top of page
Writer's pictureBedRock

Genes, Survival Machines, and the Evolution of Strategies - Essays on Investment Thoughts

Updated: Apr 8

Since I referenced w to the power of n in a book review of "The Selfish Gene" last time, I decided to take out the book again and read it once more. Perhaps when I first read it, I found the author too verbose and gave up on reading it, but this time I persevered and found the book to be very rewarding. Because the content is so extensive and based on fundamental ideas, I plan to share some interesting topics from the book in several opportunities.


Regarding the topic of strategy stability, I have described it in another article I wrote, "Scientific Exploration of Strategy Selection - Investment Reflections (49)", which you can check out if you're interested. Here, I want to explore the evolution and stability of strategies from another perspective driven by the selfish gene.


Genes and the Body (Survival Machine)

For some important background information:

  • Genes can be understood as algorithms that aim to maximize returns. In simple terms, because natural resources are limited, genetic algorithms that are more effective at replicating themselves in complex environments gradually occupy a more dominant position in the environment and squeeze out weaker genes.

  • Mutation: Mutation arises from errors in the natural replication process of genes, which provides the possibility for the algorithm to evolve continuously. From the perspective of natural selection, beneficial mutations (that maximize long-term returns) will be preserved, while unfavorable mutations will be eliminated, thereby achieving optimized adaptation to nature.

  • The body, whether animal or plant, is a survival tool selected by various genes in the process of natural evolution, also known as a "survival machine". The programming of this machine is designed to accomplish anything that is most beneficial for all the genes as a whole.

  • For a survival machine, another survival machine constitutes part of the environment, which is a roadblock to its genetic goals (maximizing returns), or a part that can be utilized. However, unlike non-living matter such as rocks and rivers, these other survival machines also represent other genetic goals. Therefore, they will retaliate against our reactions, resulting in complex game relationships.


The survival machine's behavior has a prominent feature, which is its obvious purposiveness.

The basic principle involved in this is what we call negative feedback. This works like a purposeful machine equipped with a measuring device that can detect the difference between the current state of things and the "desired" state, and the machine can operate faster as the difference becomes larger. Thus, the machine can automatically reduce the difference, and the principle of negative feedback operates in this process. When the "desired" state is achieved, the machine automatically stops.

Genes must perform tasks similar to making predictions about the future. In a complex world, predicting the future is risky. Every decision made by the survival machine is a gamble, and genes are responsible for programming the brain in advance so that decisions made by the brain are mostly positive. In the evolutionary casino, chips are survival. We believe that if the genes of those animals construct sensitive brains, making them often winners in bets, then, as a direct consequence, the likelihood of these animals surviving is greater, and these genes will be inherited.

Of course, in unpredictable environments, how genes predict the future is a problem. One solution is to give the survival machine the ability to learn in advance. This learning ability is formed through the trial-and-feedback mechanism of learning strategies (regardless of whether the Pavlov effect is formed or not). This mechanism is widely used in AI, such as adding a small random number to the decision-making program, recording all future results, and then slightly increasing the weight of favorable factors while reducing the weight of harmful factors to dynamically adjust the strategy.


Another interesting method of predicting the future is simulation. You build a model in your brain that the brain can use to predict what might happen. Survival machines that can simulate future events are much smarter than survival machines that only accumulate experience based on actual trials and errors. Actual trials are time-consuming and laborious, and obvious errors often have fatal consequences, while simulation is safe and fast. And now, using computers for simulation is a major advance in modeling ability.

The evolution of modeling ability seems to have ultimately led to the emergence of subjective consciousness. The emergence of consciousness may be due to the fact that the brain's simulation of the world has reached such a level of perfection that it includes its own simulation.

Of course, no matter what philosophical issues consciousness raises, its emergence ultimately frees the survival machine from the control of its masters, the genes, and turns it into a decision-maker with execution ability. The brain is not only responsible for managing the survival machine's daily affairs, but it also has the ability to predict the future and make corresponding arrangements. It even has the ability to refuse to obey the commands of genes.

Originally, the behavior of animals, whether altruistic or selfish, was under the control of genes. This control, though indirect, was still very powerful. Genes exert their fundamental influence on behavior by governing the construction of survival machines and their nervous systems. Genes are the main strategists, while the brain is the executor. However, as the brain becomes increasingly sophisticated, it has actually taken over more and more decision-making functions and uses techniques such as learning and simulation in the decision-making process.


The Origin of Game Relationships: Replication Algorithm + Limited Resources


The replication factors arise from accidental collisions.

