2.9 Simulated adaptation
Suppose, now, that evolution is climbing a mountain crest in the landscape (phenotypic or mental is irrelevant here) and that the lines in figure 2.4 are part of a contour line enclosing a region of acceptability in the landscape (see chapter 6 normal adaptation definition 6.5). At the start the cluster of red points represents a very homogeneous population with small variances in the parameters. Evidently, even small environmental changes in the landscape, may cause the process to become extinct. Thus the increased disorder (and diversity) is a very good life insurance for unforeseen changes in the environment. This holds even if the red process were centered at the head of the arrow. The argument is also in contrast to what racists think about disorder, diversity and survival.
Figure 2.4.
After a sufficiently large number of generations, the increase in mean fitness and disorder (adaptation) may result in the green cluster. Actually, it is only the disorder that has increased; the probability (=.65) of becoming a parent to new individuals in the population, is the same for both red and green population.
Figure 2.5 shows how the process has climbed a fictitious landscape in two dimensions (peaks are blue, hollows are red). Green points have been selected, while the red points are not. The process is started at A. The accumulation of green points shows that the process has shown an interest for several peaks at B, C, D and E. Finally the higher peak at F has been chosen. The process has been slightly forced, i. e. the bar has been elevated bit by bit (a correspondence in nature is the arms races between different species).
The highest peak at G was never found because the bar was too high and the hollow between F and G was too wide and deep. Also notice the spread of points along the ridge between D and F.
Figure 2.6 shows a slightly forced climbing on a banana shaped ridge starting at A. In the vicinity of the highest peak at B, the adaptation of the normal distribution is even more extreme. In the final stage the level of the bar is so high that only a narrow strip is available for the green points. But most of the red points in the vicinity are not so bad, after all.
Figure 2.5.
Figure 2.6.
2.10 Random search in many dimensions
Computer simulations of random processes have shown that very difficult problems may be solved in an efficient way. They are also able to compete with deterministic methods that with the aid of a sort of compass will find a way up the hill, particularly if the degree of difficulty exceeds a certain level. There is a higher risk that a deterministic method will get stuck at a lower peak, when starting in its region of attraction. That is to say that deterministic methods in a sense lack imagination and creativity.
In contrast to a one- or two-dimensional process, more dimensions will make the process more complex, but it is still possible to simulate the process on a computer.
In this context there exists a tragicomic example. In 1972 such a random algorithm was tested on a complex technical system. The model of the system included 450 components (each of which had a parameter value) of which 130 were adjustable. Some early tests seemed promising. But then a joker said: “In this project nothing should be left to chance”. This, of course, was a good intention as long as all the stops were pulled out. But there was also a hidden meaning: The algorithm worked at random, and it should therefore not be used in this project.
After one year of hard work with deterministic methods - available at that time - a passable system was found, but because parameter values of components vary in the manufacturing process, only 5% of the manufactured systems were able to meet all requirements according to the specifications.
Now those responsible for the project got nervous and caught at a straw; the random algorithm. After two months with the random algorithm, 95% of the systems were acceptable, and the system could now be manufactured and sold at a good profit. The example showed that random algorithms may create an enormous amount of information that is hardly available by other means.
Unfortunately, this example was never published, but Kjellström & Taxén, 1981, showed an example with 76 parameters, which is also a large number in this context.
In addition, the natural processes are very superior to our artificial ones in the sense that they may test millions of individuals or signal patterns in parallel. For instance, suppose that evolution replaces a generation of a certain species with one million individuals with a new generation in one year. For one of our sequentially working computers (testing one individual at a time) the same operation will take one million years. Even though this does not mean that the natural process is one million times more efficient than our artificial process, the tremendous superiority of the natural evolution consists of an ability to work in parallel with a large number of individuals. The same principle holds in the brain, even if tests of signals in control of a certain muscle can hardly be carried out in parallel.
2.11 Summary
The second law of thermodynamics is primarily known from physics and chemistry. One of its most important consequences is the equalization of the atmosphere around the planet. In other words, it causes the atmosphere to expand over a diversity of regions on the surface of the earth.
The law states that the disorder (entropy) always increases in all isolated systems, but I prefer the somewhat softened version that a system occupies its possible states in proportion to their probability of occurrence and with respect to the circumstances (4.2). The most characteristic outcome of this is a maximization of mean information, disorder and diversity (see chapters 4 and 5). As soon as the first living organism or copying machine had been created, it began to copy itself into billions of individual organisms. Because the copying process was subject to random errors, the gene pool of the organisms began to expand over a phenotypic landscape. Today we can see the result; an enormous biological diversity both within our own species and also in the number of different species.
Phenotypic evolution also has an analogue as an evolution of signal patterns in our own central nervous system. This process is expanding over the set of possible signal patterns in the brain, the mental landscape. The result is an enormous mental diversity of ideas or memes (Dawkins). As examples we may mention different religions, the commandments of God, the Sermon on the Mount, scientific discoveries, different works in literature, art or music etcetera. Analogous to poisonous spiders or snakes, we also have dangerous memes as for instance terror, rifles, atomic bombs or biological weapons.
To day youngsters are bombarded with computer games and bad video films and in this way the second law may contribute to the disorder and bullying in certain modern schools, but let us never forget about the creative side of disorder and strive to a balance between slavish discipline and complete chaos (see also figure 4.1).