Mainly for clarification, disambiguation and to match experimental evidence?
My theory, is concerned with "Macro mutations" and in its plainest form just expands on what is already known. I will use the term DNA subroutine for a gene that can be switched on or off to express the macro mutation in question.
Fact: an organism under stress exhibits considerably higher mutation rates.
Theory: Different types of stresses will confer different spectrum of mutation - ie. the mutations will favour more likely beneficial mutations for a predicted future requirement which the stress would signal.
The mechanism proposed is that over a window of genetic experience (say a million years) the DNA stores information regarding which mutations were more appropriate for the given stress and which ones weren't. A "bank" of hundreds of thousands of DNA subroutines that has built up over time and proved their worth are either expressed or switched off to save metabolic resources. If a subroutine has been switched off long enough it can be relegated to "junk DNA" status and will not further be trusted in the field due to it being no longer valid in the new context (or from an orthodox perspective become unusable due to genetic drift)
A word on micro mutations:
Micro mutations are classified as completely random errors in duplication of genes. These are strongly evident and well studied. However, the orthodox view is that the only way for these micro mutations to avoid eventually destroying the function of the gene in question is for them to be field tested by natural selection. To put it another way, the whole organism has to die before reproducing to avoid one crucial micro mutation from being copied. I find this argument incomprehensible. It is like as if the only way to avoid errors in programming the oxygen intake valve of the space shuttle is to launch it anyway, let it crash and avoid using those blueprints again. I call it the crash, burn and learn concept. It might be alright for a virus with millions of launches every second, but I don't think it would quite be so good for the Emperor Penguin.
This lends itself to my belief that there is something else at play other than natural selection. There must be some sort of error correction or testing mechanism on individual DNA subroutines, and if there is, they can just as easily apply to non-expressed genes (or, much more likely, there is a system that expresses these, but localises them for testing only, thus suppressing the evolved purpose of the DNA subroutine for that generation at least).