Scientists and the symmetry principle: an email exchange with Steven Weinberg

Earlier this week I received an email from Steven Weinberg, the physicist whose book To Explain the World was critically reviewed by Steven Shapin earlier this year. I commented on Shapin’s review in this essay, and in the process I attributed errors to Weinberg that he disclaimed in his email. Weinberg asked that I publish his email, and my reply, on this blog. I was happy to oblige. As I point out in my reply, Weinberg’s reaction to my essay is an additional reason to adopt the symmetry principle that I defended in earlier posts.

Here are the relevant parts of Weinberg’s email:

Dear Dr. Bycroft,

I found a good deal with which to agree in your February 24 blog essay ‘Why historians shouldn’t write off scientists: On Steven Shapin’s review of Steven Weinberg’s [To] Explain the World,’ which I have just seen. But I was infuriated by your reference to ‘the real error in Weinberg’s book,’ that I have ‘judged past scientists by first establishing whether their theories were true or false, and then by assuming that the true ones must be the result of sound reasoning and the false ones the result of incompetence.’

No one who had read my book with an open mind could possibly reach such a conclusion. Throughout my book I was concerned much less with whether scientific theories were right or wrong, than with whether their proponents were advancing the methodology of science. For instance, as noted by a reviewer in the latest issue of The New York Review of Books, I emphasized that the good guess of Democritus—that matter consists of atoms—did not show that his reasoning was sound, and that in fact he had offered no reasoning to support this guess. Like all the pre-Socratics, his theory had no bite—it led to nothing that could allow it to be confirmed. Also, I explained that, although Ptolemy was wrong about the solar system, he had made an important methodological contribution in adjusting the parameters of his theory to get quantitative agreement with observation. I was careful to point out the errors of Galileo about comets and of Newton about diffraction.

Later in your essay, you refer to other supposed ‘errors’ of mine. In short, of these too I am innocent.

To your credit, you were honest enough to admit that when you wrote this you had not read the book, and were relying on Steven Shapin’s description. The dangers of such a procedure are evident.

Here is my reply:

Dear Dr. Weinberg,

Thank you for taking the trouble of reading my blog essay on Steven Shapin's review of your book. I am glad that you found ‘a good deal with which to agree’ in the essay. I hope it was clear from the essay that my main target was Shapin’s review, not your book. If had intended to write a review of your book, I certainly would have read the book before casting aspersions on its content, as I do on two occasions in my essay.

On one occasion (‘As Shapin points out, it ignores...’) my phrasing is ambiguous. I seem to say that I have information about your book that is independent of Shapin's review. This is not the case, and I have inserted a footnote to this effect.

On the other occasion I say that I am making an educated guess based solely on Shapin’s review. Perhaps I should have added that this was a risky guess based on a small sample of citations plucked from the review. In any case, I have spelled this out in the footnote I just mentioned.

In making this educated guess I was less interested in evaluating your book than I was in drawing attention to what is known in the trade as the ‘symmetry principle’ or ‘methodological relativism.’ As I have argued elsewhere on my blog, this principle is badly misunderstood. Even professional historians of science have trouble saying what it is and why we should adhere to it. There is a tendency to confuse the principle with the idea that historians should never evaluate past science, or that they should never invoke the truth of a belief to explain that belief, or even that intellectual factors matter little in scientific debates. The result is that many legitimate and useful practices are criticised on the grounds that they violate a principle that few historians are able to articulate in a coherent manner. My remarks on your book were part of my effort to separate the many putative symmetry principles from the real one.

What is the real symmetry principle? I gave my version in the post in question. It is that we should not ‘judge past scientists by first establishing whether their theories were true or false, and then by assuming that the true ones must be the result of sound reasoning and the false ones the result of incompetence [or worse].’

You imply in your email that you agree with this principle. I find this very heartening. In the past, historians and sociologists have had great trouble persuading scientists that there is a version of the ‘symmetry principle’ that is worth taking seriously. The One Culture?, which I reviewed recently on my blog, and which featured chapters by yourself and Shapin, contains ample evidence of this problem. It looks like we have found a solution.

The symmetry principle would be worthless if no-one was ever tempted to violate it. I am grateful for the examples you invoke to show that you have not done so. I think the Galileo and Newton examples are equivocal. They would only be telling if you specified the causes of the errors in question. Did these men go wrong because they were temporarily disabled by a bad method or a social or psychological bias, or because even the best methods and the most clear-eyed scientists will go wrong some of the time? Both explanations are consistent with the symmetry principle, as I understand it. But the former is also consistent with a violation of the principle.

