Chapter 43 complex enhancement

complex enhancement "Either way, this vague revelation made us think we had a handle on where this interesting organizational phenomenon takes place," says Farmer. But that's by no means the whole story.For ease of argument, it may be assumed that this particular chaotic fringe realm does exist, but even so, the hypothetical new second law must explain how these systems came to and existed in this realm while at the same time What has been done in the field. This vague revelation could easily lead us to believe that Darwin had already answered the first two questions (as Holland outlined).The idea is that since the system with the most complex and perfect feedback is always able to maintain its sensitivity to the competitive world, then the rigid system can always perform better with a little relaxation , and a chaotic society can always achieve better results with a little control.So, if a system has not yet reached the edge of chaos, you would expect learning and evolution to push it in that direction, and if the system is right on the edge of chaos, you would expect learning and evolution to be able to push it in that direction. Pull it back into place when it tends to go off the rails.In other words, you want learning and evolution functions that can turn the fringes of chaos into stable homes for complex adaptive systems.

The third question is what do such systems do when they reach the edge of chaos.This is a more subtle issue.In the space of all possible dynamical behaviors, the edge of chaos is like an infinitely thin membrane, a special domain that produces complex behaviors that separate order from chaos.Just as the surface of sea water is only separated from the air by the thickness of a water molecule, the edge of chaos is also like the surface of the ocean, which is boundless and boundless, and the actors can use it in endless ways. Show its complexity and adaptability.Indeed, Holland may not be using the metaphors above when he speaks of "perpetual novelty," of the infinite space of possibilities explored by adaptive agents, but what he means is precisely that if adaptive agents On the thin film on the edge of the boundless chaos.

So how does the new second law explain this?Of course, it would involve building bricks, internal models, co-evolution, and whatever adaptive mechanisms Holland and others have studied.But Farmer suspects that at its core it will be more about pointing the way than describing the mechanism: the deceptively simple fact that evolution often leads to things getting more complex, more refined, and more structured.Farmer said: "Clouds have more structure than the original miasma after the big bang, and the primordial soup has more structure than the clouds." And we humans have more structure than the primordial soup.It follows from this fact that the modern economy is much more structured than the Mesopotamian city-states, just as modern technology is far more advanced than that of Roman times.The function of learning and evolution seems not only to pull the economic agent slowly, intermittently, but inexorably towards the edge of chaos, but also to make the agent develop along the edge of chaos in a more and more complex direction.Why is this?

"It's a tricky question," Farmer said. "We have a hard time articulating the concept of 'progress' in biology." What do we mean when we say that one organism is more advanced than another?Take cockroaches as an example, they have existed for millions of years longer than humans. As cockroaches, they have evolved to a very advanced level.Are we humans higher than them, or just different from them?Were our mammalian ancestors really superior to the ferocious Tyrannosaurus rex 65 million years ago?Is it just because I was lucky enough to escape the catastrophe of the falling comet?Without an objective definition of the concept of "fittest," Farmer says, "survival of the fittest" becomes a tautology of "survival of the survivors."

"But I also don't believe in nihilism, the notion that nothing is better than anything else. It's a silly idea that evolution didn't make us, but if you take a step back and look at the fullness of evolution in a broader light In the process, you will see the general trend of continuous refinement, complexity and functional enhancement. Compared with the difference between the earliest organisms and the most recent organisms, the difference between the Model T and the Ferrari The difference is almost insignificant. As puzzling as it is, the design of evolution does, on the whole, tend toward continuous improvement in 'quality'. This is the most fascinating and profound clue to a comprehensive explanation of the phenomena of life. "

One of his favorite examples is the evolutionary phenomenon in the autocatalytic group model he created with Packard and Kaufman.The beauty of autocatalysis is that you can follow the emergence process from the ground up.Concentrations of a small number of chemicals spontaneously and substantially exceed their average concentrations because of collective actions that catalyze and shape each other.This means that the autocatalytic group as a whole has transformed into a new emergent personality that emerges from its balanced background, which explains the origin of life. "If we knew how to carry out this process in a real chemical experiment, we could have something in balance between living and non-living things. These autocatalytic entities do not have a genetic code. But they can The primordial form is self-sustaining, self-expanding, not as perfect as a seed, but a hundred times better than a heap of rocks."

