being wrong vs. being false

The idea that all models that are wrong are necessarily all equally wrong [...] is fallacious in the extreme. Wrong does not have this black/white feature. 'Wrong' and 'false' are not the same. Of course, a wrong theory is also false, but if I'm walking to the shop, I'd rather find my location to be wrong by half a mile than by a hundred miles.

We can say that a theory is less wrong (i.e. produces smaller residuals), without implying that is is more true. 'True' and 'false' retain their black-and-white character, as I believe they must, but our knowledge of what is true is necessarily fuzzy. This is precisely why we use probabilities. As our theories get incrementally less wrong and closer to the truth, so the probabilities we are allowed to assign to them get larger. -- Parameter Estimation and the Relativity of Wrong, Tom Campbell-Ricketts

Is evolution anti-entropic?

Life, including evolution, is chemistry. Chemistry is the probabilistic re-arrangement of matter from lower entropy to higher entropy states. Aesthetic order is not relevant. A genetic chemical which mutates into a longer genetic chemical does not become the longer chemical. It combines with other chemical reactants to form the longer genetic chemical plus some other chemical products. The products have more entropy than the reactants.

A genome which causes a certain large, complicated, intricate organism to gestate does not become the organism. It interacts with a multitude of other chemical reactants to form the organism plus some other chemical products. The products have more entropy than the reactants.

An organism which reproduces itself does not become more of itself. It interacts with resources in its environment, inputting a multitude of other reactants (it itself is a reactant) and outputting a multitude of products, including offspring. The products have more entropy than the reactants.

If the organism reproduces itself better than others of its kind, it absorbs more reactants and outputs more products (including offspring) than others of its kind. Other factors being equal, this means that the new organism is succeeding because it is increasing system entropy faster than its competition.

Evolution does not in general reduce system entropy. The tendency of the system's entropy to increase (as long as you keep track of all the reactants and products) and the system's tendency towards biological evolution are the same principle.

Evolution does not require an open system. What life, evolution, or any non-boring chemistry needs is a system that is far from equilibrium.

Whether a system is open or closed is just a description of which parts of the system you're looking at. Any open system can be turned into a closed system by including its relevant surroundings; any closed system can be turned into an open system by omitting relevant parts. We can close the system of the Earth by including the Sun (as a high temperature reservoir) and the infinite void of space (as a cold temperature reservoir). Or we could closely approximate a closed system within the Earth, in which life can and does evolve, by zooming in on a hydrothermal vent and treating the vent as a high-capacity hot temperature reservoir and the rest of the ocean as a high-capacity cold temperature reservoir. -- g s @ stack exchange

Costs

Rarely is anyone thanked for the work they did to prevent the disaster that didn't happen. -- Mikko Hypponen

Two perspectives

[...] my biggest problem with Frequentist stuff is it makes an objectionable assumption that repeatedly performing an experiment is mathematically equivalent from sampling from a random sequence.

This is equivalent to a strong statement about the Kolmogorov Complexity of data coming out of your experiment. The difference between “I don’t know better than p(x) what the next x value will be” and “the universe conspires so that in repeated sampling the long term frequency of X=x is p(x)” are two completely different views of the physics of the world. Only one of them is compatible with the known facts about the world. Sometimes the frequentist viewpoint is an acceptable substitute for the “physics” (or more generally mechanistic description of a process) but that’s an assumption that should in general be tested by collecting a large sample of things.

So if you’re planning to do Frequentist statistics because you have a plethora of thousands and thousands of data points collected in a stable experimental manner and these pass at least some basic tests for randomness… then I say more power to you. This is vastly less than 1% of most science. -- Daniel Lakeland [ref]

Help

It is the poor who gives alms to the poor -- Japanese proverb