There is a tendency to contrast the artificial and technological world of algorithms with the natural and biological world of life. And yet, nowadays, algorithms have become unavoidable for studying the complexity of life. Hence the growing importance of computer tools, exacerbated by the proliferation of data: control, sequencing of genomes . However, digital science is only a tool for biology. The very nature of life, it is often said, has nothing to do with algorithms.
Well, if! For several decades now, biology and algorithmics have come together regularly. Better still, it is often by studying the life with an algorithmic angle, or by taking inspiration from the living to think the algorithms, that the most beautiful discoveries in these two disciplines took place.
To understand this symbiosis, let’s stop for a moment on the notion of algorithm. Intuitively, we tend to associate this notion with computing machines. Yet in the ninth century, when Al-Khwarizmi introduced the notion of algorithms, calculators did not exist. What was Al-Khwarizmi thinking about, the genius whose name has become the word “algorithm”?
Soviet stamp of 4 kopecks bearing the effigy of Al-Khwarizmi, issued on September 6, 1989 on the occasion of its 1200th anniversary. А. Адашев / Wikipedia
Intriguingly, Al-Khwarizmi’s algorithms were primarily tools designed for a wide audience. Al-Khwarizmi wanted to make every citizen able to solve basic mathematical problems: for example, to determine the weight of an apple, knowing that three apples weigh as much as an apple plus 200 grams. The genius of Al-Khwarizmi was to realize that solving these problems does not actually require … No genius. It was enough to listen and apply “mechanically” the instructions of Al-Khwarizmi to get there.
An algorithm is “stupid and mean”…
This is precisely an algorithm. An algorithm is a sequence of instructions that you just have to follow stupidly and nastyly. Surprisingly, however, once the training suite is followed, extremely abstract and complex problems can be solved!
So, I could give you a list of instructions to follow to solve an equation like x3 + 5x = 1. I’m not going to do it here, because listing these instructions would be rather boring. However, armed only with patience and attention, following my instructions step by step, you would eventually solve this problem. And if you had solved this problem in Italy in the sixteenth century, you would have been glorified and worshiped as one of the greatest scientists of the time.
An algorithm is therefore simply an instruction list to execute. But there is magic: very brief training lists can solve very difficult problems. Better still, if designing powerful algorithms for important problems is a thorough research work, executing the instructions of an algorithm requires no genius. This can be done mechanically. So mechanically that even machines could do it!
This observation seems simplistic. Yet it took centuries for mathematicians to realize it. In particular, in 1936, the great Alan Turing finally formalized the mechanization of the execution of the algorithms. This formalization became the Turing machine .
Over the course of the twentieth century, the new technology industry then transformed Turing’s purely theoretical machine into a panoply of gadgets based on Turing’s ideas. These gadgets are computers, servers, tablets or even phones. Their powers and storage capacities have reached phenomenal technological levels. Nevertheless, all these machines finally do only stupidly and mechanically perform the instructions given to them, governed by algorithms.
What is life?
Now that we have reviewed the algorithms, let’s move on to the other major concept of this article: the living. If we intuitively believe we know better what the living is, it turns out that this concept is actually considerably more complex and subtle than that of algorithms. For two millennia, philosophers have proposed diverse and varied definitions and conceptions of life, and few consensuses have emerged.
Even worse, even modern biologists are very much bothered if they are asked to define the living. Some will discuss the key molecules of life, such as DNA and RNA. However, the staggering information storage and stability properties of these molecules could make them a central component of the digital technologies of the future. And it will be strange to say that these technologies, because they have the same chemical support as the living, are also alive.
Puzzle of the living. qimono / pixabay
On the other hand, many biologists will insist more on certain characteristic properties of the living. These properties typically include heredity and variability. Again, however, it appears that many entities, such as computer viruses, meet these criteria but do not appear to be living.
“If people do not believe that mathematics is simple, it’s just that they do not realize how complicated life is,” said mathematician John von Neumann . Von Neumann is, along with Turing, one of the fathers of computer science. Like Turing, von Neumann understood the importance of algorithms and their omnipresence, including in the living world.
Thus, even before the discovery of the structure of DNA, von Neumann wondered whether a purely mechanical structure could acquire all the characteristic properties of the living. To answer this question, the researcher imagined a virtual simulation, called a cellular automaton . In this simulation, he introduced a structure composed of a kind of source code. Surprisingly, by obeying purely mechanical rules, the source code built a machine, which then made a copy of the source code. Better yet mutations could take place at the time of this copy.
Unbelievable! Von Neumann had just built a purely mechanical virtual entity that possessed all the characteristic properties of the living. Better still, we know today that von Neumann’s structure rests on the same mechanism as the living one, since the source code actually behaves like a strand of DNA. In seeking to determine a living algorithm, von Neumann had discovered the fundamental algorithm of the living!
The “Game of Life” is a cellular automaton.
The quest for the algorithms of life
This discovery, and so many others, especially those of the physicist Erwin Schrödinger , drew the attention of biologists to the mechanisms of life rather than an exhaustive description of the components of life. “Life is a process that can be abstracted from the media,” said von Neumann.
To the chagrin of some biologists, the algorithmic approach of von Neumann’s biology invites us to ignore many details in the workings of the living, to focus on what these details calculate. This simplistic approach may seem outrageous. Yet this is exactly the way theorists theorize when developing algorithms that will be executed by machines. Fortunately for the theorists, the details of the machines are not necessary to understand.
Moreover, the details of modern machines, from your phone to Google’s data centers, are so complex today that hardly anyone understands all the technical details. Certainly not me … Despite this, it is possible to understand a good part of the world of the modern digital by studying only the algorithms of the new technologies.
My bet, which seems to be that of a growing number of researchers, is that it is also possible to understand a good part of the biological world by studying only the algorithms of life.
And if, in this article, I have only given you the example of von Neumann’s artificial life, I give you an appointment at TEDxSaclay to discover the examples of Turing’s morphogenesis and Solomonoff’s learning .