Of course, in the original computer model, there was no such evolution of the autocatalytic group, because there was no interaction with the external environment in the original model.The model assumes that everything happens in a well-stirred chemical solvent, so the autocatalytic group is stable as soon as it emerges.But in the real world four billion million years ago, these vaguely defined autocatalytic monomers lived in all sorts of ups and downs.What happens in this situation?Farmer and graduate student Rick Bagley tried to understand this by subjecting the model to an unsteady supply of "food."The so-called "food" is a string of tiny molecules that are provided as raw materials to the autocatalytic group. "The most amazing thing is that some autocatalytic groups are like pandas that only eat bamboo, and they can't survive if the food supply is changed. Others are like omnivores, and they have many different methods of metabolism, which allow them to adapt. Changes in food. So, you change the food supply and they're basically unaffected." Such robust catalytic groups may be species that survive on Earth.

More recently, Farmer says, he and Bagley, along with Rosalamos' postdoc Walter Fontana, have refined the autocatalytic model so that it can produce the occasional spontaneous reaction that does exist. in real chemical systems.This spontaneous reaction leads to the splitting of many autocatalytic groups.But split autocatalytic groups paved the way for an evolutionary leap. "The split triggers all kinds of new things to come in. Some kind of variation will be enlarged, and then enter a stable state again until the next big collapse. We observed a series of metabolism and mutual replacement of autocatalytic groups."

Maybe that's a clue. "It would be interesting if we could include some kind of feedback loop (feedback loop for stability) in emergent structures that hasn't been there before in our elucidation of the concept of 'progress'. The point is, it's a series of evolutionary Events frame the matter of Spencer's conception of the universe, in which each emergence paves the way for the next." "I'm actually troubled talking about all of this," Farmer said. "There's a real language barrier here. Everyone's busy trying to define concepts like 'complexity' and 'emergent computational propensity,' and I'm but can only give you vague images in language that has not yet been clearly defined in mathematical terms, it seems like we are in a period before thermodynamics, we are now in the 1920s, when people knew that There was something called 'heat', but then people called it only in language that later sounded very absurd." In fact, people were not even sure what heat was at that time, let alone the mechanism of heat motion.At that time, the most reputable scientists were convinced that a red-hot poker was densely packed with a weightless, shapeless fluid called "calories," which flowed inexorably from the poker to the outer Something cold and lower in calories.Only a few people think that heat represents some kind of microscopic motion of the poker atoms. (This minority was right.) At the time, no one seemed to have imagined that complex and disorderly things like steam engines, chemical reactions, and batteries were all governed by simple, general laws.It wasn't until 1824 that a young French engineer named Sadi Carnot published his first paper, which stated what became known as the second law of thermodynamics: that is, heat Does not automatically flow from cold to hot. (Carnot, writing a best-selling book for his colleagues, quite rightly pointed out that this simple and commonplace fact places many limits on the efficiency of the steam engine, not to mention the internal combustion engine, the turbine of a power plant, or any The limitations of a working machine. A statistical interpretation of the second law, namely, that atoms are constantly trying to make themselves random, did not appear until seventy years later.)

Likewise, it wasn't until the 1940s that the British brewer and amateur scientist James Joule laid the experimental foundation for the first law of thermodynamics.This first law of thermodynamics is known as the law of the immortality of energy: energy can be converted from one form to another, including thermal, mechanical, chemical, and electrical forms, but energy is forever It cannot be created or destroyed.It was not until the 1950s that scientists explained these two laws in an accurate mathematical form. "We're quietly moving toward deciphering the phenomenon of self-organization," Farmer said. "Understanding organization is far more difficult than understanding chaos. We still haven't discovered the key concepts, at least not in the form of clear, quantitative analysis of self-organization." The concept of organization. We need to articulate this concept as clearly as we did the hydrogen atom, be able to take it apart and give a perfect and clear description of its mechanism. But we are not yet able to do that. We have a There is only fragmented understanding, and the understanding of each of its parts is isolated. For example, we now understand a lot about chaos and fractals. Chaos theory tells us how simple systems composed of simple parts can generate extremely complex Behavioral. We also know a lot about gene regulation in Drosophila. We also know a little about how self-organization occurs in the brain under a few specific circumstances. In the field of artificial life, we have created the 'toy universe' ' panorama. The behavior of these models slightly reflects the real situation in natural systems. But we can simulate them completely, make changes to them at will, and know exactly what caused them to behave now. We hope that we will eventually be able to regress The next step is to integrate all of this into a complete theory of evolution and self-organization."

"It's not a field for people who like to work on well-defined problems," Farmer said. "But what's exciting is that this field has not yet become rigid. Things are still developing, and I haven't seen anyone who has found a well-defined problem." avenues of problem-solving. But we've found a lot of first clues, a lot of little toy systems and vague concepts. So I predict that within the next 20 or 30 years, we'll have a real theory."