Randomness vs. Information
Updated: Jun 27, 2020
As promised last week, here is the fourth and final part of my recent debate on the Friendly Atheist Facebook page. To be fair and avoid any misleading impressions, I should note that even though the debate ends with my final comment, that is not because MV quit the discussion. MV and I are continuing the conversation through private messaging, but that discussion will remain private. Below is the rest of the debate.
KH, again, the genetic algorithms need design for the environment, but the design solutions , the optimisation - didn't pre-exist in the system. That is the sole reason for their use in engineering - to come up with designs that our intelligence couldn't create on our own. The information emerges from a process of mutation (ramdomness!), selection and heredity. Information comes out of the system that didn’t exist before the simulation was launched. No intelligence creates the information. The information arises from a simple set of rules and the brute force of mutation, selection and heredity on a very large scale. Which are the same processes that exist in nature, that living cells go through. "Uphill" or "downhill" is a term that is purely given in hindsight. It is hard to see in the moment what will be beneficial, and what will be detrimental. A mutation that could cause death in one environment could be useful in another - just like what you see in genetic algorithms. I find it hard to believe that you can't understand how a mutation can be beneficial. Do you understand how DNA and proteins work? Is it that hard to see that a change in a protein shape can help a cell, and thus an organism, to be more competitive in its environment, and that the organism will carry the mutation in its offsprings that will benefit from it? Just like if you try to compose a song - and suddenly you hit a wrong note by mistake. Most of the time you won't think much of it, maybe curse yourself, and keep moving on. But once in a while, the note actually sounds good! And you’ll keep it in your song. And it will make your song better.
MV: Another word on that: "Moreover, in nature, “fitness selection” does not create new information, but only “selects” existing information. " Yes and no. The new information is created by mutation - but is only decided to be useful or not, by natural selection. This is one idea that you can see happen in genetic algorithms, regardless of how or who created the simulation - new features and optimisations arise from mutation and selection. If you agree that genetic algorithms are useful tools for design, then you HAVE to admit that complexity, and even apparent perfection, can arise from the combination of simple rules, randomness and a selection process. From that then, why can't you understand that those same processes happen in nature? That the complexity and apparent perfection of living things doesn't need a guiding hand, or a pre-existing design?
MV, I am not claiming that mutations are never “beneficial.” Rather, the claim is that mutations do not unambiguously create new information…they only alter or distort existing information. My “uphill” and “downhill” terms have nothing to do with whether or not a mutation is “beneficial,” but rather, whether or not a mutation ADDS genetic information, or if the mutation simply destroys or distorts existing information. Example 1: chromosomal mutations that lead to antibiotic resistance in bacteria. The mutation causes a loss of cell function that prevents the bacteria from reacting to the antibiotic. This is typically seen as a beneficial mutation. While it may be beneficial within the environment of the antibiotic, the bacterium digressed genetically and became defective. Once you take away the antibiotic environment, the natural bacteria replace the mutant bacteria (Recommended resource: Dr. John C. Sanford, Genetic Entropy, FMS Publications, 4th Ed, 2014.). While function and information is distorted, no new function or information is truly added. Example 2: The hairless Chihuahua dog. The lack of hair may be a useful adaptation for environments of extreme heat. However, the loss of hair is a degeneration…there is no creation of new function or new information. Yet, in order to have molecules-to-man evolution, evolution must somehow generate brand new information. Copying errors and even duplication errors of DNA within single-celled organisms over millions of years are not going to create the “instruction manuals” that define human beings. For example, if you take a short literary work, such as the Declaration of Independence, and copy that document a billion times over, and allow occasional copying and duplication errors, will you ever end up with a totally different and longer literary work? Even if you allow the “brute force” of natural selection to eliminate the distorted copies that do not make any sense, are you ever going to end up with something like Harry Potter and the Order of the Phoenix? Even if you can hypothetically find a way to transform the Declaration of Independence into a Harry Potter book step-by-step, what if natural selection destroys the in-between distortions? If the "in-between" literary works do not make any sense to a reader, and therefore, the works are "thrown out" and are unable to "reproduce," then wouldn't the step-by-step process become literally impossible? With that being said, here’s my question for you: Can you give me an observable example in nature where a mutation created ADDED function and information, as opposed to just distorting existing function and information?
One final note for the blog reader: I did play around with the “boxcar2d.com” algorithm that MV sent me earlier in the debate. The concept is that random “mutations” can design a car. However, the algorithm is already pre-designed to create various shapes and components such as wheels, axles, and so forth, to develop a “car.” Yet, you will never see this algorithm turn a “car” into something more complex, such as an airplane, unless the necessary adaptive design was already built-in to the algorithm. Therefore, the algorithm does not help make a case for molecules-to-man evolution.
I engaged in another debate recently on the Friendly Atheist Facebook page, with up to ten atheists joining in. I will begin sharing that very action-packed discussion next week.