Ever since I did my first witness-quality test at University, I have strongly believed human perception is open to surprising inaccuracy; however, I never expected that inaccurate perceptions would be an evolutionary advantage:
Of course, Hoffman doesn’t include his data, so the experiment showing accuracy perception of reality loses could be flawed. However, hanging my disbelief on the idea that his theory on rampant inaccuracy is wrong due to inaccuracy is – while toothsomely ironic – a touch sceptical given my own lack of evolutionary biology experience.
A more profitable line of defence might be an analogy with Newtonian physics. I know Newton’s theories are inaccurate, but how relevant is the difference between them and more accurate models in most situations? In theory, a collision between me and a car will only average to the most likely state over many instances, but the number of variations in which the collision is subtly different but still painful are much greater than the ones where the force goes somewhere very odd; even knowing there is variation, it is as sensible to avoid getting hit by the car.
Of course, as a support for magic, flawed perception is most useful: I can slip my desires between the cracks. But such a level of inaccuracy doesn’t help there either: maybe the world is much more malleable to thought than expected, but what use is that if the outcome of changes to reality is as hidden as the true starting state?