The Democritus and Ptolemy examples are more convincing. There are comments under my blog essay that defend your book along similar lines, eg. the comments by Michael Weiss (here and here) and by Will Thomas (here). As you point out, however, it is better to read a book than to take someone's word on its contents. So I will reserve judgement on your methodology until I have had a good look at To Explain the World.

I hope that the two offending passages in my essay do not obscure my broader point, which is that the methodological gap between professional historians (like Shapin) and scientist-historians (like yourself) is smaller than the former group sometimes makes out. The symmetry principle is one aspect of this. Another aspect is the fact that scientist-historians have actually done the thing (science) whose history they study. You may be interested in a more recent essay in which I look into this issue, starting with your statement that history is something that you ‘visit as a tourist.’

Weinberg wrapped things up with the following:

Dear Dr. Bycroft,

I’m heartened by the convergence in our views. In particular, I do agree with the ‘symmetry principle’ as you have stated it, though this is one symmetry principle that had previously escaped my attention. I agree that the comments in my note to you about Newton’s fudging and Galileo’s error are not relevant to this principle. I included these comments only to exonerate myself from accusations of hero worship. As you say, the relevant examples cited in my note were those having to do with Democritus and Ptolemy.

Expand post.

Malheureusement je ne traduis plus tous les posts que j'écris sur ce blog. Si vous voudriez lire les posts qui ont été traduits, veuillez cliquer ici.

Agrandir ce message.

Is it post-modern to be present-centred? Thoughts prompted by Nick Tosh

Lorraine Daston says that historians who write with the present in mind are misguided and old-fashioned. By contrast, Nick Tosh says they are post-modern. He compares them to contemporary novelists who draw attention to themselves, and to the process of writing novels, in the course of their novels. Against Daston, I said that it is not what you know about present-day science that matters, but how you use that knowledge. Against Tosh, I say that present-centredness is not post-modern because it does not leap from the fictional world to the real world but only from the present and the past. And very often it does not even do that.

Don’t get me wrong about Tosh. His discussion of history of science as literature is sharp and original. It appears at the end of his article ‘Science, Truth and History, Part I: Historiography, Relativism and the Sociology of Scientific Knowledge,’ Studies In History and Philosophy of Science Part A 37:4 (2006): 675-701 (paywall). He starts by distinguishing two kinds of reader:

A work of history is written for members of one language game about members of another. To read such a work is not literally to jump to an alien language game, of course, any more than to ‘feel another’s pain’ is to jump to an alien consciousness, but good history writing can sometimes induce illusions along those lines. It is obviously crucial that the reader bear in mind the artificiality of such experiences. Genuine historical understanding requires more than a passive talent for empathy: it requires an active appreciation of the logical gap separating us from a past culture, and of the historian’s role with respect to that gap.

We can draw an analogy with fiction here. A naïve novel reader is simply someone who can lose himself, on demand, in a make-believe world spun by the novelist. A sophisticated reader (also) understands something about the process of creative writing (as an activity in the real world), and about the kinds of interactions that take place between author, characters and reader. The aesthetics of fiction are not just about pleasing the first sort of reader.

That should be clear enough. Tosh goes on to say that post-modern novels target the sophisticated reader:

‘Post-modern’ novels can be seen as emphasising their own artificiality. Consider, for example, this authorial intrusion one-quarter of the way through Martin Amis’s Money:

I was just sitting there, not stirring, not even breathing, like the pub’s pet reptile, when who should sit down opposite me but that guy Martin Amis, the writer. He had a glass of wine, and a cigarette—also a book, a paperback. It looked quite serious. So did he, in a way.

What Goffman called ‘breaking frame’ and the Russian formalists ‘exposing the device’ is not supposed to cheapen the reader’s appreciation of a particular novel; it is supposed to deepen his understanding of fiction in general. If one can’t escape the contrived nature of the novel-reading experience, then, as an author, one might as well stop trying to and instead make artistic points about (and through) it.

Next Tosh returns to the history of science, pointing out a striking inconsistency between theories of history and literature:

So we can distinguish two basic authorial responses to the gap between fantasy and reality: pretend, for aesthetic reasons, that it doesn’t exist, or stress, for different aesthetic reasons, that it does. Traditional novelistic etiquette emphasises the former tactic; post-modern novelists have often plumped for the latter. It is perhaps rather odd, then, that while an analogous pair of conventions exists for history writing, the cultural connotations are almost exactly reversed [my italics, not Tosh's]. ‘Breaking frame’ is such old news in the history business that it hasn’t even got a flash name. It is the absence of authorial intrusion (that is, the absence of modern moral or epistemic judgements) that is likely to get a work of history labelled ‘post-modern’ or ‘relativist’. While their philosophically-inclined colleagues theorise endlessly about cultural separation and the historical ‘Other’, post-modern historians of science do their best to keep the present out of their narratives—to make separation and otherness invisible. Actors’ categories history, like the traditional novel, is a genre that seeks to hide its own artificiality; it seeks to cooperate with the reader’s attempt to suspend disbelief...

