Looking twice at the history of science

Tuesday, February 24, 2015

Why historians shouldn't write off scientists: on Steven Shapin's review of Steven Weinberg's Explain the World

Say what you like about the science wars, they’ve got legs. A few days after finishing a review of The One Culture, the book that was supposed to end the quarrels between scientists and sociologists/historians of science, I learned that two of the contributors to that volume have been involved in a new skirmish. The opening shot was a new book by the physicist Steven Weinberg, called To Explain The World. The riposte came from the historian of science Steven Shapin, who has reviewed the book under the uncompromising title 'Why Scientists Shouldn't Write History.' Will Thomas, who drew my attention to the review, comments that Shapin’s review is ‘unduly divisive’ and that historians ‘ought to take seriously...the objections and perspectives of scientists.’ I agree, but I would go further. The reason historians should not write scientists off is not just that the scientist's perspective is valid but also that it overlaps with the perspective of some historians. To a large extent, the division that Shapin sees between scientists and historians is better understood as a division between historians. Expand post.


  1. Thanks for these thoughts, Michael. I come to this issue by way of having reviewed the book in question myself for Physics Today (a bundled review with Demetris Nicolaides's In the Light of Science: Our Ancient Quest for Knowledge and the Measure of Modern Physics that will appear next month).

    First, I did find that Shapin neglected to acknowledge some things Weinberg does well. When we get to the early modern period and start to see developments that Weinberg deigns to care about getting right, he actually does quite a nice job of reconstructing the conceptual developments in a lucid, technically sophisticated but still accessible way.

    That said, I largely agree with Shapin about his specific criticisms of the book, especially the first half. Weinberg runs roughshod over ancient science, which he largely dismisses as idle philosophizing. The suggestion, in so many words, that Ptolemy would have invented the telescope had he been any kind of scientist was just one in a series of eyebrow-raising passages in this section. I also suspect that any self-respecting historian will cringe when Weinberg savages Bacon and Descartes as steeply over-appreciated, Bacon because he was so foolish as to only care about how empirical knowledge could serve society and Descartes because his ontology was eventually tossed aside. I'll also say in Shapin's defense that the WSJ likely picked the click-bait title for the review, though I can't confirm this.

    My deepest reservation about the book, however, is one Shapin doesn't address. It's a philosophical concern, not a historical one. Weinberg is famously hostile to philosophy of all stripes and so is blind to the ways in which his book is really an argument for some implicit philosophical commitments. He aims to reinforce a certain set of scientific values--those that align well with his reductionist view of the world--and so tells a success story about those values. The greatest difficulty I had with the book is that Weinberg assumes that his philosophical commitments about scientific values are self-evident truths and doesn't recognize that a defense of those values is a philosophical act, conscious or not.

    This, I think, is a useful segue into a discussion of your point. Reading past science through the lens of present science goes beyond providing more robust tools to assess older work. It also imposes a set of epistemic norms, methodological precepts, and other values. But not all scientists agree on what those norms and values are, and so there isn't just one way to read the past through present science (and if you don't believe me, check out the Nicolaides book!). One way to understand your point--which I think is a good one--is to say that historians' commitment to approximating past values as best as possible is also a philosophical, value-laden commitment. The advantage that approach has, though, is that it dodges the question of whose values belong in the lens. We have a ground state that tells us it should be the actors' values, as best as we can discern them. I myself am happy to depart from that ground state when it's justified, but I'd need to see a good justification and Weinberg was not able to convince me that he had one. So I'd qualify your conclusion by saying that anticipations and evaluations don't have to lead to bad history, but they're pretty likely do if done without forethought about what they're supposed to accomplish. And that's hard work that ultracrepidarians of Weinberg's stripe are often disinclined to do.

    1. Thanks very much for your comments, J. D., and sorry for the long delay in my reply.

      I take your point that Weinberg smuggles philosophy into his account of past science, and that his philosophy of science is not universally shared even by scientists in his own field (physics). I have not read Weinberg's book, but it sounds highly plausible to me that his tacit assumptions about what counts as good science lead him astray as a historian.

