Why Science Slowed Down

Ben Landau-Taylor has Samo Burja back on The Students to talk about science as rigorous theory, not as a brand. Auto-captions call the guest Samuel Arbesman. He is Samo Burja, of Bismarck Analysis.

Science is theory

In the nineteenth and twentieth centuries, Burja says, fields moved. Evolution, electromagnetism, relativity: the human store of knowledge, and the theories built on it, advanced in many domains at once. There were machines too, but he keeps the distinction clean. Science is theory. Technology is capability. Today he sees stagnation in biology and chemistry, and an arguable stall in physics. The exception people reach for is AI. Even there, the core theory of why the systems work was worked out in the 1990s and the early 2000s. Models grow more capable because compute grows. That is a technological advance, not a scientific one.

Landau-Taylor has heard the other story: we figured most of it out; the remaining fruit is small. That is not Burja’s claim. The low-hanging-fruit story, he says, could have been told in 500 BC, in 500 AD, in 1500, and in 2026. In any of those years you might have said the known world was mapped and the conceptual work was done. The fact that we cannot immediately point to the next break does not mean there is none. We even know some of the gaps. We can see fruit we cannot reach.

The fruit we cannot reach

Sleep is the one that messes Landau-Taylor up. We know you get tired. We know it goes badly if you skip it. There was a book, Why We Sleep, and it was debunked. When Burja’s children ask why we sleep, he gives them the honest answer: we don’t know. We sure are tired, and it is time for bed. Landau-Taylor’s father, a physician, used to tell a medical-school joke. Before you finish, friends bring you their problems and you say I don’t know. After you finish, you get to tell them we don’t know.

Theoretical physicists will assure you they are making immense progress. Ask whether they have exhausted physics and they say no. They point at gravity at very small scales, and at gravity in general. Burja does not think dark matter is real; if it is, it is a large puzzle, missing mass so galaxies do not fly apart. Bigger colliders were meant to carry physics toward the next paradigm. That has not happened, or not yet.

Expense is not the main gate, and ethics is not either. Sleep studies happen. Instruments can matter, but Burja’s best guess is that science happens in the interrelation of theory and observation, and that it has to be the same person. The split between theoreticians and experimentalists professionalizes badly. Unless you are good at theory you do not know how to tweak the sleep study. We have gotten bad at theory as an activity: abstract hypotheses that make quantitative and qualitative sense of what we already know.

Generators and tacit method

The hardest transfer is not textbook answers. It is the generators, the spirit that lets you invent the questions. Great physicists and mathematicians are so often trained by the last generation’s greats. Sometimes someone reinvents a tradition. Ramanujan found a math textbook in India, derived his own approach, then amazed the Western mathematicians. He had discovered a new generator. The stuff is tacit, a felt sense. The modern world is biased against it. Mentorship is not a KPI. Has anyone endowed, at the Institute for Advanced Study, a chair of Einsteinian imagination — thought experiments all day, mental images for new problems, exactly as Einstein described?

Landau-Taylor remembers Terence Tao, staying at his aunt’s, rotating spheres in his mind by rolling on the floor. She walked in. He said he was working. Could that corner of the method be taught? People know Feynman recommended a shower, not as a metaphor for refreshment but because the tactile reset helps you think. The Twitter joke writes itself: the modern paper carries a thousand caveats; the eighteenth-century treatise is some thoughts had in the bath that constitute self-evident truths. One activity defends a fence. The other proclaims. Feynman’s counterexamples — take an abstract property and imagine it fuzzy — read like the more detailed Christian prayer manuals: step-by-step instructions for what to do with the mind.

Notation is one way of writing intuition down. Cartesian coordinates arrive late, in an appendix to the work in which Descartes does I think, therefore I am. He was in bed, sickly, watching a fly, trying to describe its flight. If you are bedridden you may conclude that the mind is what matters.

