Sunday, 1 July 2018

Filtering, sorting, copying, mutation

One of the most popular schemes to classify evolutionary processes is based around the idea of selection. Evolutionists distinguish selection from genetic drift and then subdivide selective processes into kin selection, group selection, natural selection, artificial selection, sexual selection, divergent selection - and so on.

I'm not opposed to the idea of selection, but do think that the popularity of the concept has pushed alternative classification schemes into the shade. One of my preferred subdivisions is to start by ignoring the distinction between drift and selection and to divide initially by whether the process involves death or reproduction - or more broadly subtraction or addition. For example, here is an article on the topic adapted from my 2011 book: Natural production and natural elimination.

A closely-related classification scheme involves some other topics which I have promoted: sorting, filtering and copying. "Filtering is now my preferred term for evolutionary processes that subtract. "Sorting" refers to processes that neither subtract not add entities, but instead rearrange them. A focus on frequencies leads to sorting processes being ignored. However, in practice, sorting is common and influences evolution as a precursor to filtering.

One of the things I like about "sorting", "filtering" and "copying" is that they are pretty crisp computer science topics. One of the problems with selection is that is is so general. Saying "selection did it" is a very weak explanation because almost everything counts as a form of selection. Yes, genetic drift exists, but that's just a form of noise. If almost everything counts as being a form of selection, selection becomes difficult to falsify as an explanation.

While filtering obviously coveres "selection by death", it also crosses over into the realm of reproductive processes as well. Males are frequently sorted on "leks" and then the genes of the worst males are filtered out of the gene pool by females. While that is a form of subtraction, it is one which is pretty closely linked to reproduction - and what is commonly described in terms of "sexual selection".

Filters can vary in the time of their application, or filtering that takes place at different times can be modeled as a series of filters. These two approaches are equivalent.

"Filtering" is close to what Dawin originally meant by "natural selection". Darwin argued that "sexual selection" was not a form of "natural selection". The modern meaning of "natural selection" was introduced after Darwin's death. It pictures "sexual selection" as a type of "natural selection".

The picture here leads to the following classification scheme for evolutionary change:

  • Entity count changes
    • Filtering (subtraction)
    • Copying (addition)
  • Entity count remains the same
    • Sorting (rearrangement)
    • Mutation (in-place change)

Organisms moving around is not normally considered to be a form of evolutionary change - so "rearrangements" are mostly omitted or ignored. However, rearrangements are important. They are not completely ignored, but they are downplayed, I claim.

The left-hand labels in this classification scheme are all the information theory / computer science ones - except for the "mutation" category. That one has gone the other way - when the concept is used in computer science the biological term is often used.

No comments:

Post a comment