Introduce a book about science

The English title of Out of Control: The New Biology of Machines, Social Systems and the Economic World sounds quite bluffing, and the black and yellow cover makes people feel that this is another doomsday book written in the 1990s.

In fact, this is a book full of romantic feelings. It describes that with the gradual abandonment of precise control of machines and the gradual introduction of biotechnology and evolutionary ideas, machines have the ability of self-adaptation, self-learning and self-evolution, in order to better the world. Think of machines as creatures. They can form a community, live together and evolve into better machines.

Humans always control the process of machine evolution, but how to get better evolution and achieve better results, we still have to learn from nature. The first thing I learned was to return to simplicity. "I finally found out that I want to get behaviors that are really similar to life, not trying to create truly complex creatures, but providing extremely rich variation environments for simple creatures. Therefore, the idea of building a machine has evolved from the initial idea of building a "central brain" for the machine to making many simple low-cost small machines and letting them complete simple tasks. For example, in the recent movie Prometheus, when entering the maze, astronauts released more than N small detectors. As long as they emitted infrared waves and did not touch other objects, they could complete the task well. It is not necessary for a machine to control all the small detectors, but they can explore the underground maze map as long as they fly endlessly.

The following is a general distributed control method developed by Brooks Mobile Robot Lab: Do simple things first. Learn to do simple things accurately. Add a new activity level to the results of a simple task. Don't change simple things. Make the new hierarchy work as accurately as a simple hierarchy. Repeat the above steps, and so on.

Biology is "no self", and your consciousness is the result of many chaotic cell activities. At the same time, your body is sending out hundreds of thoughts, such as "I'm hungry, I'm thirsty, I'm writing an article, I'm listening to music, and I think the pinyin of this word is like this". Each function is doing its job, but in the end, the important thoughts at the moment are merged into "I". This is a bit like Buddhism. Buddhism itself has no "I", and the feeling of "I" is the result of consciousness.

After the idea of independent function of machines is established, what form larger machines will get along with each other and what methods should be adopted to promote evolution, we must go to nature to find the answer. In the past, the United States set up a closed ecological circle in the wilderness for the space migration plan. The project lasted for two years, and there were only a few scientists in it. Before the closure, the whole project decided what animals and plants to bring in, where to build a pool, where to grow food, and how to ensure that they can survive the two years of closure. Even the closed circle needs artificial rainfall. As a result, I barely survived and had a lot of external energy to help me. This experiment is quite interesting and has accumulated a lot of ecological experience.

Lenovo's success is closer to biology in stand-alone experience, creating a closed app ecosystem. But more broadly, it can't "communicate and breathe" with non-Apple machines. At this point, although the Android system is chaotic now, there will be more and more hardware based on open source systems in the future, and the machine ecology will be more resilient, vital and diversified. For example, in the future, refrigerators, clocks and vacuum cleaners at home can be equipped with Android system, and the machines are interoperable, which is very beautiful.

After that, the book spent a lot of time talking about examples of co-evolution in the natural and computer worlds, and the second half basically inspired us to re-understand Darwin's theory of evolution. The traditional Darwinian evolution theory "natural selection" puts a bunch of creatures into the environment, and they will produce all kinds of messy variations. Finally, people who adapt to life will survive and creatures will evolve. But the question is: 1) Why do they look like that? The giraffe's neck became longer later, so how did it form at first? All the fish in the ocean finally survived. Why do some cells become this kind of fish and some become that kind? 2) Is directed evolution possible? Creatures don't mutate randomly. They may mutate according to the environment. For example, you will see more people with six fingers, but rarely people with heads on their backs. It's a little disgusting, please forgive me for proving it.

Therefore, people's understanding of evolution is far from enough, which cannot be explained by a simple natural selection. "All evolution is an extension of natural selection. Although there is some truth in this, it hinders our understanding and acceptance of more complicated things. Multiplication is indeed some kind of addition operation, but a brand-new power has emerged from this fast operation. If we only regard multiplication as the repetition of addition, we will never master this power. If you are content with addition, you will never get E=mc2. I believe that life has mathematics. Natural selection may be addition in this kind of mathematics. In order to fully explain the origin of life, complex trends and the emergence of intelligence, we need not only addition, but also rich mathematics, which is composed of various complex functions based on each other. It needs a deeper evolution. Natural selection alone is not enough. If you want to make a difference, you must be involved in the process of becoming more creative and productive. It must have more means than natural selection. 」

Combining biology and computer research on evolution (although scientists are very careful in this regard at present, it is difficult to see fruitful research because they are unwilling to encounter the word "God". ), the author expects to see the following characteristics in artificial evolution with openness and sustainable creativity: * * health-convenient information exchange, so that different evolutionary paths converge; Directional mutation-non-random mutation and direct communication with the environment; Leaping-function clustering, hierarchical structure of control, modularization of components, and self-organization of adaptive processes that change many characteristics at the same time-tends to a specific form (such as four rounds) and makes it a universal standard development process.

This book is far-sighted about the evolution of the network. Although it doesn't involve much, think about it. It was written in 1995 and has successfully predicted the trend in the next 20 years. For example, in cloud data storage, people used to store things on hard disks, but in the future everything will be in the cloud, and people will get access to data.