Comparing Evolution and Lens Design
On February 26th, 1969, I offered a research seminar about an innovation in lens design that would put the company far ahead of its rivals and make us all rich. I made for my colleagues an analogy between Darwinist evolution and automatic lens design. Using a projector, I showed my colleagues the following table, and explained it.
Step One
In Darwinist evolution at the beginning let there be a stable population of living animals. In the Automatic Lens Design program, a designer starts with the best design known that is appropriate for a given apparatus, like a reconnaissance camera.
Step Two
Among the animals there are random mutations. Mutations in general are bad for an animal. Often mutant offspring are born dead. Other mutations make the offspring easy prey to predators. These do not survive to maturity and so do not reproduce themselves. Excluding these, let’s take the case of a mutation that is passed on to live offspring. Other animals of the same species are born and reproduce themselves normally.
In lens design, the program makes a systematic variation of all the characteristics. This ensures that no change that might benefit the design is left out. For each variable the program makes a small change. Then, using the merit function, the program measures the global effect of the change, and stores the result. Third, the program returns the variable to its original value. The program repeats this process for every one of the variables in turn.
Step Three
Most of the mutations passed on to offspring seriously reduce the animal’s capabilities. Perhaps only one mutation in a thousand produces a benefit for an animal that helps it survive. Animals with such a mutation may live to maturity and reproduce, while other animals without the mutation may die young. Eventually the mutation will appear in almost all the animals of the species, if the mutation really gives a survival advantage. The diagram in step 2 of the table above simplifies and exaggerates the effect. The individual marked with an “x” has an advantageous mutation that allows it to produce two offspring. Simultaneously it puts other, normal individuals at such a disadvantage that they can only produce one offspring. According to the simplified diagram, in only three new generations the mutation reaches dominance in the population. In nature many generations are necessary.
The automatic lens design program incorporates a mathematical method for calculating a combined variation of all the variables. The method takes into account all the variations that individually reduce the aberrations. If a variation increases the aberrations, the method is programmed to take this variation in the opposite sense. Then the program calculates the combined variation that produces the maximum reduction of aberrations. Each variation contributes in proportion to its design improvement. The program then modifies the design according to the combined variation. This process is much more thorough and efficient than random variation.
Step Four
In both the process of Darwinist evolution and the automatic lens design program there is a return to step one, but with a new initial variety or a new design.
My research colleagues were uncomfortable. They anticipated what I was going to say. I continued with the comparison even though I could see that they were consulting among themselves to discover, if they could, any error in my logic. Using the following comparison table, I asked my colleagues to make an inventory of the advantages of automatic lens design over Darwinist evolution.
In Darwinist evolution at the beginning let there be a stable population of living animals. In the Automatic Lens Design program, a designer starts with the best design known that is appropriate for a given apparatus, like a reconnaissance camera.
Step Two
Among the animals there are random mutations. Mutations in general are bad for an animal. Often mutant offspring are born dead. Other mutations make the offspring easy prey to predators. These do not survive to maturity and so do not reproduce themselves. Excluding these, let’s take the case of a mutation that is passed on to live offspring. Other animals of the same species are born and reproduce themselves normally.
In lens design, the program makes a systematic variation of all the characteristics. This ensures that no change that might benefit the design is left out. For each variable the program makes a small change. Then, using the merit function, the program measures the global effect of the change, and stores the result. Third, the program returns the variable to its original value. The program repeats this process for every one of the variables in turn.
Step Three
Most of the mutations passed on to offspring seriously reduce the animal’s capabilities. Perhaps only one mutation in a thousand produces a benefit for an animal that helps it survive. Animals with such a mutation may live to maturity and reproduce, while other animals without the mutation may die young. Eventually the mutation will appear in almost all the animals of the species, if the mutation really gives a survival advantage. The diagram in step 2 of the table above simplifies and exaggerates the effect. The individual marked with an “x” has an advantageous mutation that allows it to produce two offspring. Simultaneously it puts other, normal individuals at such a disadvantage that they can only produce one offspring. According to the simplified diagram, in only three new generations the mutation reaches dominance in the population. In nature many generations are necessary.
