The Sociology of Science for Non-Sociologists
1. The Principle
Sociology often evokes hostile responses. Sometimes this is for quite understandable reasons, and sometimes because there is a gulf between what the sociologist thinks he or she is doing and what non-sociologists think is going on. I want to try and explain some of this. I start with an explanation of what sociology is. This is a personal view and I could not guarantee that all sociologists would agree with me.
Sociologists like me try to do sociology by taking `the social' as the fundamental unit of explanation (for those who know the field, this is closer to the view of Emile Durkheim than any other classical sociologist). Think of it this way: physics takes as its fundamental unit of explanation the smallest things there are; chemistry takes electrons and their orbits, or maybe things a bit smaller; biochemistry stops at large molecules or maybe chemical messengers and electrical signals; biology generally is not interested in much below the level of the cell; anatomy tends to deal with whole organs; and psychology deals with complete people. Sociology, however, takes as it fundamental unit the rather nebulous notion of the `social collectivity.'
Both sociology and psychology deal with the causes of peoples' actions and this often leads to confusion, not least among sociologists and psychologists themselves. The easiest way to see the difference is to compare sociology with the in-between subject - social psychology. Social psychology deals with the interaction of individuals so it looks like a very near neighbour of sociology. But sociology is the inverse of social psychology: while social psychology treats social groups as the sum of the interactions of the individuals within them, sociology treats individuals as the sum of the social groups in which they have grown up and live.
One can see the difference between the one approach and the other by thinking about a natural language such as English. (By natural language I mean that which is spoken by native speakers in their everyday lives.) Natural languages and their speakers are social collectivities in my sense. The social psychology approach would be to try to understand the English language by reducing it to the contributions of all speakers of English and proto-English that there have ever been. (I don't think anyone is actually interested in doing this, by the way, I am just talking principles.) The sociological approach would take any English speaker as a representative of the dialect of English that he or she speaks. Note that you can be a `representative' without being statistically representative; from listening to a single English speaker you can learn nearly all there is to know about the accent, intonation, vocabulary, and so forth that belong to that dialect. (In case you are worried about having stumbled on an unlucky case, you can listen to a half dozen instead of one, but you still don't need a statistically representative sample to learn almost everything).
It is also the case that if you and I want to understand why we speak the English we speak we would look at the social groups in which we have learned to speak rather than the individuals that contributed to the make up of those social groups. Even if we wanted to go into the deep history of the way of speaking of the social groups from which we draw our way of speaking we would look at the contributions to them of other social groups, not particular individuals. And even if we had learned our accent from elocution lessons (as in Shaw's `Pygmalion', or `My Fair Lady'), we would still need an explanation in terms of social groups for why one particular accent rather than another was thought to be the desirable one to learn. Voice impersonation and stage mimicry are interesting because they are so exceptional (like confidence tricksters). Finally, occasionally the dictionary tells us that the origin of particular words are specific works of literature, but even then we have to explain why only some neologisms enter public discourse while others do not and this is a group process (cf Orwell's 1984).
The determined `reductionist' might still want to argue, on principle, that English must be the product of all the individuals who have ever opened their mouths and uttered words in the presence of other English and pre-English speakers. But (a) even if all of this could be recovered in philosophical principle (and there are arguments to say it could not), it could never be recovered in fact (and don't forget about chaos); (b) to try to recover it is about as much use to those who want to understand the English language they speak as trying to understand matter at the sub-atomic particle level is to the anatomist or the weather forecaster.
Anyway, whether you agree with what I say about English or not, this kind of idea is what leads the sociologist to think, not about social groups made up of individuals, but of individuals made up of social groups (and therefore representing social groups).
2. The Micro-Politics
Mostly we don't experience life as a series of group pressures, we experience it as a series of wise or stupid, brilliant or dumb, and good or evil acts. There is very often a tension between explaining an act as at the level of the group and explaining it at the level of the individual. This tension comes up over and over again.
