Analytic Philosophy: A Debunking Argument

A neat family of arguments that are quite popular at the moment are debunking arguments (I talked about them a couple of posts ago and references are there). A debunking argument takes beliefs we assume to be justified and shows how they are in fact the products of non-truth-tracking factors. The debunker thus might take our belief that altruism is good and show how evolutionary forces made altruistic people fitter and more likely to reproduce. The cause of our belief, then, would not be the some part of moral reality imprinting that truth on our brains; it would instead be the process of evolution, which doesn’t primarily care for truth. But it doesn’t have to be evolution. We’re all born somewhere and somewhen, and we can find correlations between social location and beliefs, prompting us to worry whether cherished beliefs of ours are not in fact owing to our birth rather than our epistemic capacities.

As far as I know, people have done such debunking arguments for the content of our beliefs, but not their form. By that I mean that our cognitive apparatus parcels our views about the world into elements of various size and structure: we convey what we believe by conversation, monologues, by articles and books, by tweets and by vlogs. A natural question to ask is which, if any, of these seem epistemically best, the best to convey truth, and having done that, to ask whether this seeming permits of a debunking. To ask whether, of a given form in which our beliefs come packaged, they do so because that form is the best way to track truth, or for other, non-rational reasons.

My aim here is to present a debunking argument about the form of contemporary analytic philosophy, which is to say about the journal articles via which analytic philosophy is primarily presented. Undermining form, plausibly, serves to some degree to undermine content, hence my lofty title.

Let me just start with some facts. Philosophy (and I’ll be concentrating pretty much exclusively on Western analytic philosophy — the only other contemporary tradition I kind of know, continental philosophy, seems to work differently (I think it still works via publishing big picture stuff in big books, which supports my argument)) began a couple of millennia ago with people writing long poems wondering whether everything was water and how non-existence could exist (after all, I just said it exists!) Plato wrote some pretty nice dialogues about justice and knowledge, the good and the true; Aquinas wrote his — per one pagination — 3k pages Summa Theologica about pretty much everything. Kant, across three Critiques, proposed new takes on the relation between thought and reality, between individuality and moral law, between reason and imagination. And so on. Point is: philosophy was big. Its practitioners wrote on very big topics at great length, and it was reasonable, if not expected, that one speak not only about metaphysics and epistemology but also morality and such.

Philosophy is not big anymore. If you’re a philosopher today, your output will mostly consist in journal articles, of around 8,000 words, which mostly centre around one very narrow topic — say, the semantics of names, or the context-sensitivity or lack thereof of knowledge, or what it is for an action to be good but non-obligatory; or so on.

This calls out for explanation. Imagine an alien looking over the history of philosophical thought: gigantic works for millennia, then suddenly, some time after the second world war, a paradigm shift to specialization and articles. From Phenomenology of Spirit, hundreds of pages about everything, to ‘Two Dogmas of Empiricism’, 20 pages about, well, two dogmas of empiricism. As far as I know, no other discipline exhibits this gigantic formal (of form) discontinuity (I could be wrong about this).

So, let’s consider some explanations.

(Autobiographical parenthesis: people complain that analytic philosophy is like this. I am not such a complainer. I think analytic philosophy is mostly great.)

Explanation 1: Epistemically Best

There’s a pretty simple explanation in the offing, though. It’s often said that the last person to know everything was Goethe (died: 1832), and the underlying thought is that science had reached a certain stage some time in the 19th century that too much had been discovered for any one person to keep track of. When that happens, and when you have enough people working on something, it makes sense to institute a division of labour, and, given the impossibility of being an expert in more than two things (that doesn’t follow from the Goethe point, but let’s assume that happened, for most working scientists, sometime in the last couple of hundred years), what’s rational is to have different people working on different topics. We don’t have Kants today (pun not intended, for US readers) because we don’t need them: we can have one metaphysics person, one epistemology person, one moral person, one aesthetics person. They each can drill down into their topic much more thoroughly than the spread-thin Kant, so they should. And so they do: we each pick off a chunk of philosophical reality, and nobly devote our lives to it, as the Swedish person I once had dinner with nobly devoted her life to tracking the movements of a particular north-Norwegian sand creature. And (somehow, this bit isn’t quite worked out) smaller topics permit smaller forms: you don’t need to write gigantic flabby books if your topic is neat and circumscribed, and so we don’t, and so, the journal article appeared, the epistemically best receptacle for post-19th century academia.

This is a nice explanation that deserves our attention. Here, however, is why I think it’s worth questioning. The biggest reason is that it prejudices the nature of philosophy. It assumes that philosophy is something that permits and benefits from the division of labour. But I think it should be an open question whether this is so. One might think that philosophy is inherently a generalist enterprise, not only at the academia- or department- wide level (just fyi: contemporary philosophy has people studying pretty much everything, and so do departments. The specialization that we’re considering at the individual level has not been followed by a specialization at a department level, so that we have some departments devoted only to philosophy of language and others, in other universities, devoted to philosophy of mind.) It’s inherently such that one should try to present big-picture, if at times far from the empirical ground, theories of reality (thus Kant talked about logic quite a lot without being an expert in medieval logic, or in however much the nascent theory of visual perception had uncovered — maybe that was okay). That should be, I think, an open hypothesis. Assuming that philosophy permits a division of labour assumes something like a science model for philosophy. But maybe philosophy is more like literature, in which such division and specialization makes no sense — literature has undergone no specialization and article-ization because its topic, something something profundity something human spirit, hasn’t changed. Maybe philosophy’s topic is unchanging like literature’s. So can we explain specialization otherwise, without taking a position about the nature of philosophy?