Tosh ends with a remarkable prediction:

I will end this coda by hazarding a prediction, inspired by the history of the novel but grounded upon the above argument. The aesthetic bias that lies behind cultural history’s methodology (and, arguably, SSK’s philosophy) may not in fact be a very long-lived one. Future historians may read our cultural histories, including our histories of science, rather as we read the great classics of Victorian fiction: marvellous achievements in their own way, but ultimately limited by the strict literary form to which they adhered.

***

Tosh may be right that cultural histories, with their lush reconstructions of past beliefs, will soon be replaced by a more knowing kind of history. But if the replacement is truly post-modern, it will not be present-centred history as we know it.

One reason is that historians write about the world we all live in whereas novelists write about a different, imaginary world. This means that the ‘authorial intrusions’ of the historian as less jarring than those of the novelist. When Martin Amis put himself in his novel, he stepped from the real world to the fictional world. When Mary Boas-Hall criticised Thomas Hobbes, she stepped from the twentieth century into the seventeenth. The former jump is (we might say) metaphysical, whereas the latter is merely chronological.

Only the former jump has an air of paradox that invites the label ‘post-modern.’ Fiction does not work unless the reader believes, at some level, that the events being described really happened. The effect of the ‘authorial intrusions’ that Tosh associates with post-modern novels is to remind the reader that the events didn’t actually happen. There is no equivalent paradox for history. It is true that history only works if the reader believes that the events being described happened in the past. It is also true that, when the historian refers to himself in the course of a book, he reminds the reader that the book was written in the present. But there is no paradox here. The reader is not asked to believe that certain events both did and did not happen. He is asked only to believe that historians can find out about the past.

My other objection to Tosh is that, despite its name, present-centred history does not necessarily involve chronological jumps. Suppose a historian writes: ‘the mercury in Hobbes’ barometer was held in place by the pressure of atmospheric air, not by nature’s aversion to vacuums.’ The event being described here, the motion of the mercury, is an event in the seventeenth century, not in the twentieth. The statement does not require a chronological leap any more than the statement ‘Boyle was a devout Christian’ requires such a leap. Granted, we can easily introduce a chronological leap, for example by prefacing the sentence with ‘As we now know,’ or ‘In my opinion.’ But the critics of present-centred history of science do not make such nice distinctions—they would criticise the statement even without the prefatory phrase, calling it anachronistic.

The statement does make a leap, but it is one of narrative perspective rather than chronology. A historian who makes that statement has departed from Hobbes’ perspective, since the statement contains an explanation that Hobbes rejected. Which perspective has the historian adopted, if not Hobbes’? We could say that he has adopted Boyle’s perspective. But that would raise the question of why he has favoured Boyle over Hobbes. Worse, in many cases there will not be a Boyle waiting in the wings to serve as spokesperson for present-day theories. It looks like we are forced to say that the historian as adopted his own perspective, which suggests that he really has made a chronological leap.

But there is another option. We can borrow a phrase from literary criticism and say that the historian has adopted the perspective of the omniscient narrator, the voice that speaks for no-one but is aware of everything. In fiction, the omniscient narrator often describes events that none of the characters in the story will ever be aware of. To pick a random but vivid example, there is a passage in Steinbeck’s The Grapes of Wrath that describes the gradual decay of a house abandoned by share-croppers. The whole process is witnessed only by pigeons and stray cats. Yet there does not seem to be anything jarring or post-modern about this passage. It is completely different from the Amis passage that Tosh quotes. The ‘authorical intrusions’ of present-centred history are more like the Steinbeck passage than the Amis one.

If the statement about Hobbes’ barometer is not post-modern, what would count as post-modern history of science? Well, what makes the passage from Amis post-modern? Tosh highlights the following two features. Firstly, Amis refers not just to himself but also to process of writing novels. Secondly, this reference draws attention to the ‘artificiality’ and ‘contrived nature’ of novels. So we are looking for histories of science that refer to the process by which the histories were researched and written (not just to the beliefs of the author). We are also looking for histories that draw attention to the weakness or partiality of our knowledge of past science (which is not the same as drawing attention to differences between past and present science). Perhaps there are histories that meet both of these criteria, but I do not know of any. What I do know is that histories of this kind would be quite different from the present-centred histories of Whewell or Sarton or Weinberg. Expand post.

