Methinks It Is Like A Weasel

In his book The Blind Watchmaker, Richard Dawkins gave a simple computer simulation to demonstrate evolution by natural selection. The book was written in 1986 (when computers arent' all that common) so I suppose the computer simulation was a big deal then. Richard chose the phrase 'Methinks it is like a weasel' from Shakespeare and showed that the cumulative power of mutation and natural selection could reproduce the phrase without an intelligent designer. I found a few articles attempting to refute this simulation here and here.

It didn't take me long to produce my own simulation (just for fun). I did mine on Matlab. I did it entirely on my own and so, some of the methods I used and the assumptions I made may differ from other similar simulations. I got the following results after almost 2000 iterations:

Gen. Output
0: KTEKBMJOWEPSVNYVBXMJOBAXIETE
5: KTEGBMJIWYPSVSYVBIM OBAFFETE
10: WHELBMJWEIPSISNVBIM OBMHJSBJ
30: LHSLBNJWEIP ISDLBKD GBUHFSEJ
50: KHTLDNJREIU ISDLHKE A UEFSEJ
80: KGTGINKREIU ISDLBKE A XEFSEL
100: KFTHINKR IU IS LIKE A XEFSEL
200: MFTJINKR IU IS LJKE A XEASEL
400: METKINKS IT IS LIHE A VEASIL
800: METHITKS JT IS NIKJ A WEASEL
1600: METHINKS IT HS LIKB A XEASEL
1900: METHINKS IT IS LIKE A XEAWEL
1928: VETHINKS IT IS LIKE A WEASEL
1929: METHINKS IT IS LIKE A WEASEL

Here's how I did mine. I started with a pair of randomly generated phrases. Each phrase is 28 characters long, each character representing a gene. I bred the pair and produced 10 offspring. The offsprings had an 85% chance of inheriting the gene from the fitter parent, 10% chance of inheriting from the less fit parent and 5% chance of the gene mutating. Then I sorted the offspring according to their fitness. Fitness here is defined as how closely they matched the target phrase. For each gene that matched, the offspring received 1 point and for those that did not match, a penalty of the percentage difference was incurred. I bred the two fittest offspring to produce another 10 offspring. This was repeated until the target was reached.

One of the articles said that the random generation 'is repeated for only the positions where a match did not occur', but the author is mistaken. I never specified in my code for the random generation to stop once the match occured for any one gene. I computed the overall fitness score and allowed each gene to change even if it has already matched the target gene.

I think the articles in 'Answers in Genesis' are both wrong to think that this simulation is meant to prove evolution. No, it doesn't prove evolution. Richard Dawkins never claimed to prove evolution using this simulation. All this exercise does is to show that using SIMPLE rules - using reproduction and random mutation, we are able to produce meaningful objects which at first seemed highly improbable. If I were the designer, I could, in this case, type in the meaningful phrase, but as this simulation demonstarted, even without a designer, the meaningful phrase could be produced. In nature though, there is no target phrase. Evolution is blind. Nature is much more complex in that the measure of fitness has to take into account many more factors which are ever-changing. Nonetheless, the principles are the same and that results in all the wonderful creatures and plants we see in nature.

Technorati tags simulation, evolution, natural selection