A faculty of the mind

Science, Burja says, may belong in a class with Buddhist meditation and Christian prayer: three instances of training a faculty of the mind. In the modern university, science looks like a bureaucratic process. There is a bureaucratic process that goes by that name. Peer review as a centralized mailing-out of comments is recent. For most of scientific modernity until the mid-twentieth century the traffic was more like a blog post and a comments section. Einstein mailed special relativity to a friend. The friend said it checked out. Reputation tracked judgment; degrees rode along secondarily. Between 1900 and 2000, PhDs per capita in the United States rose by a factor of thirty. Burja does not think we got thirty times as many people who are as good at the thing.

Judgment resists proceduralization. A peacetime military fills up with careerist generals; war is the forcing function that shows who can fight. The tension is reputation against career. Once science is a bureaucratic career, something is lost.

What counts as a war

What is the scientific equivalent of a war? Landau-Taylor asks. Burja turns to Vannevar Bush and Science, the Endless Frontier. Bush’s taste in people was decent. He ran American technological development in the Second World War, a short war, better at pushing technology than science. Afterward he argued that the centralized structures built to win that war could be repurposed to push technology — and, implicitly, science — forever. A centralized philosopher-king has to be good. Successors lacked Bush’s taste, and power fragmented. DARPA is an exception on the technology side because program directors have discretion. A director with a limited term funded an approach to self-driving cars. NASA, in theory, has a scientific mandate. In practice it is an exploration agency, oriented around missions and ships, not theory. Classified work is another leak. Manhattan-Project science, or whatever is discovered while building a better bomb, stays inert unless it is load-bearing for still-classified technology.

Republic of Letters

There has to be a community ideologically driven to discover the truth and to expose those truths to one another. The scientific revolutions from the sixteenth to the twentieth century happened when esoteric correspondents — they had started in alchemy — began writing to the public in plain language. The Republic of Letters was half internet, half esoteric conspiracy: anyone could write in, and not all truths were for everyone. The people who built the internet and who built AI, Burja says, hold an esoteric philosophy as strange as alchemy. Alchemy proposed to turn lead into gold. The singularitarian worldview proposes to turn man into machine. The earlier internet people were cypherpunks: free man from the state, information wants to be free. You are only good at science if the ideology includes an almost irrational commitment to truth.

Some of the Greek philosophers were not merely predecessors of science. They were scientists. Archimedes calculated the water displacement of a ship with methods that approach calculus. Eratosthenes measured the circumference of the Earth with trigonometry. Joseph Needham, looking at China and the West, found the same hinge. Until about 1700, China was technologically ahead. Then the picture reversed. Needham’s conclusion was quantified empiricism joined to a fully descriptive theory — seen, as far as Burja knows, in modern science and in those Greeks.

Professor Lucio Russo’s The Forgotten Revolution argues that science was discovered around 300 BC, forgotten, and rediscovered. He goes further: the Renaissance had Greek scientific works we have since lost. Burja first thought that had to be false. Correspondences in the Republic of Letters name titles we no longer have. Libraries still burn. A few years ago an important part of the National Library of Brazil burned; last recordings of extinct Amazonian languages went with it. And after a copyright ruling, AI labs have been destroying unique books. They could keep the copies they bought, fight, or refuse and pay the fines. That would be a live-player move. At ten copies, or five, or one, you are destroying the record as a side effect of a legal fight.

Russo was a physicist and mathematician who read Latin and Greek. To evaluate ancient science, the same person has to be scientist and historian. Knowledge lives in heads, not in books alone. The Romans wanted the pop-science Greeks. Science was a minority taste among the Greeks. It still is.

Proof of possibility

Technology can tinker without theory. Science, when you have it, supplies a proof of possibility: theoretically this rocket, theoretically a space elevator on the moon even if the materials are missing. Recent physical technology — biology, materials — looks like tinkering. AI may self-play mathematics. Novel science is harder. The models are trained on human literature about the world. Burja hopes no one gives up the calling because machines can do a version of the job. It remains a calling even then.

Pythagoras, they agree, would not have.

Landau-Taylor thanks him. A great conversation, as always.

“A pleasure, Ben.”