The Bactra Review: Occasional and eclectic book reviews by Cosma Shalizi   166

Why Democracies Need Science

by Harry Collins and Robert Evans

Polity Press, 2017

Common Sense Keeps Breaking Through

This book has two big themes. One is that the scientific enterprise rests on certain values, which are attractive and should be pursued, and are consonant with a democratic polity and society (though not quite the same as those of democracy). The other is that there should be a "College of Owls", staffed by sociologists of science (and a few sociologically-minded scientists) which can advise democratic governments about scientific controversies, in particular about whether there really is a serious controversy on a certain subject.

I have some thoughts, but they since articulating them into a coherent whole is too much trouble for right now, I will give the book less thought than it deserves.

Owls, Why Not?

Democratic (and other) governments have long institutionalized means of getting advice on scientific and technological questions. (In the US, this is largely but by no hardly exclusively done through the National Academy; the Office of Technology Assessment should be revived.) Expanding this to getting advice about understanding the state of scientific controversies is probably a wise development, and I am fully persuaded by Collins and Evans's arguments that this would not pre-judge what governments should do. (It is perfectly coherent to say "Yes, we understand the science says X, which would seem to imply Y, but we are going to do Z, because considerations A, B, and C are all more important.") I am less persuaded that this can really be cleanly separated from evaluating the science itself, but that may be incidental. If every report from the Academy was paired with a report from the Owls, about the state of the scientific debate, and whether any appearance of controversy is worth taking seriously, this might well be very valuable.

If Collins and Evans gave any explanation for why an un-democratic regime could not also have a College of Owls, and use it just as effectively as a democratic regime, I missed it. They do, in their last chapter, argue that the values underpinning the scientific enterprise are congruent with those underpinning (modern, liberal) democracies, but at best this just says that it can be hard to conduct scientific inquiry under non-democratic regimes, when they choose to interfere. Well, so it is, as we know in part from heroic examples of resistance on the part of scientists. But it can also be hard to conduct scientific inquiry in a democracy, when the demos chooses to interfere. Indeed, one strand of very learned opinion claims that inquiry does better under the protection of free-thinking princes than subject to the scrutiny, not to say the persecution, of the demos. (There are two ways we know that "Socrates is mortal", after all.) Now, maybe liberal democracy is a lot better than either autocracies with occasional free-thinkers in charge, or il-liberal democracy. But if one is going to write at length about the harmony between science and democracy, one should really think through these matters (as many have, for a long time).

An Old Familiar Song

It is somewhat extraordinary to write a whole book about the moral values of science, and their alignment with democracy, without mentioning the explicit testimony of many practicing scientists who were also democratic activists (e.g., Andrei Sakharov, or Fang Lizhi). It is also a bit strange for scholars steeped in English-language science studies to make a big deal about the values implicit in the scientific enterprise, and the need to choose those values, without even mentioning Karl Popper. I am sure that at some point Collins and Evans have read The Open Society and Its Enemies and The Poverty of Historicism, since they cite them, but they have evidently forgotten that these are crucial parts of both books. Still, coming to the same conclusion independently should, if anything, reinforce our confidence in the conclusion.

Trust Us, We're Experts in Why Experts Are Really Quacks

Collins and Evans are in a particularly bad position to argue that democratic polities should pay any attention to the findings of sociologists, because they have spent their entire (long, successful) careers as sociologists attempting to undermine the idea that science gives us reliable information about the world. Thus they are at pains to argue that science's values are good ones even if science doesn't work [1]. Their defense of the those values comes down appeals to incredulity; paraphrasing only a little, "How can you possibly prefer the views of those who haven't even tried to carefully look into a matter to those who have looked?" And when you put it like that, it does sound bad to me.

[I]f one wants to know about some feature of the world, does one prefer to listen to the opinion of one who has observed that feature of the world or one who has not observed that feature of the world? ... name="breliability">[2]

[T]heory, or any other kind of a priori belief, cannot generally be preferred to observation in matters of observation. Or, if theory, or prophecy, or soothsaying is preferred to observation, the onus is on those who prefer it to explain why. Should it be that they believe that the whole world is an illusion, or that the word of the deity as located in old books is to be preferred to observation then, at least, the choice will be clear and it can be stated. There will be nothing left but a choice.