For example, discussions of crime and punishment are torn between the idea that the criminal is a product of his or her social group and the idea that a criminal act is a matter of personal responsibility; the political wind blows back and forth and different groups try to make different kinds of political capital out of it. Thus, the US National Rifle Association defends Americans' right to carry guns by refusing to countenance anything but personal responsibility for an act. They say: `Guns don't kill people; people kill people'. And they are right in a number of senses. They are right in a moral sense; each murderer is still a murderer. They are right in a statistical sense; hardly any gun-owner ever kills anyone. It looks, then, as though explaining firearm-related homicides is just a matter of explaining the actions of a few individuals. What is still puzzling, however, is why the British homicide rate (homicides by young men are the crucial indicator), is much lower than that in the USA. Could it be that almost no one has a gun in Britain or is it just that American young men are more murderous?
The political disagreement over the level of explanation is often mistaken for a scientific disagreement. Looking at the cartoon one can see that there is no scientific clash between the two kinds of explanation of homicides with firearms, all the tensions are political and moral. In this case there can be little doubt that the individual who kills is blameworthy because the act of shooting was a proximal cause of the death. A distal cause, nevertheless, was the distribution of firearms, for which, one might argue, the political system was responsible. A complete scientific explanation, if there were such a thing, would need to study both distal and proximal causes.
This example is over-simplifies because there are other kinds of cause operating in this case. The sociologist would be still more interested in the fact that there are lots of guns in Switzerland but a low murder rate so there must be cultural factors as well as material factors operating distally. The gun analogy works well for diffusing the supposed scientific tension between distal and proximate causes and for stressing that sociology, as I have described it, is almost exclusively interested in distal not proximal causes even where a proximal cause is operating. I'm not a great believer in `complete' explanations - I think different kinds of scientist should concentrate on their speciality, at least where the science is difficult and untidy, as in the case of sociology. The synthesis of all these different ideas about cause should happen, I believe, at the social level, where different sciences interact.
3. The Sociological Study of Science
The two cases we have looked at - natural languages and the firearms debate - are different in that in the matter of firearms one can see that both kinds of cause work together whereas in the natural language case there is almost nothing you can do with proximal causes. The case of the analysis of science is different again in that in some peoples' minds the two kinds of cause are locked in a scientific battle - competing for being the one true type of cause. The science wars could be said to be a fight about the causes of scientific beliefs (what scientists take to have been proven). To caricature, on one side are those who think that scientific beliefs are to be explained as the outcome of passages of theory and experiment carried out by individuals with other scientists merely checking results for accuracy; on the other side are those who think that collective scientific beliefs are just as much the outcome of social processes as the English language, with individuals' contributions, like the neologisms of a literary author, being taken up or not according to whether they fit with the current lurch of scientific thinking.
The end of the science wars will almost certainly be a matter of continuing disagreements among a range of compromise positions which are closer to each other than they were at the start (some of which will be represented in a book edited by Jay Labinger and myself, called The One Culture?, which will be published by University of Chicago Press in spring or summer of 2001). But whatever the outcome I still think that the social scientist should concentrate on one kind of cause at a time according to where his or her expertise lies. As a sociologist I concentrate on social causes and leave the psychology of science or the science itself to others (some people think these approaches are at war - as implied, I think they are just different ways of looking at the world). This implies, among other things, the principle of methodological relativism.
Methodological relativism, which is discussed elsewhere on this website as well as in the book mentioned in the last paragraph, means ignoring scientific truth and falsity. For the sociologist to do otherwise would be to set him or herself up as someone who already knew the answer to a continuing scientific debate. (The historian must act as though the debates he or she studies have not been resolved - for a discussion of types of history of science see the Afterword to the second edition of the first volume of The Golem series.) The sociologist, then, does not take scientific sides so long as there is any significant group of scientists still arguing the matter. Often, the scientific truth seems so obvious, if different, to scientists on opposite sides of the debate, that the sociologist appears to both sides to be missing the whole point.
As well as the scientific debate about the proper explanation of scientific beliefs, the political and moral tension between our experience of life as a series of wise, stupid, brilliant, dumb, good and evil acts, as opposed to a set of structural and cultural pressures, is just as marked in science as anywhere else. The Nobel Prize and the other marks of scientific success are not given for being the subject of social pressures or finding yourself in the right social milieu, they are given for genius, brilliance, and their lesser variants. And if you lose your job, or fail to get tenure, it is not because social events were unkind to you, it is because you didn't publish enough, or contribute enough to the team, or whatever. And that is how all academics, including sociologists, experience our lives.
