The Knowledge Machine: How Irrationality Created Modern Science

by Michael Strevens

Further, the rule does not reside where the methodists have expected to find it, in scientists’ heads. It does not tell scientists what to think privately; it merely regulates how they argue publicly. It is not a method of reasoning but a kind of speech code, a set of ground rules for debate, compelling scientists to conduct all disputes with reference to empirical evidence alone.

science is a machine for motivating disputatious humans to carry out tedious measurements and perform costly and time-consuming experiments that they would otherwise not care to undertake.

The apparently unassuming code of public behavior, the evidence-only constraint on scientific argument that constitutes all the method science needs to set humanity marching inexorably toward truth, deserves a grand name; I call it the iron rule of explanation. Much of The Knowledge Machine is given over to understanding where the iron rule came from, what it amounts to, and by what means it leads science toward enlightenment. That will be my attempt to settle the Great Method Debate.

Yet these same arguments and explanations show how peculiar and often inhuman in its workings the knowledge machine can be. It gets the job done not in spite of but in virtue of its proprietary blend of inarticulacy, closed-mindedness, and systematic irrationality. No wonder humanity was so reluctant for so long to pull the lever that brought it buzzing and spluttering to life.

The job of scientists is to go through the list of all possible theories and to eliminate as many as possible, or, as Popper said, to “falsify” them.

I don't think this is Popper's premise. His point isn't "go around falsifying as a hobby." It's that if you care about truth, you need to check the possible theories, and that checking has to be adversarial and have some real possibility of disproving it.

Hans Reichenbach, a professor of philosophy in Berlin in the early 1930s. Like Popper, Reichenbach escaped to the English-speaking world as totalitarianism engulfed his Germanic homeland. Reichenbach had not thought much about Hume’s worry that the future may fail to resemble the past until 1933. In that year, the Nazis burned the Reichstag, took control of the University of Berlin, and expelled many of its Jewish professors and staff, Reichenbach included. “Then,” Reichenbach is said to have observed, “I understood at last the problem of induction.”

Big oof.

The same is true for scientific investigation on a more modest scale. In a typical physics experiment, it may take years simply to get the apparatus to function properly; in cognitive psychology or the life sciences, it may take years to run pilot studies and to rehearse experimental designs seeking something that will deliver a significant outcome.

Kuhn’s motivator is more subtle and a little sinister. Individual Kuhnian scientists are not critics at all; they accept the prevailing paradigm with barely a contrary thought. But in their enthusiasm to squeeze every last drop of predictive power from that paradigm, they crush the life out of it.

I think this is not right. The key point for Kuhn is the puzzle solving. It's not that you crush the life out of the paradigm, it's that you use it as much as you can and then at some point you get stuck.

Eddington could explain the blurriness of the astrographic photos, but he gave no reason to think that they would systematically err so as to give Newtonian rather than Einsteinian values for light’s degree of gravitational bending. Further, the crisp photos from the 4-inch telescope gave a value for gravitational bending that was considerably greater than that predicted by Einstein, to a degree that they could be considered to support Einstein’s theory only if that telescope, too, was assumed to be systematically biased. Eddington appeared to be engaged in some rather special pleading, then: he assumed systematic errors in one direction for one of the Brazilian telescopes and in the other direction for the other telescope, so as to reach the conclusion that the results they delivered were quite consistent with Einstein’s theory of relativity.

Researchers funded by Coca-Cola, PepsiCo, and other soda manufacturers have been five times more likely than others to find that there is no connection between drinking sugar-sweetened soda and obesity. Those funded by cigarette companies have been seven times more likely than others to find that secondhand smoke has no deleterious effect on health. And whereas non-industry-funded investigators of the efficacy of new drugs may find that the drugs do what they are supposed to do in about 80 percent of studies, investigators funded by the drugs’ creators find a positive result almost 100 percent of the time. It looks as though something is guiding the science above and beyond cold, hard facts, something that closely shadows cold, hard cash.

The fate of the theory teetered: in some places drift was looked on with favor, in others with doubt verging on incredulity. In 1943 it was attacked, however, by the eminent evolutionary biologist George Gaylord Simpson, and it sank under the weight of his reputation.

“The natural world has a small or non-existent role in the construction of scientific knowledge,” writes the sociologist of science Harry Collins; similarly, the sociologist Stanley Aronowitz says, “Science legitimates itself by linking its discoveries with power, a connection which determines (not merely influences) what counts as reliable knowledge.” The facts are, in short, pawns in a game in which the “strongest team decrees what counts as truth.”

This seems too silly. Somehow scientific engineering works.

With these numbers in hand, Kelvin published his calculations in 1863 using a well established theory of cooling to show—by the time he gave his final estimate in 1897—that the earth could be no more than 20 to 40 million years old and “probably much nearer 20 than 40.”

