The Skilled Breeder Scenario

Imagine a dog breeder who wants to develop a new breed with a specific
set of qualities.  This dog breeder is quite wealthy and has spared no
expense for his breeding program.  He has hired a number of molecular
biologists and they sequence the DNA of each dog in his kennel.  Thus,

he is able to select which dogs to breed with perfect skill.  However,
this dog breeder considers recombinant DNA techniques to be “cheating”
and relies strictly on natural processes for any variation of the genes
in his population of dogs.  However, he is happy to alter the
environment of the dogs.  For instance, he has been experimenting with
dog food that contains more soybean and sweet potato and less meat.

Are there limits to what the skilled breeder can achieve?  For example,
can this dog breeder eventually breed a herd of horses from his dogs?
If so, how long would it take?  Could this happen if the kennel/stable
contained a billion dogs and the breeder was given 100 million years to
do this?

From a information theory perspective, there is no barrier for the
breeder to eventually produce a herd of horses.  There are plenty of
opportunities, through the selection process, for the breeder to inject
information into the system.  Imagine that each pair of dogs has 10
puppies in a period of 4 years.  The breeder must select which puppies
will be bred in the next generation and which pairs are to be bred.
Each puppy selection adds log_2(5)/5 = 0.46 bits of information to the
system, so 0.58 billion bits of information are added to a population of
a billion dogs each year through the selection process.  In addition, if
the breeder chooses specific pairs of dogs to breed, quite a bit more
information is added to the system.  Thus, there is plenty of
opportunity to add complexity to the dog population.

At first glance, an estimate of the resources needed to breed a horse
from a dog seems to be straightforward.  First, compare the DNA of a dog
with the DNA of the horse and find the minimum edit distance between the
two.  Next, find the rate for a local mutation (add a random base pair
or delete a random base pair) to occur and be passed to the next
generation.  Combining these two will provide an estimate of the number
of generations and population size needed.  The estimate could be
further refined by taking into account the possibility of non-local
mutations and the fact that genes could be modified in parallel.

However, there is another requirement that changes the picture
drastically.  We also have the requirement that all intermediate forms
must be viable.  It is not a priori clear that an unbroken chain of
living creatures exists between the horse and the dog.

In summary, the skilled breeder scenario assumes 3 things about the
evolution of a population:
1.  Variations in the population are through random (undirected)
processes.
2.  Selection is done intelligently.
3.  All intermediate forms must meet some minimum criterion for
viability.
In the following, I will discuss the skilled breeder scenario in several
contexts and discuss why it it important for understanding origins from
a Christian perspective.

Fitness Function

One can think of an evolutionary process as a search through the space
of possible genes to find the global minimum of a fitness function.  In
this case, the fitness function is as well-behaved as possible, thanks
to the skill of the breeder.  However the search mechanism (random
changes plus minimum viability) does not span the space of possible
genes.  In fact, given two different viable genes, there may not exist a
path between them.

Dawkins’ Weasel Example

Richard Dawkins proposed a simple model of evolution,
METHINKS*IT*IS*LIKE*A*WEASEL, [Richard Dawkins, The Blind Watchmaker]
which has been discussed at length by Dembski [William Dembski, No Free
Lunch].  Dawkins’ model is not an example of the skilled breeder
scenario since it lacks any criterion for minimum viability of
intermediate forms.  An interesting variation of the weasel model would
be to demand that any intermediate string of letters consists of a
sequence of English words (imagine starting with some initial sequence
of words and maybe allowing the total length of the sequence to vary
during the evolutionary process).  This modified weasel model would be a
good example of the skilled breeder scenario: it is no longer obvious
that the target sequence can be produced.  And if it can be produced at
all, it will take considerably longer than an estimate based on minimum
edit distance.

Protein Folding

In the realm of protein evolution, we can also talk about a skilled
breeder scenario.  Consider a gene sequence that is used, after RNA
transcription, to construct a protein.  One can imagine making a series
of random mutations to the gene sequence and selecting sequences that
are close to some target sequence.  In this case, the minimum viability
condition would be that any intermediate sequences must also produce
functional proteins.  Since the vast majority of amino acid sequences
are not functional, the minimum viability condition is a significant
constraint.  See Ann Gauger article “Protein Evolution: A Guide for the
Perplexed” [http://www.biologicinstitute.org/post/57103144912] for
further discussion.

Darwinian Evolution

The Darwinian evolution is equivalent to the skilled breeder scenario
minus the skilled breeder.  The Intelligent Design argument states that
an increase in specified complexity of the population is impossible
since there is no source of that specified complexity.  However, even if
we ignore the issue of specified complexity, Darwinian evolution may
still be impossible.  That is, if the skilled breeder scenario cannot
produce new species within a realistic time scale and population size,
then Darwinian Evolution certainly cannot.

Origins and the Christian

Old-earth creationists generally hold the view that God introduced new
species on the earth over the course of millions of years.  However, the
mechanism for this is unclear.  Were new species simply created \em{ex
nihilo}?  Were they (usually, modifications of existing species, where
God intervened in the mutation process?  Or were they modifications of
existing species, where God intervened in the selection process?  If the
skilled breeder scenario cannot produce new species within a realistic
time scale (assuming an old-earth view of creation) and population size,
that tells us something about the mechanism the Creator may have used to
introduce new species.