As mentioned earlier, genes can be understood as an algorithm that maximizes profits. In other words, if a possible self-replicating replication factor appears in natural evolution, the numbers of this factor in the world will show exponential growth, leading to a significant advantage relative to non-replicating factors. The production of genes, replication factors, and life itself is essentially an effect of entropy reduction, a reverse process from disorder to order.

However, resources are always limited.

It is easy to understand that the replication process of replication factors requires effective combination with other substances (resources) in the outside world (of course, the replication process cannot be created out of thin air). This process of utilizing external resources will encounter bottlenecks as the number of replication factors expands exponentially. The Origin of Game Relationships

It is this contradiction between replication factors and limited resources, as well as the contradiction between different replication factors for resources, that gives rise to a variety of bizarre game relationships: competition, exploitation, cooperation, deception, and so on.

Game relationships are numerous and complex, including those within the same species, such as mates, offspring, direct competitors for essential resources (food, mates), as well as those between different species, such as food, mutual use, symbiosis, and so on. However, when to adopt what strategy is highly complex, related to one's own state and environmental judgment, such as when male elephant seals fight for mates, whether to do so immediately or to grow larger and stronger first.


Evolutionarily Stable Strategy

ESS is a theory developed from game theory to explain how different strategy choices in natural evolution will ultimately evolve into a very stable state, which is very profound. This state is so stable that if a member's strategy deviates from this state, it will be punished by the natural evolutionary process.


The book provides an example of two ways of competing for resources in a group, one called "dove" that uses eye contact to intimidate the other party. If one side retreats, it will receive 50 points for food, but eye contact incurs a time cost of -10, resulting in a total of 40 points. The losing side will also consume energy and time and can only get -10 points. Therefore, if a group uses only dove strategy, because the victory and defeat are zero-sum games, one side will win at the cost of the other losing, so on average, only one side can get 15 points.

Suppose another competition strategy called "hawk" is produced due to mutation, which uses direct fighting to compete. Since the entire group is initially made up of doves, the mutated hawk side easily wins the competition, directly scoring 50 points. As time passes, the hawk gene will develop significantly in the group due to its enormous success.

However, as the proportion of hawk genes in the group increases, the probability of two hawks meeting increases significantly. If the losing side is seriously injured in the intense confrontation, its benefit will be -100. Therefore, it can be seen that the average score of two hawks is only -25. Although doves are easily defeated when facing hawks, they do not participate in the competition and therefore do not get hurt. Their score is 0 (and they do not waste time with eye contact), which is higher than the result of two hawks meeting.

Therefore, when there are too many hawk genes in the group, the genes that use dove strategies will be relatively more successful, which will increase their proportion in the group.

Thus, it can be seen that the proportion of hawk and dove genes in the group will tend towards a ratio, and a higher ratio will lead to too many hawks and a decrease in net income, while a lower ratio will have the opposite effect. After simple calculations, the proportion of hawk and dove genes in the group will reach a stable state at 7:5, and there will be strong regression forces both above and below this ratio. This is the Evolutionarily Stable Strategy (ESS).

After the group reaches this stable state, the average income is calculated to be 6.25.


Evolutionarily Stable Strategy (ESS) embodies some important characteristics:

  • ESS is not equivalent to the optimal strategy, but it is the most stable. It is easy to know that the optimal strategy should be the situation where everyone uses the dove strategy, and the average income is 15, which is far higher than 6.25. However, this strategy is very fragile in natural evolution for the hawk strategy produced by mutation, and cannot form a stable state. Therefore, although natural evolution forces both dove and hawk genes to strive for optimization, the development direction is based on the most stable state.

  • Of course, when frequent mutations occur, the stable state ESS needs to be constantly adjusted. When a new continent was discovered, European colonizers introduced many animals and plants all at once, which quickly introduced a large number of mutation elements into the existing ecosystem. Technology, as a driving force for innovation, introduces a large number of mutation elements to the existing social organization and also pushes the stable ESS of the original structure to require constant adjustment. Therefore, when analyzing patterns, we must pay special attention to the formation of stable states and the impact of the driving force of change.

  • The core of the Evolutionary Stable Strategy (ESS) is not about being optimal, but about being immune to conspiracy theories and being the most stable. For example, in some interaction strategies, tit-for-tat may be the most stable strategy, as discussed in "Scientific Exploration of Strategic Choice - Investment Thoughts Essay (49)," and in subsequent evolution, new competitive strategies may emerge, such as retaliatory strategy, where if the opponent is a dove, then you will be a dove, but if the opponent is a hawk, then you will be a hawk. This strategy is obviously going to become the most powerful component.


1 view0 comments

Comments


bottom of page