Interestingly this is already giving up on young-earth creationism.

Science is driven onward by arguments between people who have made up their minds and want to convert or at least to confute their rivals. Opinion that runs hot-blooded ahead of established fact is the life force of scientific inquiry.

A rule that strives to lay down the law about the significance of scientific evidence, then, must also lay down the law about the likelihood of all relevant auxiliary assumptions, in the same way that a procedure for determining chain strength must estimate the strength of every link. The rule’s judgments can be objectively valid only if its estimates of the auxiliary’s likelihoods are objectively valid. An objective rule for weighing any and every piece of evidence is therefore possible only if there is an objective fact of the matter about the likelihood of each relevant auxiliary assumption, given the available evidence.

The bayesians think they can do this.

On one side of the debate is caloric theory, created and developed by French scientists from the 1780s through the 1820s, according to which heat is a kind of stuff. This “caloric fluid” flows steadily from warmer to cooler substances, causing them to become hotter and—newly engorged with fluid—to expand. By 1830, caloric theory has achieved a series of spectacular successes: the accurate calculation of the speed of sound through air, the delineation of a mathematical formula capturing precisely the rate with which heat flows through a material such as a metal rod, and an illuminating theory of the efficiency of steam engines and other heat-driven motors.

Montague and Capulet disagree as to which theory best explains the evidence. And they disagree on how particular pieces of evidence—most notably, heat’s ability to travel through a vacuum—bear on each other’s theories. Is this the version of Romeo and Juliet in which the lovers survive, only to have family squabbles lead slowly but inexorably to irreconcilable differences?

The unending script, the code of conduct for scientific argument according to which Montague and Capulet may continue their debate indefinitely, is provided by the methodological precept that I call the iron rule of explanation. What the rule says is simple enough: it directs scientists to resolve their differences of opinion by conducting empirical tests rather than by shouting or fighting or philosophizing or moralizing or marrying or calling on a higher power. That is all; it makes no attempt to interpret the evidence, to decide winners and losers.

perform an experiment or measurement, one of whose possible outcomes can be explained by one hypothesis (and accompanying cohort) but not the other.

It is in general very hard to do this because you don't know the membership of the cohort. The Quine-Duhem problem is real.

the rule imposes a standard for what counts as empirical testing that never changes, questions of legitimacy are settled once and for all.

There are no duels or divorces in science, no schisms to parallel the historical schisms in religion, politics, and philosophy, where opposing camps stop talking or worse. Always there is something that even the most bitter adversaries can agree to do next: another test.

what about interpretations of QM?

The English scholastic philosopher Robert Grosseteste developed the notion of a controlled experiment in the high Middle Ages, centuries before Bacon.

In modern science, the notion of explanation is free of philosophy and all other ideology; it is as pure as the elemental metal from which the iron rule takes its name. Consequently, explanatory power means the same thing to every modern scientist, regardless of upbringing and inclination, so that every scientist agrees on what the iron rule says, on what satisfies the rule’s criterion for a meaningful scientific test.

You want to use your microscope to settle your differences. It will reveal a physical structure in cuttlefish tissue that is indiscernible to the naked eye, you say, yielding invaluable clues about the workings of the heart and other organs. “Futile,” responds Aristotle. It is not that physical structure has no role to play in bodily functions (although Aristotle does in fact deny that organic tissue has microscopic structure). It is rather that physical structure is not at all the right kind of thing to account for biological function. Even if it were to explain how a cuttlefish moves, it could not explain why the movement is tuned so finely to promote life. Only psuche can do that.

Aristotle might have a point here. The big misses, like the function of the brain or the circulation of blood, wouldn't have benefitted from a microscope, and would have benefitted from some systematic investigation of the purpose of various organs. We absolutely have ways to be empirical about the _telos_ of body systems and Aristotle knows about them.

Rather than deep philosophical understanding, Newton pursued shallow explanatory power, that is, the ability to derive correct descriptions of phenomena from a theory’s causal principles, regardless of their ultimate nature and indeed regardless of their very intelligibility.

what about kepler?