      I would add some caveats, though. Firstly, we should distinguish anticipations from evaluations. Shapin is worried about both, but your concern only seems to apply to evaluations, since only for evaluations do present-day norms and values play a direct role.

      Secondly, contentious norms can easily be turned into uncontentious descriptions. Suppose Weinberg gives Descartes a bad wrap because Descartes used mechanical models. Many scientists would disagree with Weinberg's assumption that mechanical models are bad science. But no-one can deny that Descartes used mechanical models. Of course it is not very interesting to be told that Descartes used mechanical models, because we knew that already. But it is conceivable that, in some cases, Weinberg's norms lead him to make descriptive claims that are more interesting than that one.

      The third caveat is an extension of your statement that 'historians' commitment to approximating past values as best as possible is also a philosophical, value-laden commitment.' You point out that this commitment 'dodges the question of whose values belong in the lens.' My comment is that the historians' claim to consider only actors' norms is no guarantee that the historian will practice what he preaches, just as Weinberg's claim to consider only the shared values of modern science is no guarantee that other scientists do, in fact, share his values.

      The final caveat is to do with how the historian justifies his evaluations of past science. Some historians use past science as evidence for their methodological preferences, eg. by arguing that the followers of Newton's instrumentalist method were much more successful in generating true theories than were the followers of Descartes' mechanical method. A historian who does this is not tacitly assuming that instrumentalism is best but rather demonstrating, with historical examples, that it is best. In doing so he makes assumptions about the truth or falsity of the theories adopted by Cartesians and Newtonians. But claims about whether a past scientific *theory* is true or false are less controversial than claims about whether a past *method* was sound. So this kind of evaluative history is immune to the criticism that you raise.

      The same is true of another kind of evaluative history, the one that is probably the most current among professional historians of science. I have in mind historians who write about the 'losers' of past science -- those who got it wrong, and seem to have done so for the wrong reasons -- and show that there are at least *some* widely endorsed present-day methodologies according to which the loser's reasons were actually better than the winner's reasons. Such historians do not say the loser's methodology is the only one endorsed by today's scientists, or even that it is a good methodology. Their aim is just to point out inconsistencies between today's methodologies and the received wisdom about the abilities and contributions of past scientists.

      To sum up: yes, evaluations are 'pretty likely' to lead to bad history 'if done without forethought.' But history always requires forethought, even history that is not explicitly evaluative. And many historians, including those who are scientists by day, are capable of the required reflection.

    2. Your comment complements the original post insofar as it shows that Weinberg, like Shapin, does not speak for his entire discipline when he writes about past science. Just as historians disagree with Shapin about how to do history, so scientists disagree with Weinberg about how to do science.

      The positive side to these disagreements is that Shapin and Weinberg represent quite extreme positions in their respective fields. I suspect that most historians are more forgiving about anticipations and evaluations than Shapin, and that most scientists are less dogmatic about scientific method than Weinberg. Certainly they are the two authors who adopted the most extreme positions in 'The One Culture,' the book about the science wars that I mentioned at the beginning of the above post.

    3. Hi Michael– Just saw this now, and deep into term-time drama, so just a quick note. I sense that I more or less agree with you on the general point. Anecdotally, I know and interact regularly with a number of physicists who do very good history precisely because they've dedicated themselves to doing the requisite reflection. And if their training gives them a somewhat different perspective on how to ask questions or make assessments, then I find that it offers useful diversity to the field. The question for me is what type of reflection Weinberg is doing in this particular case, and it seem to me that it is primarily of a kind designed to support his very specific philosophical predilections.

      Incidentally, I'll likely be spending some time in your corner of the world shortly, as my better half has taken a postdoc in your shop and starts next month. Would be keen to buy you a pint and talk through some more.

    4. Sure thing. Not sure how to get in touch, though...

      By no means would I defend everything Weinberg has written, or even everything I've read by him. Nor would it surprise me if, as he's aged, he's become more dogmatic.

