[hjhjhjhjhj12]

You're conflating general descriptions of algorithms with implementations of said algorithms. I disagree; an implementation of an algorithm CAN very much so be tied to one particular program. If you ignore function call side-effects entirely (cache states, timing delays, modified globals, etc.), then you would be able to transplant functions from one program to another freely. Essentially, to make functions that are fitting to quickly transplant into other programs (in libraries or not), it actually takes intentional design to make sure their side-effects are limited in scope and predictable.

I think this is revealing of an implicit assumption that we disagree on. I do not accept this idea that genes, in a sense, are like functional-programming functions. Considering the fact that there exist genes that express only in the presence of other genes or specific environmental factors, its very likely that genes DO have side-effects in our metaphor of genes as functions.