Malheureusement je ne traduis plus tous les posts que j'écris sur ce blog. Si vous voudriez lire les posts qui ont été traduits, veuillez cliquer ici.

Agrandir ce message.

Lorraine Daston on history as fiction – critical thoughts

Nick Tosh and Lorraine Daston have, separately, made some intriguing remarks about the affinities between fiction and the history of science. Their remarks follow on nicely from my last post. Daston compares historians to readers of mystery novels who know in advance how the story ends. She means this as a criticism. By contrast, Tosh praises authors who freely acknowledge the gap between themselves and their subjects, a stance that Tosh labels ‘postmodern.’ My position is somewhere in between: it is OK for readers to know how the story ends, but this is not a particularly postmodern phenomenon.

Daston made her remarks in a 2012 interview with the Canadian radio station CBC:

Imagine you’re the kind of person who cheats when you read mystery novels, and you read the last page first, to find out who did it. And then when you read the rest of the mystery story, you know that everything is building towards this climax. You read it in a very different way than the person who has to retrace all of the red herrings that have been planted in your way by the author, to throw you off the scent of the real villain. We the historians of science were always the people who read the last page first, and we knew how the story ended.

This statement is plausible on the surface: you read a novel differently if you know the outcome, and historians of science usually know the outcome of the scientific debates they study. But note that Daston has compared two different ways of reading a detective novel, and that she uses this comparison to assess two ways of writing history. Foreknowledge may change the way we read, but does it change the way we write?

The analogy to detective fiction suggests that it does not. Mystery writers do know the outcome of their stories, and despite this they revel in the kind of red herring to which Daston refers. And why not say the same about readers? If the prescient writer can appreciate the red herrings, why not the prescient reader? As per my last post, mystery novels can be gripping even when we know in advance who committed the crime and why.

Daston continues:

Moreover, we were complicit in this with the scientists themselves, who wish to have a story about why we believe what we believe now, and why it is right, because the scientist knew, in a way that the historian did not know so viscerally, that everything—or most things—that we now believe in science will be overturned, if not in a generation then in two generations. This is the progress of science, but it is the pathos as well. Therefore the reassurance of that kind of history of science was: it is inevitable to believe what we believe now, indeed there are no other possible positions one could have...it was that deep reassurance which more than we the historians needed, the scientists needed.

In other words, by writing history with the outcome in mind, historians made those outcomes seem more rational than they really were. This reassured scientists because deep down they knew that most of the outcomes were not particularly rational. My objection is the same as the one I stated above. Historians and their readers can include all the red herrings—all the accidental developments, the roads that led nowhere, and the roads that might have led somewhere but were not taken at the time, and might still be taken—they can include all of this even though they are perfectly aware of what scientists now believe.

What I find especially odd is the idea that knowledge of the status quo is a conservative force, one that stifles dissent and disagreement. In the political realm, those who seek change are usually well advised to study the way things are. This knowledge can be rhetorically as well as strategically valuable. It allows the revolutionary to present his cause as a lonely struggle against massive opposition—not so much opposition that the cause is hopeless, but enough to endow it with romance and urgency. Animal rights activists do not ignore the mistreatment of animals. On the contrary, they spell it out in gory detail. Feminists do not turn a blind eye to laddish behaviour—they hunt it down and string it up. Many skeptics about climate change do not deny that there is a scientific consensus on the topic—instead they try to show that the consensus is rotten.

The grain of truth in Daston’s statements is that historians who overemphasise the present condition of science are usually the ones who exaggerate the naturalness of that condition. My point is that the source of this exaggeration is not knowledge of present-day science but the way this knowledge is used. The principal error is to assume that the theories of present-day science were as defensible in the past as they are today—that there was as much evidence for them in the past as there is today. In history and in mystery novels, it does not matter whether we know how the story ends. The important thing is to not mistake the story for the ending.

Nick Tosh follows in my next post. Expand post.

Malheureusement je ne traduis plus tous les posts que j'écris sur ce blog. Si vous voudriez lire les posts qui ont été traduits, veuillez cliquer ici.

Agrandir ce message.

