You're in for a treat. The basic situation is that physiologists are watching it in real-time, neo-Darwinists have a hard time accepting it because "it's not necessary for evolution to work", physiologists then wish they could batter the neo-Darwinists with their microscopes until they relent.

Noble is going over more but as a programmer, the information theoretical argument is close to my heart: DNA transcription, if left completely to its own devices, has a quite high error rate. Correction mechanisms (which evolved completely randomly at some time) then take that error down to practically nothing, and after that randomness is again introduced. Which means that evolution (or, well, how many a genome evolves) is not a random process, but a process employing randomness, enabling it to strategically choose where to mutate. The genome of a bird, for example, if it senses that the current phenotype can't get at nectar, is well-advised to mess around with beak shape genes instead of mitochondrial DNA, and this "different environmental stressors cause different genetic transmission" is indeed what physiologists are observing and I don't just mean epigenetics. It's not that the same result couldn't be achieved by pure, blind, randomness, it's that a genome able to employ strategic randomness is more fit, can adapt faster and more successfully. And as it's a metasystem transition it gets locked in, there's no backsies because any deviation from that kind of achievement is always less fit than one that retains the capacity, same as you don't see DNA-using life suddenly ditching it, being content with only RNA.