The sociologist does not want to change this any more than the structural analyst wants to take blame away from the person who fired the gun. As in the NRA debate, the kinds of cause discussed do not have to compete. It is just that the sociologist is not professionally interested in who pulled the trigger and why, and its scientific equivalent, which individual did well and which individual did badly in the scientific race and why. This can be hard for the scientist to see because matters of individual invention, discovery, leadership, and, of course, failure, are the obsessive stuff of day to day life in science (and the rest of academe), and it can seem that a high level analysis that does not deal with proximal causes is an insult to science's amour propre as well as scientists' humanity.
Psychologists have it much easier than sociologists because the things they study are walking around, just about the same size as they are, neatly bounded by envelopes of skin, and the causal world psychologists deal with is a match for the causal world experienced by the professional. The sociologist is always struggling to find a higher level explanation when what happens in front of one's eyes seems to happen for very obvious local reasons following directly from the actions of individuals - and that's how respondents see it too. You can only have any hope of finding a bigger picture by approaching things in a determined frame of mind and ignoring what interests many respondents. Maintaining the right frame of mind takes you away from the day to day world of the scientist even as you are trying, for methodological reasons, to get into it. In paper 4 I said:
One of my failings ... is that I cannot generate the levels of disdain for physicists that one physicist can generate for another; I know what I am missing because ... in my own field I can readily generate such levels of disdain for some of my fellow social scientists! (p 299)
In analysing science, not only am I not interested in generating this kind of disdain, I have actively to avoid it because ordinary assessments of worth conceal the bigger kinds of causes I am after: the very visible trees conceal the shape of the wood.
One can see the approach exemplified in paper 4. When I discuss the disagreement between the Louisiana group and the Frascati group I am not interested in who was right and who was wrong, but in the sets of pressures and forces, institutional and cultural, which provided the opportunity and potential for them to act in their different ways. In each case they might have resisted those forces and acted differently, and therefore the way they chose to act was, in the end, their responsibility, and if they had been different personalities they might have acted differently. But my job is to exploit the situation for what I can get out of it in the way of understanding distal causes, not to explain the specific actions. I tried to make the point clear in the very last paragraph of paper 4:
I have described the relationship between the evidential cultures of the two groups, their view of the heavens, and their institutional positions, as a homology. I choose that word because it would be quite wrong to draw crude causal connections between the structures and pressures within the respective environments and the scientific choices made by the individuals. The history of scientific heroism is the history of scientists resisting institutional pressures, and there is no reason to suppose that the principal actors who speak from these pages are not just as capable of resisting external pressure as a their famous forebears. Of course, given that every scientist discussed here would readily concede that, in the last resort, blinding is the only sure way to eliminate unconscious bias in data analysis, they would also have to accept the possibility that structural forces might affect their larger judgments in subtle and invisible ways. It would be very difficult to turn this `might' into anything definite, as the complex workings of law courts reveal. Fortunately, we do not have to prove the matter one way or another for any individual scientist, we have only to describe the pressures and the plausible patterns of action within the scientific community. (p 336)
There are a few methods which help one in trying to understand structural and cultural patterns. One of the best is to look for areas of social strain. Like a geological fault, a social `fault' exposes the normally hidden `strata' which make up social collectivities. There are some famous `experiments' which indicate the underlying idea. In a `breaching experiment' you try to understand social rules by breaking them. For example, Harold Garfinkel, who invented the idea in the 1960s in UCLA, persuaded his students to go home to their parents house and act with extreme good manners and politeness as though they guests rather than sons and daughters; breaking normal expectations, even in this ultimately harmless way, soon engenders anger. My favourite of these breaching experiments was done by a colleague from Manchester University who asked his students to demand two tickets when they took a bus ride, one for themselves and one to reserve the seat next to them. And you can try such an experiment yourself next time you go to the supermarket. Ask yourself who owns the goods taken from the supermarket shelves between the time they are in the trolley and the time they are paid for. You can find out by taking goods out of others' trolleys or from the pile of items preceding your pile on the checkout belt. On that evidence it looks as though whoever puts something in their trolley `owns' it, at least in one sense. On the other hand, you can just about get away with picking up an item from the belt in front of you and looking at it so long as you replace it. And if you drop and break an item while walking round it turns out you didn't own it because you can get a replacement from the shelf for nothing - and so on.