  2. I also haven't read Weinberg's new book, nor can I read Shapin's review without piercing the paywall. But I have read Weinberg's old essay The Revolution That Didn't Happen, his assault on Kuhn's Structure. I'd be surprised if this didn't traverse some of the same ground.

    Weinberg does a very good job dismantling Kuhn's notion of incommensurability. My impression is that incommensurability has not worn well even with historians. Weinberg and Kuhn also butt heads over the idea of objective truth; I don't expect to see this issue resolved anytime soon (or even within this millenium).

    Weinberg's article also talks about Kuhn's "road to Damascus" moment, or as Kuhn puts it:

    My jaw dropped with surprise, for all at once Aristotle seemed a very good physicist indeed, but of a sort I'd never dreamed possible.

    Here I wonder if Kuhn and Shapin don't share much the same outlook, along with virtually all modern historians, and in stark contrast to Weinberg's viewpoint (and, I suspect, the viewpoint of almost all modern physicists).

    Aristotle's physics is useless to modern physics. In contrast, Newtonian physics still forms the bedrock of the curriculum. I find this a striking fact. Of course "Newtonian" does not equal "Newton's", but still, Newton's work "cleans up nicely"; Aristotle's does not.

    What are we to make of this? Most historians seem to ignore the issue entirely; Weinberg uses it as an excuse to lambaste Aristotle. Surely both approaches leave out a key aspect of the history of physics.

    1. Hi Michael,

      Similar ground, but not exactly the same ground. I think that the question of whether truth (as judged by what today's scientists believe) is a guide to rationality is independent of the questions of whether scientists are moving towards the truth and whether some of their theories are incommensurable.

      I say that historians should not assume that scientists who held true theories were more rational than their erroneous rivals. What follows? Only that there are some cases where the false theory was the more rational one to hold. Not many cases, and perhaps only a small minority -- but enough that it is worth the historian's while to check each case by studying the historical record. Having this historical scruple is compatible with believing that today's scientists are right about most things they say about nature.

      It is true that Kuhnians have an extra reason for denying that today's beliefs are a good guide to the rationality of past beliefs. The extra reason is that, according to Kuhnians, the definition of rationality -- the criteria used to assess theories -- has changed over time. But even a non-Kuhnian like Weinberg should recognise a more basic reason, namely that the evidence that bears upon theories has changed over time.

      As you point out, Newton's physics persists whereas Aristotle's does not. Does this mean that Newton's methods were better than Aristotle's? In this case perhaps, but not in general -- good methods are no guarantee against error. Does it mean that Newton had more raw intellectual talent than Aristotle? Surely not. Does it mean that Newton's theory is true whereas Aristotle's was false? I would say 'yes', if only because if we are going to assess theories at all, we should do so by today's standards; and if Newton's theory has survived that means it is better than Aristotle's by today's standards. However this conclusion -- Newton was right and Aristotle was wrong -- does not change the answers to the first two questions.

  3. This comment has been removed by a blog administrator.

  4. Weinberg's book finally became available through my library's digital network. Popular! Weinberg does not run afoul of the symmetry principle, so far as I could tell. One trio of paragraphs, early on, seems intended specifically for critics like Shapin:

    Of course, one has to try to understand the historical context of scientific discoveries. Beyond that, the task of a historian depends on what he or she is trying to accomplish. If the historian’s aim is only to re-create the past, to understand “how it actually was,” then it may not be helpful to judge a past scientist’s success by modern standards. But this sort of judgment is indispensable if what one wants is to understand how science progressed from its past to its present.

    This progress has been something objective, not just an evolution of fashion. Is it possible to doubt that Newton understood more about motion than Aristotle, or that we understand more than Newton? It never was fruitful to ask what motions are natural, or what is the purpose of this or that physical phenomenon.

    I agree with Lindberg that it would be unfair to conclude that Aristotle was stupid. My purpose here in judging the past by the standards of the present is to come to an understanding of how difficult it was for even very intelligent persons like Aristotle to learn how to learn about nature. Nothing about the practice of modern science is obvious to someone who has never seen it done.

