>>Can one really make quantitative predictions using evolutionary theory? On a
>>micro-change level perhaps you can??? But on a larger scale I would have
>>thought genetics and "the relationship between DNA and functionality" is
>>not sufficiently well understood to make any realistic quantitative
>>predictions.
>"...genetics and "the relationship between DNA and functionality" is not
>sufficiently well understood..." What do you mean by this?
I am not a biologist, so correct me if I am wrong, but my understanding
of the current state of genetics knowledge is something like the following.
We know that the construction of an animal's body is guided by information
encoded in DNA strings. We know how the reproductive process combines DNA
from the mother and father to form new DNA strings. We know about various
ways in which DNA strings can be mutated. I believe that currently
geneticists are trying to map the human DNA strings. As I understand it,
we have some knowledge about exactly how DNA directs the construction of an
animal's body, but this knowledge is far from being complete. Similarly, we
are only just starting to be able to work out how modification of DNA
affects the development of the body.
So, we have a reasonable understanding about how new DNA are formed from old.
We have some understanding about biological systems, such as how the heart works,
the lungs work etc. What we don't have (correct me if I'm wrong), is a good
understanding of exactly how the information encoded in the DNA, translates
into functioning biological systems like the heart etc. That is what I mean
by not well understanding "the relationship between DNA and functionality".
We don't well understand this relationship, nor do we well understand
how changes in DNA strings (brought about by reproduction and mutation)
are going to affect the phenome (resulting animal body). As a result, it makes
it very difficult to make quantitative predictions from evolutionary theory.
I would suggest that it even makes qualitative predictions difficult. The
programs Glenn talks about - genetic algorithms - have been used with varying
degrees of success in a range of areas. The success of genetic algorithms
is highly dependent on:
1. the formation of new DNA strings from current strings in the population;
2. the mechanism for forming phenomes (the "body") from a given DNA string; and
3. the "fitness function", ie the criteria one uses to determine which members
of the population should be allowed to mate more.
If we don't have the right combination of the above three, then genetic algorithms
will perform very poorly. So, if we want to see how well "real genetics" should
work, we really need to know more about 1, 2 and 3. Likewise, if we want to
know what kind of intermediates evolution should present us with (and whether
it should present us with any intermediates), we really need to know more about
2 and 3.
Mark.