A breakdown in social relationships, or a scientific dispute, is a naturally occurring breaching experiment and that is why these things are so interesting to sociologists. It is, perhaps, the concentration on breakdowns, disagreements, and other stresses and strains that makes it appear that the sociologist is mainly interested in gossip and trouble rather than hard evidence and success. But this concentration is there because it is the best way to get at the bigger picture - the things that guide our actions in ways that we never notice until things go wrong. This is why I began my research by looking at scientific disputes.
Likewise, in the work I am writing up on the changing management regimes of LIGO, I am interested in understanding the meaning of the change from small science to big science, and I can understand this best, I believe, by looking at the strains it engendered. For the majority of scientists who worked in LIGO there were few strains and the transitions were a troublefree experience. That, however, does not tell me much (just as the experience of the majority of non-killers among the gun-owners of America tells me little about the opportunities and potentials that result from widespread gun ownership). If I want to reach out toward the larger picture I can do it better by looking at those who did experience the strain of change even if they are a small minority. And, to repeat, one does not look at them to try to work out why they were different to all the others - one looks at them to give some construction lines to guide the painting of the larger picture.
Another method (of which I think I am the only exponent), for exploring social collectivities is to look at non-social creatures' interactions with societies. I have done work on intelligent machines because they are, in my sense, non-social entities. If, for example, you compare the native speaking and understanding abilities of a computer and a human you see the difference between something that has learned its language through being part of a social collectivity and something that has `learned' it another way (see, for example, my book Artificial Experts). This, by the way, has nothing to do with hardware or wet-ware as the example of the domestic dog shows. A dog has more wetware than the most optimistic computer designer dreams of and plenty of opportunity to learn its owner's natural language but it still can't do it. Thus dogs are not members of our social collectivity. Chimps and dolphins might be - time will tell. (One can argue about whether dogs, ants, bees and networks of computers have their own self-contained social lives in the sense I am using the term - I don't think they have but I don't provide the arguments here).
Another way of tapping into social collectivities is looking for boundaries between them. One does this by looking for differences in what people take for granted when they are embedded in different social groups. Differences between religous groups are one example, and the differences between Louisianian and Frascatian gravity wave scientists that I explore in paper 4 are another. What Thomas Kuhn was saying in his famous book The Structure of Scientific Revolutions is that major scientific change is not just a formal or technical matter it is a change from one kind of social collectivity to another. `Paradigm incommensurability' to use Kuhn's term, was the kind of misunderstanding that takes place across the boundaries of social groups; what looks like a scientific disagreement is more like a social disagreement. This is still the best way to understand many deep scientific disputes and it makes it quite easy to see why they are so difficult to settle. (And it also explains what Max Planck said about science: that new ideas triumph not by convincing opponents but by the death of those who hold the old ideas and their replacement by a new generation.)
The meaning of lies
If you spend your time looking for these kinds of boundaries it is again pretty irritating to those involved in the dispute. The Kuhnian idea as I have expressed it is not much use to working scientists because it is the job of scientists to get things right. That means proving the point to those who cannot see it. And that, of course, is just how I act when I am in a dispute with my social science colleagues. But it is not how I think when I look at scientific disputes; in my fieldwork my first concern is to rediscover what one might call the `socio-logic' of each of the separate parties and then try to show how each party can be seen to be acting reasonably within their particular collective way of seeing the world. As I put it in paper 4:
... it is scientists' responsibility to make science and, therefore, to make the nonsense with which the sense can be contrasted; it is not sociologists' responsibility to make either. (p 297)
That is to say, it is the job of scientists to show that their scientific opponents are talking nonsense or acting foolishly wherever there is a disagreement whereas it is the [first] job of the sociologist to [try to] show how each party makes sense in their own terms. This is likely to be irritating to the insider. Thus, normally, each party to a scientific dispute sincerely holds that their view is right and would be seen to be right by any competent scientist. To continue with our firearm motif, the parties to the dispute line up opposite each other and fire away with scientific arguments and whatever else is in the armoury. The sociologist, meantime, is busy trying to make sense of everyone's position and so is likely to find him or herself somewhere in the middle of the disputing parties, drawing the fire of all!
It would be all too easy for the sociologist to side with the dominant and powerful view, whether in matters of science or organisation, but the result, though it might look more and more wise as history unfolds, and would bring accolades from the scientific victors, would be of negligible sociological value.
