Research-driven development. Engine wear

Research "Is it harder to find ideas?" (“Is it getting harder to find ideas?”), which was released in September 2017, and then, in an expanded version, in March of this year. The authors, four well-known economists, show in it that ever-increasing research efforts bring less and less economic benefits.

John Van Reenen of the Massachusetts Institute of Technology and Nicholas Bloom, Charles I. Jones and Michael Webb of Stanford University write:

“A large amount of data from a wide variety of industries, products and companies indicates that research spending is increasing significantly while research itself is rapidly declining.”

They give an example Moore's lawnoting that "the number of researchers now required to achieve the famous doubling of computational density every two years is more than eighteen times that required in the early 70s." The authors note similar trends in scientific papers related to agriculture and medicine. More and more research on cancer and other diseases does not lead to more lives saved, but rather the opposite - the relationship between increased costs and increased results is becoming less and less favorable. For example, since 1950, the number of drugs approved by the US Food and Drug Administration (FDA) per billion dollars spent on research has dropped dramatically.

Views of this kind are not new in the Western world. Already in 2009 Benjamin Jones in his work on the growing difficulty of finding innovation, he argued that would-be innovators in a given field now need more education and specialization than before in order to become proficient enough to simply reach limits they could then cross. The number of scientific teams is constantly growing, and at the same time, the number of patents per scientist is decreasing.

Economists are primarily interested in what is called the applied sciences, that is, research activities that contribute to economic growth and prosperity, as well as to improve health and living standards. For this they are criticized, since, according to many experts, science cannot be reduced to such a narrow, utilitarian understanding. The Big Bang theory or the discovery of the Higgs boson does not increase the gross domestic product, but deepens our understanding of the world. Isn't that what science is all about?

Front page study by Stanford and MIT economists

Fusion, i.e. we already said hello to the goose

However, it is difficult to challenge the simple numerical ratios presented by economists. Some have an answer that economics might as well seriously consider. According to many, science has now solved relatively easy problems and is in the process of moving on to more complex ones, such as mind-body problems or the unification of physics.

There are difficult questions here.

At what point, if ever, will we decide that some of the fruits we are trying to achieve are unattainable?

Or, as an economist might say, how much are we willing to spend on solving problems that have proven very difficult to solve?

When, if ever, should we start cutting losses and stopping research?

An example of facing a very difficult issue that at first seemed easy is the history of litigation. development of thermonuclear fusion. The discovery of nuclear fusion in the 30s and the invention of thermonuclear weapons in the 50s led physicists to expect that fusion could be quickly used to generate energy. However, more than seventy years later, we have not progressed much along this path, and despite many promises of peaceful and controlled energy from the fusion in our eye sockets, this is not the case.

If science is pushing research to the point where there is no other way for further progress other than another gigantic financial outlay, then perhaps it's time to stop and consider whether it's worth it. It seems that the physicists who have built a powerful second installation are approaching this situation. The Large Hadron Collider and so far little has come of it... There are no results to support or disprove the big theories. There are suggestions that an even larger accelerator is needed. However, not everyone thinks this is the way to go.

Golden Age of Innovation - Building the Brooklyn Bridge

The Liar's Paradox

Moreover, as stated in the scientific work published in May 2018 by Prof. David Woolpert from the Santa Fe Institute you can prove that they exist fundamental limitations of scientific knowledge.

This proof begins with a mathematical formalization of how an "output device"—say, a scientist armed with a supercomputer, large experimental equipment, etc.—can obtain scientific knowledge about the state of the universe around him. There is a basic mathematical principle that limits the scientific knowledge that can be gained by observing your universe, manipulating it, predicting what will happen next, or drawing conclusions about what happened in the past. Namely, the output device and the knowledge it acquires, subsystems of one universe. This connection limits the functionality of the device. Wolpert proves that there will always be something he cannot predict, something he cannot remember and cannot observe.

“In a sense, this formalism can be seen as an extension of Donald McKay’s claim that the future narrator’s prediction cannot account for the narrator’s learning effect of that prediction,” Woolpert explains at phys.org.

What if we don't require the output device to know everything about its universe, but instead require it to know as much as possible about what can be known? Volpert's mathematical structure shows that two inference devices that have both free will (well-defined) and maximum knowledge of the universe cannot coexist in that universe. There may or may not be such "super-reference devices", but no more than one. Wolpert jokingly calls this result the "principle of monotheism" because while it does not forbid the existence of a deity in our universe, it does forbid the existence of more than one.

Wolpert compares his argument with chalk people paradoxin which Epimenides of Knossos, a Cretan, makes the famous statement: "All Cretans are liars." However, unlike Epimenides' statement, which exposes the problem of systems having the ability to self-reference, Volpert's reasoning also applies to inference devices lacking this ability.

Research by Volpert and his team is carried out in various directions, from cognitive logic to the theory of Turing machines. Santa Fe scientists are trying to create a more diverse probabilistic framework that will allow them to study not only the limits of absolutely correct knowledge, but also what happens when inference devices are not supposed to work with XNUMX% accuracy.

David Wolpert of the Santa Fe Institute

It's not like a hundred years ago

Volpert's considerations, based on mathematical and logical analysis, tell us something about the economics of science. They suggest that the most distant tasks of modern science - cosmological problems, questions about the origin and nature of the universe - should not be the area of ​​​​the greatest financial costs. It is doubtful that satisfactory solutions will be obtained. At best, we will learn new things, which will only increase the number of questions, thereby increasing the area of ​​ignorance. This phenomenon is well known to physicists.

However, as the data presented earlier show, the orientation towards applied science and the practical effects of the acquired knowledge is becoming less and less effective. It is as if the fuel is running out, or the engine of science is worn out from old age, which only two hundred or one hundred years ago so effectively fueled the development of technology, invention, rationalization, production, and finally the entire economy, leads to an increase in the well-being and quality of life of people.

The point is not to wring your hands and tear your clothes over it. However, it's definitely worth considering whether it's time for a major upgrade or even a replacement for this engine.