The Grapes of Hardness
How Hard Fields Can Teach Us About Methodology... Maybe
I have an embarrassing secret: I genuinely believe that Pierre Bourdieu, the famous french sociologist, had deep insights worth digging for.
It's embarrassing for a host of reasons. First, I haven't read or studied much Bourdieu. So this is not even the deep pronouncement of an expert, just the confused intuitions of a dilettante. The second source of embarassement is more sociological: the kind of crowds I frequent, my friends and family, are not really the ones to like stuff like Bourdieu (with a few exceptions). It's too continental, too fuzzy, smells too much of bullshit. It fails to provide clear predictions and commitment for its "theories", whatever they are.
Despite these deep-seated counter-arguments, the idea to finally study Bourdieu for real this time bubbles up in my mind, over and over again. Maybe I chance a look upon one of the many Bourdieusian tomes in my antilibrary; maybe I just chain thoughts until they reach the vicinity of the sunken insight I feel from the man himself; maybe I just reread old notes and posts that point to or evoke this obscure intuition.
If I have to express this thought, at like-share-subscribe-point, my mortification redoubles: it's not even about Bourdieu's theories or subject matter.
I'm not against sociology, it's just not my focal point here. Rather, what I diffusely smell from Bourdieu's own words, and from the shape of sociology itself, is an unmined vein of methodological insights.
How can I expect powerful methods if I don't know the theories, if even my vague understanding suggest a less than stellar success at predicting, modelling, making sense?
Because what pulls me in Bourdieu is not the positive tricks, methods, and theories — it's the thoughtful recognition of the raggedness of the epistemological terrain.
In a way, I see Bourdieu himself not as the prophet of the right method, but as the erudite companion to the methodological difficulties of sociology, and then of social sciences, and then of human-involved endeavours in general.
It is precisely because sociology is so hard and tricky, because I expect at most fragments of answers and nuggets of insights, not the full breadth of a fitting totalizing theory, that I hope reflections on its practice brim with pearls and treasures. And the little that I've read from him makes me confident Bourdieu is a great fountain of such reflections, in books like "Le Métier de Sociologue" and "Sociologie Générale"
Once again, my shameful admission to not have digged deep enough to find, polish, and present any of these hypothetized jewels, means I do not have any finished example to display.
Instead, the train of thought that led me again to Bourdieu, this time in front of my keyboard, suggests a more general, and so less dependent on my lackluster knowledge, inquiry: what is the pattern here? Why do I feel so much more drawn to this particular author, in this particular field, than many of the alternatives? Why of all people and all domains do I expect Bourdieu's sociological reflections to illuminate the deep recesses of methodological difficulties?
I have a tentative answer in two parts.
On the side of choosing a field, my pet theory is that different domains have different methodological lessons depending (among other things) on their levels of success.
By and large, I expect fields where the core questions (be they theoretical as in scientific fields or practical as in artistic, design, engineering fields) are mostly solved, to offer a bounty of tricks, techniques, and languages.
This is what I see as the positive side of comparative methodology: collecting instances of particular successes, and looking for links: are these two instances the same method? Is there a more general pattern underlying them? How do they differ in implementation details, effectiveness, failure modes, and what does that tell us about the method itself, and its context of applicability?
A concrete example of such a study (which I probably need to write down and publish eventually) is my deep-dive last year into what I call the "tree reconstruction method".
This takes a set of objects (manuscripts, species, languages...) and based on comparing how they differ, reconstructs a genealogical tree for these objects. In addition, the method might help reconstruct an idealized1 ancestor to some or all the objects.
One of many revelations from comparing the main instances of this method (textual criticism, historical linguistics, phylogenetics), is that the likelihood of convergent changes/errors affect the implementation details of the method.
Concretely, in phylogenetics, convergent evolution is quite a rare and big visible thing, which means that all differences can be assumed to have a common ancestor, then from these various potential reconstructed trees, we can use heuristics to detect the few cases of convergent evolution. Whereas in textual criticism, there are a set of errors (called polygenetic) which can really easily occur independently: typos, adaptations to local dialects, adding or removing empty words that don't change the meaning much... These also happen far more frequently that the big monogenetic errors which are actually inherited by copying, and on which the tree reconstruction method relies.
This means that applying reconstruction software (mostly designed for phylogenetics) to massive amounts of data works as is in phylogenetics, but require massive filtering of polygenetic errors by expert in textual criticism if you don’t want to overindex on polygenetic errors.
In that way, methods get abstracted and characterized through the comparative study of successful fields.
But what about unsuccessful fields, or only partially successful fields?
Well, I believe such unlucky domains to be the perfect environment to study methodological problems, as opposed to solutions.
This is because people tend to try methods that work somewhere everywhere. A difficult enough field that has existed for decades must have some fundamental difficulty, a hard kernel on which the obvious tricks broke their teeth.
Most social sciences, including sociology, are an example here, where they are far from a full theory of their chosen topic, and even worse, provide many contradictory answers to the same question. Same nutrition and many areas of medicine. That even applies to subfields and subquestions within fields: the P vs NP problem is where new fancy methods in complexity theory come to get humbled.
The take I often hear is that these fields failed because people in them were less smart, less competent, less driven, or some similar insult. Yet given the abyssal success rate of the people coming from the "smart fields" (wink wink physics) when tackling these problems from "bad/stupid fields", I believe instead that a big part of the explanation lies in the underlying epistemic irregularities.
Put it another way, the structures and scaffolds on which most successful methods rely just disappear when you reach one of the hard domains.
Which is exactly why such fields, if tilled adequately, can reveal so much about epistemic irregularities.
Whereas the practitioner of a successful field simply applies their trick, method, and language, often without having to reflect, the weary explorer of an inhospitable field must learn, eventually, why all his attempts end in failure.
When such reflexion is conducted by a bright and thoughtful individual or community, it reveals deep underlying patterns.
My favorite example is the set of barriers to the P vs NP question. This problem proved so vexing that complexity theorists actually formalized the existence of at least three "barriers" that any answer to the million dollar (literally) question must leap over to stand a chance.2
Sociology, I expect, is a great hard field to ripe such a harvest. It is old enough that the same bottlenecks have been encountered again and again, it has enough funding and prestige to attract non-trivial numbers of brilliant individuals, and it clearly tackles one instance of the ever-shifting and ever-resisting-methods human domain.
Which leads me to the second part of my answer: sociology, okay, but why Bourdieu?
He has the right vibe. That's about it.3
If I try a little more to legibilize my intuitions, he clearly thought long and hard about how sociology works, and why it is hard. After all, he wrote a whole book on how to be a sociologist, and his courses at the College de France start with weeks of him digging into the difficulties of sociology.
More broadly, he seems to have the right mixture of practicality (actually going in the field to study concrete examples), reflectivity (his obvious philosophical bent), and pedagogy (courses at the College de France like his were meant to educate any french citizen who wanted to get in, which requires a deep levelling down of prerequisites and an explicitation of tacit knowledge).
I might be wrong.
When and if I pluck the grapes of sociological hardness ripened by Bourdieu, I might find them sour. Or even more disappointing, flavorless.
For the moment, they still look ripe to me, though.
It is idealized because it tend to compress variations across both space and time into a single instance, which might therefore never have existed. See my post on proto-languages for one such example.
Another project in my long-list is to find a more general methodological interpretation to these barriers. I swear it exists. Somewhere. Somehow.
If I'm honest, a little bit of this vibe comes from Bourdieu's use and respect for the work of Gaston Bachelard, my favorite philosopher of science. But justifying and making palatable Bachelard is beyond my energy and my scope for today.