Perhaps the most disturbing thing of all is that the sociologist is more interested in big social patterns than in getting at `the truth.' Think of it this way. Suppose you are a stranger to Western culture and you are watching a ballroom dancing competition in Birmingham (England). You see two people dancing the tango in a very intimate and sexy way (though you don't know it is a tango).
You turn to your `informant' and ask what is going on. We imagine that the informant might give three possible answers:
(A) They are trying to win a medal in a dancing competition.
(B) They look like they're dancing but really they are enjoying some surreptitious sexual titillation.
(C) They are dancing to bring on rain.
If we know only a little about the British way of life we know that C is not true, and we know that if the informant is being reasonable, C would not have been offered as an explanation. This leaves us with A and B. If we were a `private detective' employed by the jealous partner of one of the dancers it would be vital to find out whether it was A or B that was true. On the other hand, if we were just trying to find out about social life in the West Midlands of the UK, either A or B are equally informative because they express two reasonable possibilities within life as it is lived in that area. I wrote a paper about this called `The Meaning of Lies' in which I tried to show that for the sociologist interested in the way social life is lived a `good lie' (as opposed to a joke or a piece of nonsense), is just as useful a piece of data as the truth, because a `good lie' has to be plausible. Just as widespread ownership of guns in the USA provides the opportunity for shootings, the social collectivity of Birmingham, England, provides the opportunity for ballroom dancing and surreptitious sex but not for rainmaking. As I said above the sociologist, in trying to investigate forms of life, is interested primarily in the possibilities for action that they provide not the specific actions that were carried out (see the passage from page 336 of paper 4 quoted above).
To save any misunderstanding, let me add that the one or two occasions when I have known that respondents were deliberately lying to me I have felt that both parties were dirtied by the interchange. The real point is that the sociologist needs to avoid trying to fulfill the role of policemen, lawyers, prize-committees, writers of scientific review papers, and all those others who are responsible for doing the judging as scientific history unfolds. It is not a matter of valuing lies - no one values them - it is a matter of not being preoccupied with uncovering or establishing `the truth.'
Let me spell this out with reference to a particular instance drawn from the history of LIGO. In the mid-1990s the new management team fell out with some of the members of the group working on the 40 meter interferometer at Caltech. Eventually this disaffected group left. In talking to both sides in this dispute I have collected a variety of accounts of what happened, one, some, or all of which could be true. They include:
(A) The new management team wanted scientists to work to industrial style schedules whereas they felt creative work should dictate its own pace.
(B) The disaffected group were incapable of adapting to the demands of what was becoming a large complex project.
(C) The disaffected group were primarily concerned with the long term goal of establishing gravitational wave astronomy whereas the new management team needed short term results quickly.
The first thing to notice about this list is that it does not contain anything equivalent to `they were trying to bring on rain;' all three explanations are plausible. If one is trying to understand, not the specific argument between the new management and the disaffected group, but something structural, like the change from small science to big science, each of the three possibilities are informative. We can learn a lot about this kind of structural change in general by exploring all three explanations. Maybe the members of the groups experienced the problem in different ways so that there is no `truth of the matter;’ that is, A, B, and C, may be each partially true, completely and uniquely true for some actors and not for others, or even completely untrue for some actors, but it does not really matter. Let us concentrate on `C’ and imagine it was partly true for some actors and not at all true for others. It is still the case that a conflict over short term and long term goals could have been experienced by anyone in this change from small to big science - it is something that is quite possible in the way that someone objecting to the colour of vacuum tanks or the style of vegetation round the front door (each the equivalent of the rain dance) is not. And if one was trying to anticipate what might go wrong in similar change in the future, the potential clash between different peoples’ perception of the value of short term and long term goals is something you would want to take into account. This would be far more interesting, certainly for the sociologist, and I think for the scientist anticipating a big project, than a discussion of awkward personalities.
Thus we circle back to the main point: it is not interesting to the sociologist to provide a law-court-like judgment of who was right and who was wrong; what is interesting is to understand the potential for action that was provided by changes and structures at a much higher level - particularly those that might repeat themselves in other cases. And one gets to this by understanding how the various respondents view their social situation (and one can illustrate it with their talk).
