The Knowledge Machine: How Irrationality Created Modern Science

Unlike the humanities, including philosophy—where the idea of progress is a controversial topic—it is an essentially indisputable fact that science makes considerable progress over time. Why this is the case—and how science actually works—is what Michael Strevens seeks to explain in The Knowledge Machine.

The basic argument is that scientific knowledge grows through the application of the “iron rule of explanation,” as Strevens calls it, that demands that all scientific argument be settled by empirical testing alone, and that the results of empirical testing are to be recorded in formal scientific journals for future reference and use.

The iron rule is peculiar in the sense that it demands adherence to empirical testing and show more does not consider the relevance or significance of any non-empirical knowledge, whether philosophical, religious, spiritual, or aesthetic. While individual scientists are free to theorize in whatever manner they like—and are swayed by the same philosophical, moral, and political influences and biases as everyone else—the iron rule of explanation guarantees that formal arguments are presented without reference to any of these ancillary considerations.

The net effect of this “procedural consensus,” over time, is what Strevens refers to as “Baconian convergence,” or the idea that repeated empirical testing over time converges on the one theory that best explains all the accumulated data. This is why physicists, over time, have eventually come to accept the legitimacy of the general theory of relativity, for example, whereas a religion like Christianty schisms permanently into a thousand different parts.

The reason for this is that philosophical and theological reasoning, while attempting to be more ambitious and all-encompassing, has no ultimate method of verification or falsification through testing. If I think, as Isaac Newton did, that Jesus was created by God and subordinate to God, whereas you think that Jesus and God are one and the same, how are we supposed to resolve this philosophical difference? I can provide my logical and coherent reasons and you can provide yours—along with our respective interpretations of the relevant scripture—but without a procedural consensus whereby we can ascertain the truth beyond mere logic, there is no way to settle the argument.

On the other hand, if I believe the Newtonian theory of gravity is correct and you believe the Einsteinian theory is correct, we can (if we were capable) settle the dispute by measuring the angle that light is bent by the sun’s gravity during a total solar eclipse, as Arthur Eddington and others did. Newton’s theory predicts one measure; Einstein’s predicts the other. We can both agree to what the measurements will tell us beforehand; then, after the experiment is conducted and the measurements are verified, we can settle the dispute.

Of course, as Strevens points out, it’s not exactly this simple. An individual scientist must still engage in “plausibility rankings” and determine how to weigh conflicting evidence. There is a strong element of subjectivity in the interpretation of evidence and the process is far from completely objective. But the main reason why science is effective is not due to the unwavering rationality of any individual scientist; rather, it is attributable to the process of several scientists over time abiding to the iron rule of explanation—and publishing detailed empirical findings—that allows the process of convergence to occur over time and the correct theory to materialize. This is why science advances, and why we are now able to launch satellites into space and communicate with each other around the globe electronically at the speed of light.

So far, so good, but why did the iron rule of explanation—which has proven to be so effective—take humanity so long to develop? Why did it develop in 17th-century Europe and not, for example, in ancient Greece or China? The reason is, according to Strevens, that the iron rule is, at bottom, irrational. It asks the practicing scientist to effectively ignore all other forms of human inquiry that is not strictly empirical. This would have seemed absurd, especially in ancient Greece, the birthplace of philosophy, or in Medieval Europe, obsessed with theology as it was. To give up all philosophical and theological reasoning in the attempt to explain how the world works was too radical an idea for most times and places.

This is why science had to wait for the peculiar historical and cultural circumstances of early modern Europe. Only then, and only over time, did it begin to make paradoxical sense that knowledge of the world can only grow by significantly limiting its scope to empirical testing and data alone. Since then, the iron rule of explanation and its procedural consensus has resulted in Baconian convergence and a growing and sophisticated understanding of the workings of the world.

Strevens, I think, has hit on something profound in this book, and his explanation for how science works is ultimately convincing. However, I take some issue with the title of the book and on his calling the process of scientific discovery irrational.

The iron rule is not, in itself, irrational; its tremendous success over the last few hundred years should attest to that. By limiting scientific argument to empirical data alone, our knowledge of the world has increased astronomically in a short period of time.

The irrationality, then, does not lie in the iron rule itself; rather, it lies in the belief that the iron rule applies to problems outside the realm of science. If you believe that philosophical, ethical, and political problems can be solved with empirical argument alone, then yes, your overextension of the iron rule is indeed irrational.

But if you limit the scope of the iron rule to scientific, empirical problems, then there is nothing irrational about the rule because there is nothing in the rule that says you cannot compartmentalize scientific problems. Isaac Newton should have demonstrated this; he abided by the iron rule in his scientific work while simultaneously pursuing other philosophical and mystical pursuits. There is nothing irrational about this. The irrationality, rather, comes from someone like Stephen Hawking, who said that “philosophy is dead” because he couldn’t apply the iron rule to philosophical problems—problems it is not meant to address.

