My old pal Marco has begun a quixotic crusade against Ockham’s Razor, claiming that it has been used unfairly to browbeat dissenters and prop up flimflammery. Exhibit #1 in his crusade is the hypothesis that natural selection on random mutations is sufficient to explain the staggering diversity of the natural world.
Okay, so there are such things as horizontal gene transfer and the intriguing symbiotic relationship at the beginning of eukaryotes. And trivially, some stretches of DNA will be less stable than others, and some cellular environments will be more exposed to mutagens than others, so mutations are trivially “non-random”. He insists this is not what he means. There must be some “non-random” mechanism to generate “good” mutations. I said, no, this is bogus. I said this at great length through a long rambling argument in the comments on his blog before I got tired of saying it, but having a pretty good stamina for argument I recovered after a few weeks and thought I would come back and say it again here.
1. Heritable random mutations happen all the time. The environment of cellular replication is rife with things that can cause mutations, the mechanisms intended to pick these up don’t always work, and if you irradiate living things it is easy to generate non-viable mutant offspring, because most change is bad. This is the basis for every post-nuclear-holocaust movie ever and my political philosophy. And yet:
2. Changes that are good happen all the time. There are lots of examples of this unfolding in real-time as we speak. Of course, these good changes might not have anything to do with random mutations. But there is no *clear and pressing need* to postulate any mechanism beside random mutation. If there is no clear and pressing need to introduce a new parameter, we don’t do it, because then we are off chasing will-o’-the-wisps all the time. This is why Ockham’s Razor is good. On the off chance a random mutation makes an organism that isn’t non-viable, the random change will propagate through the population.
3. The 747 argument is bollocks. Marco is (I think) convinced for a need for non-random mutations by the dodgy statistical arguments of various creationists and steady-state-universists that random changes are insufficient to explain dramatic changes in speciation, because there isn’t enough time in the universe for little changes to add up to the big complicated changes we see. This is assuming an oversimplified linear y = mx + b of what living systems are like and how changing them works. Living systems are complex systems where arbitrarily small changes on the molecular level can have arbitrarily large effects on the macroscopic level. Think of all those genetic diseases that can be traced back to one little residue on one protein being skew-iff. The butterfly effect is the explanation for butterflies. And any arbitrarily large change that is not too large to make an organism unable to reproduce is allowed.
4. You can only find out what the programme does by running the programme. There is no way for a system to know in advance that a change is going to be good. You don’t know for sure when you add a new ingredient to your omelet, no matter how many other foods you’ve added dried cranberries to and it worked out fine. The Vice-Chancellor doesn’t know for sure when he brings in a midnight to dawn teaching period, no matter how many highly-paid consultants recommend it. To find out what changing a line of code does, you need to run the programme. To find out what changing the genotype does, you need to generate the phenotype. There are no shortcuts.
So, there is no need (points 1-3) for “non-random” mutations and there is no conceivable mechanism (point 4) for “non-random” mutations.
This may seem like a series of disconnected ex cathedra pronouncements. Very well, it is a series of disconnected ex cathedra pronouncements. That is just for brevity. I am prepared to defend them all to the death with copious citations and ninja logic.