Collectivities and Individuals
Notice that if you look for boundaries between collectivities by looking at the differences between what people take for granted on either side, you end up talking to individuals. One irony of sociological method is that most of what you do when you are trying to look for those big vague, collectivities, is talk to isolated human beings. This is another reason that sociology of this kind, even though it is directed at high level structural causes, looks like gossip-mongering or journalism. Indeed, the method is essentially journalism, and would be journalism, if it were not that an entirely different kind of project lies at the heart of it. The journalist answers to the reader, or to what is sensational, or, if it is investigative journalism, to the kind of truth that law courts (and Nobel Prize committees) are interested in; also the journalist is likely to provide an account that will match the account of history. But in the case of sociology of the kind discussed here, the quotations from individuals are selected to illustrate the structural argument being discussed. It is rather as though one quoted typical pieces of natural English speech from Leeds and from Brooklyn not for their content, and not to find out who had the `truest' accent, but to show that there was a boundary between Yorkshire and New York within the overall English speaking collectivity.
The other difference between sociology and journalism is that the former is a science - albeit a rather untidy one - and it tries to find generalities that can be applied to other similar cases and episodes; journalists have a licence to take the unique character of the episode being studied as their prime concern. That is why sociology is less snappy and clear than journalism, tends not to tell quite such a good story, and can be long drawn out and slow to emerge. In my less self-confident moments I sometimes bemoan the fact that a good journalist could find out 90% of what I find out in about 10% of the time. But still the end product is usually quite different.
A third method of trying to understand collectivities is to try to get inside them yourself. This method is known as `participant observation,' or more accurately, `participant comprehension.' I wrote a paper about this when I did a participatory study of the science of child `spoon-benders' (Collins, H. M., 1984. `Concepts and Methods of Participatory Fieldwork', in Bell, C., & Roberts, H. eds., Social Researching, Henley-on-Thames: Routledge, 54-69.) For the sociologist of science, participant comprehension is a tough method because to get inside the way of life of your respondents it looks as though you need to get yourself up to research front competence in the science in question. Fortunately, and I offer this as an empirical observation backed up by no theory or philosophy, you don't.
The empirical evidence on which this claim is based is firstly my own case studies and their reception, and secondly the experience of others doing similar work. From my own work I draw on three of my studies. The work on spoon-bending children put me in a position to become a fully accredited scientist in the area. We designed an experiment in which we sat children in a psychology laboratory faced by `dummy' observers and filmed by `dummy' cameras while we watched and filmed them from the other side of one way mirrors. We developed various other technical `wrinkles' that made it very difficult to cheat and our set-up became, for while, the standard for this kind of testing. We also wrote up an experiment in which we found 5 out 6 children who claimed to be able to bend spoons paranormally could convince the observers with them in the laboratory even while we filmed them palpably cheating from our vantage point behind the one way mirror. That little experiment was published in the correspondence columns of Nature and I briefly became an international celebrity (for doing what I thought was a trivial piece of nonsense that just happened to say what all the sceptics wanted to hear). So that is my example of complete participation in the form of life I was trying to investigate.
A little before I had tried to do a sociological study of scientists investigating the theory of amorphous semiconductors. On this project I did thirteen interviews before accepting that I could understand so little of what was being said that I could not even tell who was agreeing and who was disagreeing with who, and I gave it up. So that's my example of a complete failure to comprehend the science enough to do the sociology.
My third example is my study of gravitational waves in which my scientific understanding lies somewhere in between the two cases already described. My shortfalls in scientific ability are obvious - I can't do any of the maths, I couldn't possibly publish in the field, and there are areas which are conceptually beyond me (at least for the time being) - for example the difference between signal recycling and RSE. And, of course, the theory of astrophysical sources is beyond me and I am happy to be able to leave that to my colleague Dan Kennefick (see his paper on star crushing) who has a PhD in theoretical physics and continues to publish in the area. But my experience is that my ability to do the sociology of those bits of the field which I do write about is pretty well on a par with that in the spoon bending study. That is to say, it does not seem to me that the sociological and historical mistakes that I will undoubtedly make will have anything much to do with my lack of ability to understand the science. And in my defence let me point out that the fiercest critic of my gravitational wave studies - Alan Franklin, the physicist turned historian/philosopher at Boulder physics department - has agreed that though he thinks I have a great deal of the sociology wrong he does not think any of that was due to lack of understanding of the science (for discussion of the debate with Franklin see this website under science wars). These experiences suggest, and I stress again that they are simply that - experiences - that for the sociologist of scientific knowledge the potential for doing sociology has a very steep increase once a certain level of scientific level of expertise has been gained and there is not a great deal to be done in the way of improving your sociological abilities by learning more science once the step has been passed.