For elective modernism [i.e., for Collins and Evans], then, logical positivism [sic; read "empiricism"???] lives on as a basis for our aspirations for how to understand the observable world. And, of course, it always did. We have always preferred observations over non-observations where it was the observable that was at stake; we have just forgotten how to say it out loud. (pp. 42--43)

But the same arguments would work equally well for other values --- "How can you possibly prefer the views of those who have disdained the voice of the gods, as reveled by oracular sacrifices, over the opinions of pious people who pay due attention to the oracles? We have always respected the oracles within the domain of the oracular, we have just forgotten how to say it out loud." (I have chosen an example which is not a live option for anyone to make the point; readers may substitute their own least-favored counter-empirical source of opinion as they like.) This sort of exhortation would work just as well on behalf of haruspicy as of empiricism, unless organized-actually-looking-to-see is at least somewhat more informative than haruspicy. As my quotation indicates, however, Collins and Evans desperately want to avoid any such claim; rather, they want the reader to make a choice, explicitly without any justification.

Now, I actually think a case could be made for a position something like this. It's what the late great anthropologist-philosopher Ernest Gellner sometimes called believing that "positivism is right, for Hegelian reasons". Namely: world-views, and the ways of life which go along with them, have to be taken as more or less total packages. ("Fundamental changes transform identities": Plough, Sword, and Book, p. 194.) Precisely for this reason, attempts to find neutral criteria which could truly rationally compel someone to accept one such package over another have at best foundered on circularity. But some of those packages are a damn sight more attractive than others. The package which Gellner sometimes abbreviated as "positivism", where concepts are submitted to empirical control, where description and prescription are carefully separated, etc., etc. --- in short, the scientific world-view --- has proven to be an immensely attractive one, because it works in the world. It has transformed the ecology of humanity, and is (as he put it somewhere) visibly devouring every other form of society. At this level, we may have no option beyond a total choice without rational justification. [3]

It does not follow that we should accord deference to every enterprise which purports to be part of the larger scientific world-view, particularly when it is very shaky about things like whether its concepts actually refer to anything in experience. This is, however, exactly what Collins and Evans want, by saying that their approach will "defend science [even] when it is ineffectual". they illustrate this with an extended discussion of why policy-makers should continue to rely on the forecasts of macroeconomists, despite their dismal empirical track-record (pp. 56--60). They admit that this doesn't say why we should pay more attention to econometricians than to astrologers (though Elliot wave theorists might have been more to the point). But the idea that, for example, if reliable macroeconomic prediction is unavailable, maybe we should try to design policies that don't rely on predictions does not cross their horizons.

There may, however, be a more direct reason why our authors want to defend science even when it doesn't work. After all, they want democratic polities to pay a lot of attention to sociologists when it comes to controversies about science and technology. But if what Collins and Evans have been saying for decades about how science is completely socially driven and detached from the real world is right, why should the sociology of science be miraculously exempt? Why, then, should politicians and voters pay any attention to its advice?

The authors are somewhat aware of the self-undermining dangers of advocating the position parodied by Larry Laudan as "skepticism about everything except the social sciences". There are two passages where they actually come close to en[I]t is important to distinguish between a descriptive 'is' and a prescriptive 'ought'. A well-known example in the [science and technology studies] literature --- Shapin's study of the controversy surrounding phrenology in nineteenth-century Edinburgh --- makes the point. Shapin shows that the outcome, in which phrenology was effectively vanquished and proponents of the status quo retained their positions of influence and authority in the academic world, was closely linked to, and, in part, explained by, their connections and influence in the wider cultural and political life of Edibnburgh society. This is the 'is' of the matter. The question is what follows from this description? One could argue that the scientific debate would have been concluded more quickly and efficiently if the influence of the various social and political factors had been brought immediately to centre-stage. But we argue that this is incompatible with the idea of science. We argue that, despite the inevitable influence of local political factors in the technical debate about phrenology, the guiding principle should be to eliminate these effects so far as possible. One conclusion that can too easily [4]

Having come to the brink of grappling with the crucial issue, they then veer off (into asking "What is science?", with obligatory Wittgenstein-on-family-resemblances to avoid giving an answer). I honestly don't think they ever give a better answer to this fundamental question than "self-evidence", or, (paraphrasing) "it would be pointless to keep doing the activity if it were just politics".