Histories of science as murder mysteries, or: Steven Weinberg as Henning Mankell

The conclusion of my last post was that scientists who write history are like visitors to one city who live in a different city (as opposed to historians, who study cities while living in the countryside). The point of the analogy was to show that knowledge of present-day science need not get in the way of good history-writing. There is another analogy that gets at the same point from a different direction: scientists who write history are like authors of murder mysteries who reveal the identity of the killer in the first chapter. This may sound like a criticism, but there are successful authors who actually write like this, starting with the Swedish maestro Henning Mankell.

SPOILER ALERT - crucial details of detective novel revealed below, most of them from the first chapter

Consider Mankell's Villospår (1995), also known under the much less chilling title Sidetracked. Four of the murders in the story are narrated from the point of the killer, which means that the reader is always several steps ahead of the detective, the stern but sensitive Kurt Wallander. Within the first 20 pages we know that the killer acts alone, rides a small motorcycle, attacks his victims with knives and axes, is engaged in a vendetta on behalf of his sister, and buries the scalps of each his victims under the window of the hospital ward in which his sister is confined.

Wallander knows none of this. If he knew all of it, he would solve the mystery after the third murder, if not before. This would make the book shorter and much less interesting. It would mean that, as readers, we would not find ourselves in the edifying position of admiring Wallander’s deductions while knowing that his conclusions are wrong. We would not take nearly as much pleasure in the suspenseful meetings between Wallander and the person who (as we but not Wallander knows) committed the grisly murders. Nor would we know that Wallander placed his life in danger by speaking to this person. Most of all, our attention would not be drawn away from the facts of the murders (these facts are boring because Mankell hands them to us on a plate) and towards the process by which Wallander discovers those facts. This process, including the many dead-ends along the way, is the real mystery of the novel.

Mankell’s approach in Sidetracked is not unlike Steven Weinberg’s in To Explain the World.* Weinberg makes no secret of the fact that the earth follows an elliptical path around the sun, that inertial motion is rectilinear rather than circular, that the speed of light is finite, and numerous other present-day orthodoxies that were out of reach of many of the past scientists he discusses.

These disclosures do not ruin Weinberg’s narrative any more than they ruin Mankell’s. On the contrary. They teach us that skilled scientists can be wrong for the right reasons. They lead to dramatic scenes in which scientists come tantalisingly close to a discovery without quite getting there, where they unwittingly endanger their lives or careers by meddling with phenomena they do not understand, and where they become acquainted with a new entity without correctly identifying the entity (for an example of the latter, consider Joseph Priestly’s isolation of oxygen). Most importantly, a plain statement of the facts of nature has the paradoxical effect of drawing our attention away from those facts and towards the circuitous route that leads eventually to their discovery. This route, not the facts themselves, is the real story of the history of science.

Granted, there are relevant disanalogies between histories of science and murder mysteries. One is to do with the contemporaneity of those who know the facts and those who inquire about them. In Sidetracked, as in most murder mysteries, the facts about the crime are known (by the criminal) even before the inquirer (the detective) knows that a crime has been commited. In most histories of science, the facts about nature are not known by anyone until they are discovered by one of the principal inquirers in the narrative.

The second and related disanalogy is that, in a murder mystery, the discovery of the facts is almost always part of the story—the murderer, the means and the motive are correctly identified by someone in the book, usually near the end. By contrast, histories of science need not end with the identification of a true theory or the correct identification of a phenomenon. Often they do, but there is nothing in the rules of the genre that requires this.

These disanalogies do not seriously undermine the point I am trying to make, which is that histories of science can benefit from knowledge that is unavailable to most or all of the scientists they deal with in any given narrative. The first disanalogy is harmless because in both cases—histories of science and murder mysteries—the reader has access to facts that are unavailable to the inquirer. It does not matter whether there is someone else, living at the same time as the inquirer, who knows those facts as well as the reader. The second disanalogy is harmless because there are enough histories of science that do end with a correct solution to a problem to make the comparison to murder mysteries worthwhile.

I am not the first to compare histories of science to murder mysteries. Lorraine Daston has also done so, though not in the same way as I have just done. And Nick Tosh has some interesting remarks on the literary merits of historians who, like Weinberg, wear their present-day knowledge on their sleeves. I respond to Daston and Tosh in my next post.



*As I imagine Weinberg’s book, which I have still not read. It matters little whether my hypothetical description of Weinberg’s book is accurate. The features I describe are endemic in the kind of histories of science that are called ‘Whiggish’ or ‘present-centred.’ Expand post.

Malheureusement je ne traduis plus tous les posts que j'écris sur ce blog. Si vous voudriez lire les posts qui ont été traduits, veuillez cliquer ici.

Agrandir ce message.