    So there you have it. I do detect a slightly disparaging tone towards those whose "aim is only to re-create the past", but this is repaid with interest in Shapin's review. Anyway, what would one expect from a modern scientist? De gustibus...

    1. Thanks Michael for keeping this blog alive with your comments!

      The paragraphs you quote are indeed promising, especially this one: 'Nothing about the practice of modern science is obvious to someone who has never seen it done.'

      I do have two reservations, though. Firstly, Weinberg doesn't quite state the symmetric principle. Nothing in the passages you quote implies that scientists can be wrong for the right reasons or right for the wrong reasons, eg. that a precise experiment can sometimes give a worse answer to a given question than casual speculation. Secondly, does Weinberg practice what he preaches? Does he treat his historical examples with the even-handedness that he promises early on? I really should get the book myself and try to answer that question.

    2. I'll try to avoid the Scylla of a ridiculously long reply without falling prey to the Charybdis of drastic oversimplification.

      Weinberg wants to judge past scientists' successes by modern standards in aid of "understand[ing] how science progressed from its past to its present". Shapin is monumentally unsympathetic to making such evaluations, and so quotes Weinberg out of context throughout his review. I think you gave a good defense of this aspect of Weinberg's book, so let's look at the symmetry principle.

      Point 1: Weinberg gives lip service to the goal of Shapin and others to "re-create the past, to understand 'how it actually was' ", but this does not really interest him for its own sake. Now, if his primary goal is "to understand how science progressed from its past to its present", the symmetry principle may not be all that important. Take Aristotle's thoughts on motion. Weinberg's judgment: these played out almost entirely as impediments on the road to Newtonian mechanics. *Why* Aristotle took the approach that he did --- whether his reasons were good or bad --- takes second place to the *effect* of these teachings, almost completely baleful (in Weinberg's view).

      Point 2: The symmetry principle says we should not reflexively say historical figure X had bad reasons for asserting Y, just because Y turned out to be wrong. But neither should we reflexively say "symmetry principle!" every time someone casts aspersions on historical figures.

      Case in point: Descartes. He utterly botched the laws of collision. They were logically inconsistent, empirically grossly deficient, and innocent of the distinction between mass and volume. Having read Blackwell's detailed and sympathetic discussion (Isis, Vol. 57, No. 2, pp. 220-234), I agree with Weinberg that this is not first-rate work; it seems notably inferior to Descartes' contemporaries and near contemporaries.

      Shapin ridicules Weinberg for his low opinion of Descartes. He makes fun of what was clearly intended as a throwaway joke: "Finally, on the basis of observation of several lovable pet cats, I am convinced that Descartes was also wrong in saying that animals are machines without true consciousness."

      Weinberg was making a higher-level point: "His philosophy was and is much admired, especially in France and among specialists in philosophy. I find this puzzling. For someone who claimed to have found the true method for seeking reliable knowledge, it is remarkable how wrong Descartes was about so many aspects of nature." There follows a damning catalog of Descartes errors, so judged by modern standards. Since Descartes claims to have discovered the secret key to scientific truth, I think we are entitled to skepticism on *that* claim, even without a meticulous study of where Descartes blundered and where he had bad luck. (By the way, why the heck *did* he pick the pineal gland as the seat of reason?)

      If someone offers an infallible formula for making money on the stock market, I think we'd be ready to question his genius if the formula turned in a long string of losses.

      (Just to beat up on Descartes a little more: Richard Arthur has made a strong case that "a great deal of what is normally ascribed to the genius of Descartes should instead be ascribed to his erstwhile friend and mentor [Beeckman] ... one can fairly say that a
      significant part of what subsequently became known as Cartesian natural philosophy was plagiarized from Beeckman." I don't know that Weinberg was aware of Arthur's work, however.)