There is no “theory of everything”; reality is complex, like a six-sided cube you cannot view all from the same perspective. Different problem types require different approaches, and science has developed, according to the iron rule, its own successful approach. Philosophical, historical, ethical, legal, and political problems all have their approaches as well, and, while they all influence each other, no single domain has authority over all the others.

Science has simply limited its scope to empirical testing to solve certain kinds of problems. While the scientist that thinks this particular approach can solve all types of problems is certainly irrational, scientific problem solving, led by the iron rule, is not. show less

This was an interesting book. It does not fit nicely into my usual genres - or into my expectations of how a book should be organized. The organization worked; the book was interesting; but in a a sense its credibility was contradicted by its topic.

The scientific method seems to be strange and difficult for most people to understand. I'm not entirely sure why; it seems pretty natural to me. Perhaps I'm merely properly indoctrinated, in spite of never having been a working scientist. But on the other hand, perhaps I'm merely on the autistic spectrum, and have an unfair advantage in keeping certain common human thought distortions out of my thought process.

At any rate, the thesis of this book is that the scientific method is both show more unnatural and irrational - except that it works far better for its purpose (developing accurate knowledge of the world) than other, more natural means of understanding. Strevens suggests that developing such a method - and using it long enough for its practical advantages to become obvious - was always extremely unlikely, but had somewhat of a better chance in Europe in the aftermath of the religious wars - people had gotten into the habit of compartmentalizing domains of knowledge, separating religion from politics, because the alternative seemed to be yet another round of religious wars. Thus separating observed data from everything else, while doing science, or at least while reporting on scientific work, seemed less unnatural at that time than it ever had before.

The book starts by discussing the views of Karl Popper and Thomas Kuhn about how science works, first contrasting them with each other, and then demonstrating that neither adequately describes the actual behaviour of some/all/most scientists. He then gives his own version of how science really works - or at least of the basic rule that must be followed for science to work at all.

In particular, he postulates something he calls the iron rule of explanation: scientists may only attempt to resolve their differences of opinion by conducting empirical tests. No fighting, no shouting, no revealed scripture, no appeals to past masters. Just empirical results. They may - and generally will - differ about what those results mean - but that can only be addressed by gathering and reporting more results. They may use any means they like to come up with theories, or ideas of what data to gather - but when they report the data, they are expected to sterilize it - not tell us that they think their results are in accord with Scripture, or that their experiment was inspired by a dream. At least, not tell us this in the scientific literature - they can say whatever they want in a memoir, or at the pub over a beer.

Paradoxically, leaving out extra sources of information results in better results. (At least, it's paradoxical if you truly believe dreams, divine revelation, etc. etc. provide useful information on the topic at hand.) Strevens uses a stronger word here - "irrational" rather than "paradoxical"; in my opinion, that word choice weakens his argument somewhat.

Unfortunately, Strevens is arguing for his position using the methods of the humanities. And not even the more data-minded of those methods. (We see examples, not statistics.) To me, this seems to be rather weak evidence.

Of course his approach is somewhat inevitable - he's a philosopher, after all. But I suspect he also doesn't get it - in the sense that he can't imagine being motivated to do science himself. Non-scientifically, I suspect that because he doesn't truly get it, his model of the motivations of scientists is likely to be somewhat inaccurate. (I imagine him as me, trying to understand the values and motivations of fashion designers, beyond the obvious one of making money.)

In any case, that doesn't really matter. His iron rule is a worthy addition to the existing theories of how science really works. But I don't think his methodology - or that of Kuhn for that matter (I haven't read Popper) is going to convince anyone of anything they didn't already want to believe. All 3 descriptions of how science really works are probably best regarded as "just so" stories - science speak for explanations not subject to empirical investigation.

That said, all three of them may be useful for potential scientists in understanding what they are, ideally, supposed to be doing. Or more correctly, how they are supposed to go about seeking knowledge, and some of the ways in which that search can go astray.

I'm not sure what (if anything) is useful for those prone to bastardize the word science, or one of its sub-fields, with pseudo-intellectual mashups like "creation science" or "indigenous physics". Mostly I just ignore them. show less

Although I’m persuaded by Streven’s core idea that modern science depends on a strict separation between empirically collected data and speculative explanatory theories, I feel like he pushes too hard to find examples everywhere in history. While some of his examples are perfect (e.g. how Eddington revealed all his data about the Einstein eclipse, even when it apparently contradicted his conclusion), I question how much the “typical” scientist really adheres to this rule. This is especially obvious in the “soft” sciences, where “data” is itself subject to interpretation, but even in a field like microbiology, it’s hard to deny how much of the data collection is influenced by “paradigms” like the Central Dogma.

That show more said, I’ll take away the main point: that science is nothing more than data methods of the data collection logic that ties it together. show less

Good explanation of what science *is* and is *not* and speculations about why it developed in Europe 500 years ago and not some other place or time. He talks about Kuhn and Popper, agrees with some of their theories and disputes others. For a book about science it’s pretty flowery and poetic - well done but sometimes I wished he were a little more direct.