. These demands are not “rescuing devices” for evolution, this is what models are for - to sort out the underlying principles of the empirical world. As attributed to Einstein, everything should be as simple as possible, but no simpler. A population model and set of parameters cannot be so simple that it demands that viruses, flies, mice, and given the antiquity of the earth, everything else, be dead a hundred times over or else we would not be here to do the modeling.