      Weinberg does express admiration for some of Descartes' other work: obviously his mathematics, first-rate by any standards, but also his explanation of the rainbow, where Descartes ignored his own philosophical precepts and blended hypothesis, deduction, and observation in the approved modern manner.

    3. Let me throw in a quick defense of Weinberg with regard to two other revered figures: Plato and Galileo.

      My digital copy of Weinberg has expired, so I can't quote verbatim, but he doesn't say that Plato *is* silly, just that he's *sometimes silly* (or can be silly, something like that). (In contrast to Aristotle, whom Weinberg characterizes as always serious.) This is a passing remark about Plato's *tone*, and not at the heart of Weinberg's view of Plato, which is indeed balanced. (By the way, I think the definitive take-down of Plato comes from IF Stone's book on Socrates, purely on matters of politics and integrity.)

      On some matters, Galileo *was* behind the times. Weinberg specifically mentions his disregard for the newly invented algebra; possibly also his ignoring Kepler's laws. (If Weinberg didn't mention the latter, he should have.) "Behind the times" is a gentle criticism; many see rampant egomania at work instead.

    4. Hello again Michael,

      I agree with your Point 1 and Point 2, and your most recent comment backs up Point 1 as it applies to Weinberg.

      Point 2 is close to my heart. As I put it in an earlier post, we should not confuse the Symmetry Principle with the Exclusion Principle

      Point 1 suggests an unexpected affinity between Weinberg's project and the sociology of scientific knowledge. Both focus on observable phenomena -- the interplay of theories for Weinberg and social behaviour for the sociologists. Both eschew the study of the inner workings of the human mind -- in Weinberg's case the reasoning that leads to theories and in the sociologists' case the psychological springs of social behaviour. There is a kind of behaviourism in both projects. Perhaps.

      I do have a quibble about Point 1. There is a version of the symmetry principle that applies to effects rather than to reasonableness: we should not assume that all true theories benefit later science and that all false theories harm later science. Has Weinberg made such assumptions? Does he give any examples of false theories that have led to new discoveries despite themselves, or of true theories that have set up unexpected road-blocks on the route to modern science? For example, does he point out that the successes of Newton's corpuscular theory of light delayed by a century the progress of the wave theory?

      My quibble about Point 2 is to do with how it applies to Descartes. Eighteenth-century thinkers had no trouble admiring Descartes while rejecting many of his theories. 'Descartes is to be admired always and followed occasionally', as Fontenelle put it. Voltaire and Condorcet admired Descartes even though they had abandoned his metaphysics and cosmology and epistemology for those of Newton and Locke. They admired his courage and critical spirit and his determination to tear down an old philosophy and build an entirely new one on the foundations of his own reason.

      Many of Descartes theories about nature were wrong. This shows that his method was not perfect. But perhaps it was better than anything else on offer at the time. And his philosophy had some positive long-term consequences, ones that appealed to the likes of Voltaire and Condorcet. Maybe the lesson is that though Point 1 is sound in principle it is hard to apply in practice, and that Weinberg has saddled Descartes with an overly narrow definition of 'the consequences of a theory for later science.'

  5. I've just seen the latest spate of comments on this. As a "watch this space" announcement, Joe Martin has put together a panel on the book for the American Physical Society spring meeting in Baltimore next year. Weinberg's on it, I'm on it, Gerald Holton is provisionally on it, and Jennifer Ouelette. It should be very interesting!

    So, obviously, I now have to read Weinberg's book. After reading the first part, I can say that I think Shapin's reading of the source of Weinberg's errors is incorrect. Weinberg does, in a sense, judge the past, but he is acting as a sort of a tour guide, describing the past in terms of how it differs from the present. The tone is not unlike that in C. S. Lewis's The Discarded Image. That said, Weinberg makes some clear errors, such as his claim that physics had shed religious overtones by sometime in the 18th century. The title of my talk is tentatively "Diagnosing Historical Science, Diagnosing Histories of Science."