Scientific argument, by contrast, in the sense that matters to the iron rule, is what appears in science’s official channels for broadcasting research, namely, the scientific journals and conference presentations. It is only in such venues, on the printed page and the projected slide, that objectivity is required. There are, of course, many other places in which scientists argue: in their lab meetings, in their conversations over a beer after a hard day’s research, and in their public but unofficial communications, such as television interviews, popular books, and talks at public libraries and corporate retreats. The iron rule does not require objectivity in any of these endeavors. When I say that the rule is concerned exclusively with scientific argument, then, I use that term in a narrow sense—shorthand, in effect, for “official scientific argument.”

this only works because and to the extent that the official communication is genuinely the thing people are listening to when they make up their minds.

a “vast pile of experiments,” in the words of Thomas Sprat, one of the first members of the Royal Society, who in 1667 characterized its purpose thus:

pre newton

The story of snowflakes shows as well as any that the iron rule’s ideal of objectivity for arguments appearing in official scientific communications is only ever partially achieved. But it is not merely a pretense, not simply propaganda. The presentation of scientific evidence and argument may fall short of perfect objectivity while still largely performing its underlying functions: to archive for future generations the outcomes of tests and their explanatory relations to theory and to channel scientists’ energy and attention away from opinion, persuasion, and invective—directing it instead toward the production of exquisitely detailed empirical fact.

does this argument really go through if everybody is fibbing?

To Whewell, the existence of the Christian God was as clear and certain as any observable fact. Yet he accepts that theological considerations ought nevertheless to be disbarred from official scientific dispute in accordance with the rule—not because they are bad reasons, not because they are purely personal reasons, not because of any perceived logical defect, but simply because they are the wrong kinds of reasons for doing science.

Science, then, is built up like a coral reef. Individual scientists are the polyps, secreting a shelly carapace that they bequeath to the reef upon their departure. That carapace is the sterilized public record of their research, a compilation of observation or experimentation and the explanatory derivation, where possible, of the data from known theories and auxiliary assumptions. The scientist, like a polyp, is a complete living thing, all too human in just the ways that the historians and sociologists of science have described. When the organism goes, however, its humanity goes with it. What is left is the evidential exoskeleton of a scientific career.

The iron rule solves this problem not by attempting to glamorize what’s clearly menial, but through a more indirect, more devious stratagem. It sets up scientific argument, as I have explained, as a kind of game in which hypotheses are defended and attacked. In that game, only one kind of move is legitimate: the empirical move in which a hypothesis is attacked for failing to explain some observed matter of fact and defended by showing that the failure is merely apparent, due to malfunctioning equipment, unfavorable conditions, or faulty assumptions.

why did people accept this game? did they?

On the one hand, the idea that the microcosm mirrors the macrocosm was a thesis that went nowhere.

This thesis is underrated. Something like it underpins the notion of "scale-free phenomena" and that idea is still valuable. Often we do see similar dynamics at different scales.

The price of peace in Europe was a permanent bisection of the moral domain into nonoverlapping spheres of obligation, holy church and sovereign nation, each with its proprietary principles and its separate networks of duty and just desert.

huh? this is an era of state religion

thinkers took up Newton’s project, continuing to deepen its fundamentals and to extend its range—developing theories of heat, light, electricity, and the structure of matter—within the same austere framework used by the master himself.

it's a paradigm

TO INVESTIGATE THE dissemination of the iron rule, you might take a scientific approach, carefully monitoring the classroom activities and office hours of science teachers in the schools and universities, tagging along on their hole-digging field trips, and eavesdropping on the advice that senior scientists impart to their underlings over the bustle of the lab bench. I propose, by contrast, a more philosophical approach. Ask yourself: if you had to get the iron rule into the heads and hearts of the next scientific generation, how might you proceed?

Most striking to those in my own profession is the scientific high and mighty’s continual denunciation of philosophy.

This section comes with examples, mostly American, sneering at philosophy. But all these examples are too late for the author's purpose. Modern science ran for several centuries with close overlap with philosophy. Einstein, Bohr, Schroedinger, and peers were highly philosophically literate and that very much shaped their work.So disdain for philosophy can't be the key mechanism for inculcating the iron rule.

Were Descartes a university, it would be a rambunctious and vibrant place, spilling into the hallways and the stairwells. Every member of the faculty would read, discuss, and argue about the work of every other. The physicists and the philosophers, the theologians and the psychologists, would participate equally in a shared discourse on the principles that rule the world. Were Newton a university, you would not hear a sound; the common room would be thick with dust. Each faculty member would be found at all times shut up in their office or lab, pursuing their own researches by their own means, reading just those books that speak directly to their own subject matter, writing only books answering to the same specification. They would meet once a year to discuss parking and the budget for coffee.

Newton was super autistic, yes.

The standard product of this system is an empiricist all the way down, an individual who not only in their public writings but also in their private thinking takes a “scientific attitude” that is directly opposed to the humanistic attitude. The scientific attitude demands tangible evidence. It scoffs at philosophy and is uneasy with a sense of beauty or meaning that cannot be put into words.

I'm not sure this is true.

to a certain degree; it also explains why it is bent to precisely that degree.

In The Beginnings of Western Science, 362–4.

“The Society has reduced its principal observations”: Sprat, History of the Royal Society of London, 115.