The automatic lens design program incorporates a mathematical method for calculating a combined variation of all the variables. The method takes into account all the variations that individually reduce the aberrations. If a variation increases the aberrations, the method is programmed to take this variation in the opposite sense. Then the program calculates the combined variation that produces the maximum reduction of aberrations. Each variation contributes in proportion to its design improvement. The program then modifies the design according to the combined variation. This process is much more thorough and efficient than random variation.
Step Four
In both the process of Darwinist evolution and the automatic lens design program there is a return to step one, but with a new initial variety or a new design.
My research colleagues were uncomfortable. They anticipated what I was going to say. I continued with the comparison even though I could see that they were consulting among themselves to discover, if they could, any error in my logic. Using the following comparison table, I asked my colleagues to make an inventory of the advantages of automatic lens design over Darwinist evolution.
Equal Starting Points
In lens design, there is a designer who starts the process, choosing the most appropriate type of lens for the application. But with living species, the initial design must be completely natural, without any designer. The lower types of living organisms, for example, amoebas, can hardly take individual thought about improving themselves. According to Darwinism, before the appearance of humans, there was no mind that could guide the process of Darwinism or give it a purpose or goal. The starting point had to be an extremely unlikely lucky accident that produced the first living organism from complex chemical combinations of atoms. Biology has to start from the basis of chemistry, and chemistry is based on physics. Chemistry does not define very clearly what combinations of atoms and molecules will lead to life, because the number of combinations is almost inexhaustible. Nevertheless, for the sake of argument, we started with the supposed common ancestor of all living things on Earth.
Many Variables
Let’s think for a moment about the number of variables that must be managed. A lens with four elements has the position and curvature of the two surfaces of each element. Right there we have 16 variables. Then the material that fills the space between successive surfaces may vary in its optical characteristics, such as refractive index and dispersion. This gives 14 more variables for the 7 spaces between the 8 surfaces. We now have 30 variables. But the genome of an animal may have 30 000 to 100 000 genes. In short, the number of variables for lens design is large, but the number of variables for a plant or animal is very, very large. This difference favors lens design over evolution, because the program has to explore fewer variables than natural selection does.
Slow Evolution, But Rapid Artificial Design
Let’s now consider the rapidity of adaptation. Darwinists say that the process of biological evolution is very slow. For this reason, there is no hope of observing the appearance of a new species during the lifetime of any scientist. Thus, the process of evolution is not subject to the usual kind of scientific proof, which involves experimentation and finding results that confirm the theory. Darwinists say that we must accept the difficulty of observing even one instance of the appearance and eventual dominance of a favorable mutation.
More than 140 years have passed since the publication of The Origin of Species. Darwinists have seen confirmation of the slowness of change. The little fly that lives on rotten fruit, Drosophila, can reproduce itself in twenty minutes. There have been many attempts to induce high rates of mutation with radioactivity and other pressures on small captive populations of this fly. Of all the mutations observed none has come to be dominant in the world population of fruit flies. In addition, if all the mutations were to appear in one individual fly, the fly would still be of the same species. Thus, even in the case of an insect that reproduces very rapidly, experimenters have never yet observed the origin of a species.
To be generous with the Darwinists, I put the rapidity of adaptation at one step per century. Computers are much faster. At the time I gave the talk they could do thousands of arithmetic operations in a millisecond. (That is very slow by today’s standards.) The example I chose for my colleagues was a computer that could do 1 000 steps per second while designing a lens. Then, to put a generous limit on the process of Darwinism, I allowed 4 000 million years since the crust of the Earth cooled sufficiently to support the existence of life. At one step per century, there was time for 40 million steps. The computer could do the same number of steps in less than 12 hours.
A Comparison that Shows Up a Difficulty
With evolution the next step comes from random mutation. One has to wait a very long time until the appearance of the next mutation necessary for development. But the computer is programmed to consider all of the variations of the lens parameters, and to make a systematic selection of the best combined variation. Once again, this is a great advantage to automatic lens design and a big disadvantage to evolution. Evolution has only a chance of finding a beneficial mutation, but automatic lens design is sure to find the best possible variation.
Both processes have a merit function that indicates the best lens or the best mutation. A naturally selected mutation makes an individual the fittest for survival. The best lens is the one with the smallest aberrations.
But in several other aspects the process of Darwinism has disadvantages relative to automatic lens design. The programmer defines the direction of the next step in automatic lens design, but evolution has no direction. The same holds true for goals. The program’s purpose is to design lenses that focus better, but natural processes have no goal. Direction, purposes, and goals can only exist in an intelligent mind.