    1. Interesting, indeed. I mean the panel and your reading of Weinberg. It strikes me that there is something odd about criticising judgements of past science on the grounds that they are presentist. Sometimes this criticism is valid, but only when the judgements are done badly. When they are are done well they do not imply that the past was like the present. Instead they imply the opposite. 'Descartes' theory of collisions was wrong' is another way of saying 'Descartes theory of collisions differed greatly from my theory of collisions.'

    2. I found this blog while looking up the Shapin review of Weinberg's book, which a colleague had suggested to me for a particular purpose (I actually wanted a cantankerous historians' reply to a scientist doing history!). One point that I think you over-looked is that Shapin's table-thumping account of what historians would say to Weinberg was actually a criticism of table-thumping, whether by Weinberg or historians. Shapin moves on from the faux table-thumping argument to suggest what is truly interesting is that Weinberg's idealizations and celebrations of science indicate the inability of historians to control their own territory. Can historians indulge in talk of anticipations? You say yes, which is fine, as John Pickstone once eloquently made the same case. But you neglected to insert the nuance Pickstone did and what, I think, Shapin is hinting at. The issue is actually illustrated by Weinberg. The problem with whig history is the abridgments (to use Butterfield's term) involved actually distort the past actors, so that when you do get to anticipations, you're comparing red and green apples when in fact the green past apples were more like oranges. You ask for someone to supply what is wrong with whig history, but one could just say 'go back to Butterfield' or even 'read John Schuster's brilliant The Scientific Revolution'. Abridgements make anticipations too easy and contrasts too strong, in effect. What any of this has to do with symmetry is a little obscure, because your definition of symmetry ran whig history and methodological symmetry together. But one would be hard pressed to say Weinberg had not violated symmetry understood as invoking social causes for error and letting truth shine by its own light. Symmetry, said Bloor, does not prohibit evaluations, just evaluations masquerading as explanations. And isn't that what Shapin is hinting at, that Weinberg's evaluations of what constitutes proper knowledge got right into how he idealized what we ought to think is real science down the ages?

    3. Hi Darrin, thanks again for the comment.

      On table-thumping: you are right that Shapin writes 'Table-thumbing isn't interesting'. But it is hard to know how to read this, since some of the views that Shapin attributes to the table-thumping historian are views that Shapin would probably accept. Consider the following view, which I mention in the above post: 'They’d express bemusement at Mr. Weinberg’s insistence that science advances by rejecting teleology, even as he depicts its history as a triumphal progress from dark past to bright present.' Based on other things Shapin has written, I find it hard to believe that he is dismissing this kind of 'bemusement' as table-thumping. On the contrary, he probably shares that bemusement.

      As you point out, Shapin goes on to say that historians do not seem able to control their own territory. Scientists feel entitled to write the history of science, whereas (to use Shapin's analogy) linesmen do not feel entitled to write the history of rugby. I address this point in the above post, where I say that Shapin's analogy between scientists and linesmen is misleading.

      Regarding anticipations, you seem to be saying that historians can talk about anticipations as long as they do it carefully. In other words, there is a risk involved in talking about anticipations (as there is a risk in *not* talking about anticipations), but we can minimise or even eliminate the risk by observing the distinction between what happened in one person's mind and what the consequences were for what happened in other people's minds.

      Is that what you are saying? If so, I agree with you. I cannot say whether Weinberg has avoided the risks of anticipations, since I have not read his book, but if you have examples from Weinberg's failings in this area I'd be glad to hear about them.

      As for Shapin's views on anticipations, I am not convinced that he is 'hinting at' the 'subtle' view that I just summarised and that you describe in your comment. Shapin wrote: 'The historians might also thump the table, insisting that searching for anticipations and foreshadowings is both wrong and illogical—”ahistorical” as they’d say.'

      The view that Shapin attributes to historians here is not subtle. It states in so many words that the search for anticipations is wrong and illogical, period. Is Shapin endorsing this view, or is he dismissing it as 'table-thumping'? It is very hard to say based on what Shapin has written. What we *can* say is that Shapin attributes this view to historians. Hence the above post, the aim of which was to point out that there are professional historians who do not share this view.