The other place where they come close to dealing with the issue is this passage:

[T]o preserve science as a distinctive form of life, scientists have to ignore, in a determined way, what the reflect analysts of science say. One cannot do good science without disbelieving social constructivism. Individual scientists have to believe they are seeking the truth and that there is a chance of finding it, even while social scientists insist it is the social group that ultimately determines what counts. Furthermore, scientists must ignore the social constructivists if the formative aspirations of science on which this entire thesis turns are to be robust: if it is all social construction, why act with scientific integrity, rather than go straight for some political goal? As we said earlier, we desparately need to preserve the moral imperative that guided science under Wave One [i.e., before the current wave of science studies], and whatever drives it.
Still more confusingly, then, this need to preserve an old-fashioned view of science is just as important for scientific social science as it is for natural science. To carry out the social analysis of science with energy and integrity requires that the analyst must often ignore his or her own findings in the very act of creating them. When the social scientist is analysing science as socially constructed, he or she has to believe that the truth about social life is being discovered, not that they are putting forward an interpretation for group acceptance or rejection. (pp. 76--77)
They go on to counsel the social scientist to learn to "compartmentalize", though "dissociate" seems closer to the mark.

As this last passage makes clear, they do not consider either that (i) if an analysis undermines itself (or seems to do so), might be because there's a mistake in it somewhere, or (ii) there is honor in giving up an activity once you've realized it's all fraud and bullshit.

Selective Certainty

As the just-quoted passages show, Collins and Evans are pretty categorical in telling the reader what science studies has found out about science, when what they really mean is something more like "as some of those who study science have claimed, to vigorous objections from others who also study science". (Said differently, people like Philip Kitcher, Deborah Mayo, Larry Laudan, Naomi Oreskes, etc., do not get to count as "reflective analysts of science".) For that matter, their statements about general sociology are equally dogmatic --- "forms of life", identified with "collective tacit knowledge", are simply stated to be the ultimate, (currently) un-analysable "basic constituents of the social universe" (p. 67). Needless to say, however, this mixture of Wittgenstein and Durkheim is not exactly without challengers within sociology (cf. Turner's Social Theory of Practices), nor are analytical sociology and cultural epidemiology contradictions in terms.

I admit that I find this sort of thing more vexing that I should, because these flourishes are so un-necessary to the main thrust of their practical arguments.

Kvetching at the Returned Prodigals

To be very clear, if Collins and Evans now want to urge, contra "conclusions[s] that can too easily be drawn" from their own earlier work (and that of their friends and allies), that there is really value in the scientific enterprise, that it shouldn't be just all politics and can be something else, well, welcome! I think a bunch of their specific ideas are good, and their hearts are obviously in the right place. Having made their careers on ideas which they now see can veer dangerously out of control (to mention two cases they [don't: the expose of the Merchants of Doubt, and the consulting career of Steve Fuller), they are trying to course-correct in the direction of common sense. In so doing, they can't bring themselves to want to jettison the ideas that made their writings exciting --- any intellectual should be able to sympathize with that. Focusing all my attention, and so yours, on their all-too-human foibles is, as Krugman once put it, quarreling with allies while Sauron sends forth his forces from Mordor.

I wish I had a synthesis of all these considerations to put here by way of a conclusion, but I don't.

[1]: They never explain how they came up with exactly their list of values, nor what grounds they would give to convince someone who came up with a different list (or how different another list of values would have to be to actually disagree with theirs). ^

[2]: In the part I've omitted from this quotation, Collins and Evans say that a preference for observation over theory, a priori belief, etc. cannot be justified by observation always being right, because sometimes particular reports of observations turn out to be just erroneous. This is true, but they completely miss the possibility that we might prefer observation because it is generally overwhelmingly more reliableTheory and Reality: An Introduction to the Philosophy of Science, to name just one place.) In other words, for them, because it's not perfect, it can't be better. ^

[3]: These are themes Gellner returned to in many of his books, including, perhaps, his best work, Plough, Sword, and Book (see pp. 193ff). The most extensive discussion is however The Legitimation of Belief. ^

[4]: Since Collins and Evans also say that the "Second Wave" of science studies comprehensively destroyed the fact-value distinction (pp. 17, 70), that distinction was evidently not their own distinction between "descriptive 'is'" and "prescriptive 'ought'", as made in this passage. What that objectionable and exploded something else was, though, they do not say. --- It is also worth noting, in passing, that they treat Shapin's account of phrenology in Edbinburgh as unproblematically correct and realistic, rather than subjecting it, symmetrically, to the same sort of undermining explanations that Shapin, like Collins and Evans, applies to scientists. ^

200 pp., bibliography, index, a few conceptual diagrams and tables
In print as a paperback, ISBN 978-1-50950961-4

Philosophy of Science / Sociology

Drafted 25 June 2017, revised 11 December 2017