And what are the results of the two processes? Darwinists say that their process produces viable species. They claim that the process of Darwinism produced all the present living species of the Earth. In the same way, our programmers supposed that the automatic lens design program would produce better lenses.
In lens design, there is a designer who starts the process, choosing the most appropriate type of lens for the application. But with living species, the initial design must be completely natural, without any designer. The lower types of living organisms, for example, amoebas, can hardly take individual thought about improving themselves. According to Darwinism, before the appearance of humans, there was no mind that could guide the process of Darwinism or give it a purpose or goal. The starting point had to be an extremely unlikely lucky accident that produced the first living organism from complex chemical combinations of atoms. Biology has to start from the basis of chemistry, and chemistry is based on physics. Chemistry does not define very clearly what combinations of atoms and molecules will lead to life, because the number of combinations is almost inexhaustible. Nevertheless, for the sake of argument, we started with the supposed common ancestor of all living things on Earth.
Many Variables
Let’s think for a moment about the number of variables that must be managed. A lens with four elements has the position and curvature of the two surfaces of each element. Right there we have 16 variables. Then the material that fills the space between successive surfaces may vary in its optical characteristics, such as refractive index and dispersion. This gives 14 more variables for the 7 spaces between the 8 surfaces. We now have 30 variables. But the genome of an animal may have 30 000 to 100 000 genes. In short, the number of variables for lens design is large, but the number of variables for a plant or animal is very, very large. This difference favors lens design over evolution, because the program has to explore fewer variables than natural selection does.
Slow Evolution, But Rapid Artificial Design
Let’s now consider the rapidity of adaptation. Darwinists say that the process of biological evolution is very slow. For this reason, there is no hope of observing the appearance of a new species during the lifetime of any scientist. Thus, the process of evolution is not subject to the usual kind of scientific proof, which involves experimentation and finding results that confirm the theory. Darwinists say that we must accept the difficulty of observing even one instance of the appearance and eventual dominance of a favorable mutation.
More than 140 years have passed since the publication of The Origin of Species. Darwinists have seen confirmation of the slowness of change. The little fly that lives on rotten fruit, Drosophila, can reproduce itself in twenty minutes. There have been many attempts to induce high rates of mutation with radioactivity and other pressures on small captive populations of this fly. Of all the mutations observed none has come to be dominant in the world population of fruit flies. In addition, if all the mutations were to appear in one individual fly, the fly would still be of the same species. Thus, even in the case of an insect that reproduces very rapidly, experimenters have never yet observed the origin of a species.
To be generous with the Darwinists, I put the rapidity of adaptation at one step per century. Computers are much faster. At the time I gave the talk they could do thousands of arithmetic operations in a millisecond. (That is very slow by today’s standards.) The example I chose for my colleagues was a computer that could do 1 000 steps per second while designing a lens. Then, to put a generous limit on the process of Darwinism, I allowed 4 000 million years since the crust of the Earth cooled sufficiently to support the existence of life. At one step per century, there was time for 40 million steps. The computer could do the same number of steps in less than 12 hours.
A Comparison that Shows Up a Difficulty
With evolution the next step comes from random mutation. One has to wait a very long time until the appearance of the next mutation necessary for development. But the computer is programmed to consider all of the variations of the lens parameters, and to make a systematic selection of the best combined variation. Once again, this is a great advantage to automatic lens design and a big disadvantage to evolution. Evolution has only a chance of finding a beneficial mutation, but automatic lens design is sure to find the best possible variation.
Both processes have a merit function that indicates the best lens or the best mutation. A naturally selected mutation makes an individual the fittest for survival. The best lens is the one with the smallest aberrations.
But in several other aspects the process of Darwinism has disadvantages relative to automatic lens design. The programmer defines the direction of the next step in automatic lens design, but evolution has no direction. The same holds true for goals. The program’s purpose is to design lenses that focus better, but natural processes have no goal. Direction, purposes, and goals can only exist in an intelligent mind.
And what are the results of the two processes? Darwinists say that their process produces viable species. They claim that the process of Darwinism produced all the present living species of the Earth. In the same way, our programmers supposed that the automatic lens design program would produce better lenses.