Of course, all this applies only if combined with the principle of methodological relativism.
In spite of this empirical discovery, it is always good to learn more science because it gives one a wider range of potential sociological subjects - eg one might move into the sociology of the theoretical aspects of the topic if one could master the maths - but, to repeat, the sociology of the rest of the physics does not depend on learning the maths. (Unsurprising perhaps, since, as I have recently learned, even some physicists do quite well at the physics without being very good at the maths.)
Another piece of evidence which touches on the level of scientific understanding needed for sociological understanding comes from looking at the range of people who study the sociology of science. At one end there are scholars such as Andrew Pickering and Dan Kennefick, who either were, or are, practising and publishing physicists, and at the other end people such as myself who can claim little more than a basic level of scientific literacy. Yet it does not seem that we can map the sociological competence of these different analysts on to their scientific competence in any systematic way; neither the professionals themselves nor their critics would make such a claim.
Some critics, of course, are likely to associate what they see as sociological failings with lack of scientific understanding when the output is read as making scientific judgments. This kind of mistake can be made in perfectly good faith; the critic feels sure that if only the social analyst knew more science they would see the world in a way that more nearly corresponded with their own view. Thus I have been accused of failing to know enough science to understand that Joseph Weber's new theory of the cross-section of gravitational wave detectors was correct as well as not knowing enough science to realise that it was wrong. Faced with scientists disagreeing with one another the social analyst would be very unwise to claim to know enough science to make a decision, for this would be to claim to know more science than at least one of the parties. Thus the analyst is bound to be in the position of appearing to know too little science to know what's right scientifically. To repeat a point made earlier, however, and as I have said, social analysts who have a highly developed understanding of the relationship between their bread and their butter (ie, how best to earn favour among the powerful), can avoid problems by adjusting their apparent understanding of science appropriately. For the rest of us, the right course of action is given by the principle of methodological relativism.
Reading a paper
Something I have learned from discussion with my scientist colleagues in the science wars, and in this particular case from David Mermin, is that a deep source of misunderstanding can be the nature of the output from work of this kind. Sociology is always tentative, clumsy and lacking in exactness and one should not look in it for the kind of clarity or purity one finds in a mathematical subject. In the exact physical sciences one can judge the quality of a published paper by finding a single mistake. A mathematical proof is like a house of cards and it collapses if even a single card is removed. A sociological paper, on the other hand, though it is never going to be as beautiful and sharp a structure as a house of cards, can survive the loss of quite a few bits and pieces. It is more like a rough heap of sand and the hard thing is to get it into any kind of shape at all.
Of course, not all natural sciences achieve the clarity and beauty of a house of cards. There are natural sciences that deal in heaps of sand as well. There is long term weather forecasting, there are aspects of cosmology and astronomy, and there is anything to do with the large scale history of our planet and the future of its living inhabitants.
Sociology has to deal with all the kinds of difficulty found in these sciences plus the right of the subjects to change their actions and provide interpretations of them that compete with those of the analyst. Because the exercise is set in the lived world, the outcome is also unlikely to cleave absolutely rigidly to its high-level explanation goals; some of what happens is bound to be the result of individual choices rather than social pressures and this will have to be said from time to time or one would not be able to fit the high level explanation with the events as we know them. So while the sociologist will try to get everything as right as possible, the paper goes out to the world in the expectation that its meaning and consequences will always be the subject of further interpretation and negotiation. (One might say that one has to operate in the expectation that one will encounter `evidential collectivism' paper 4.)
Thus, one may be certain to find mistakes in a sociological paper and much more that looks like mistakes. And if one thinks that one or more mistakes damn a whole exercise then it will be a rare sociological paper that will escape criticism and remain interesting. Given that interesting papers are, for all the reasons discussed, likely to look like an attack on the position of any reader who is a subject of the analysis, contemporaneous sociology of science is not a profession one should recommend to a friend.