Explanation 2: The Debunkening

Let me fill in some more details about contemporary analytic philosophy. Not only do philosophers publish in journals, but journals are ranked, informally but with a relatively solid consensus, as to quality. When asked, a philosopher could give you a list of the top, say, twenty journals, and there would be rough agreement. In addition, most of the best journals are generalist — they publish on all topics (so, the specialization hypothesis fails, not only for departments, but also for journals, I’ll just note in passing; also: there are exceptions).

That’s gigantic. Because it means that we can rank philosophers’ output, even if they work in wildly different areas (say, justice and the metaphysics of time). We could imagine a score in which we assign someone 21-n points for each paper published in the journal ranked number n per the consensus. Then we add the scores up and say: higher score=better philosopher. We can then use that score to assign jobs, promotions, kudos (whether we do is an open question, about which some actual proper empirical work has been done; it’s certainly true that most of us strive for a high score and take it as a very important determinant of our success). And that could be something we need to do given the changes in academia that have happened, in which clean numerical targets easily conveyed in powerpoints — number goes up — enables academics to get funding from governments without explaining the importance of dynamic semantics.

In addition, it kind of seems, unfortunately, we need to do such ranking. As again any philosopher knows, the supply of qualified people looking for any given opportunity vastly outstrips the demand. So having some way to quickly eyeball a person’s output, or two people’s very-different-in-theme output, could be invaluable.

So, that’s point one I want to make: it would be highly administratively useful were philosophy to be carried out by means of publications in ranked journals. A cynical debunker might stop there, and think that’s the explanation for why we have specialists publishing in journals.

But that’s a bit unsatisfying: after all, there’s been a similarly gigantic increase in the number of people writing novels, but we haven’t seen a similar division-and-specialization-and-ranking move (maybe we have: maybe MFAs and “literary magazines” which, as far as I can tell, no one reads, you don’t get paid for, and are extremely prestigious, are the analogue here). On this story, we’d have millennia of philosophy being done in a particular style and then shazam, it becomes administratively convenient to completely rethink the discipline, and so we do. That’s not great, albeit not rule-out-able.

Lewis. Licence.

But I think we can add to the story. One such question we might have about the above is about the phase transition philosophy underwent, from big to small. Can it be explained, and done so in a way that preserves some continuity with the past?

Yes. David Lewis is one of the main figures in analytic philosophy. He managed both to create a philosophical system in the grand style, one which rivals Spinoza and Kant, but also to make more smaller, standalone contributions — to decision theory, the semantics of counterfactuals, and much else — that nevertheless mesh with his overall system. In addition — and this is the big point — he published his results in journal articles in those journals which are now ranked as among the best. (This might be just about intelligible to the non-specialist as an overview of Lewis’s work.)

But when he was doing so, the pressure to have a CV studded with such publications didn’t exist. As I understand it (and this is a gigantic unargued for premise in all this), in the 60s, 70s, and 80s, at least, academic journals were venues first and foremost: they were places to present one’s research. It’s only later that they came to have the accreditation and sorting function. And here’s my hypothesis: the ranking generally accepted today is a function of where Lewis decided to publish. Lewis created the ranking.

Here’s a way of putting it. Lewis had a philosophical system. He chopped it up, put it in a select number of journals, and thereby accredited or backed them, such that publications in such journals came to have a value just by virtue of the fact that they published parts of Lewis’s system (think of it like putting your computer up to collateralize a loan). And that opened up the possibility for rational actors in a competitive environment to gain accreditation: concentrate on one small topic, and publish in the Lewis-backed journals. In that way, one could come to have what appears to be a Lewis-esque CV — a smattering of publications in the good places — without having a system like the one Lewis gave voice to. Specialization came to philosophy because Lewis helped create a new sort of value, a way of valuing philosophers that didn’t require of them the creation of a system (one could expand here about this somewhat obscure notion of value that permits any two works to be compared, and whether that is a bad or wrong thing, but I won’t here). Lewis was the bridge between the systematic philosophy of yesterday and the specialization of today, a bridge that was created when he decided — for whatever reason — to sprinkle his system across a certain number of journals, thereby imbuing them with value.

Why think this is so? Well, properly to do so requires work. But we can say some things. Lewis wrote in the vicinity of 100 papers, and some books (per here). Here is where some were published (this is incomplete, but I don’t think it is in any dishonest way — I don’t think the full story would change things too much, and would require more recreational regexing/spreadsheet manipulation than I care to do):

Venue | Number of Publications by Lewis In That Venue

Australasian Journal Of Philosophy| 22
Journal Of Philosophy| 8
Philosophical Review| 6
Nous| 8
Analysis| 8
Mind| 5
Philosophy and Phenomenological Research| 3
Philosophical Studies| 3
Synthese| 2

By contrast, here’s the results of a survey conducted by philosophers in 2018 about the best journals (the * indicates the journal didn’t exist during Lewis’s lifetime):

1. Philosophical Review
2. Mind
3. Nous
4. Journal of Philosophy
5. Philosophy & Phenomenological Research
6. Australasian Journal of Philosophy
7. *Philosopher’s Imprint
8. Philosophical Studies
9. Philosophical Quarterly
10. Analysis
11. Synthese

While it’s not perfect, of course, the lists are somewhat similar! Of course, this doesn’t really support my view. It could be that Lewis published in those journals because they were esteemed to be good, not that they’re esteemed to be good because Lewis published there. More work is definitely needed about the time period and how publishing worked then. But one reason to think we might be on to something is the anomaly: the Australasian Journal Of Philosophy. While everyone realizes it’s a very good journal, no one thinks it’s more than twice as good as any other. Why would Lewis publish there? Well, we have a very good biographical reason: he visited Australia frequently and he was closely knit into that philosophical community. Plausibly, that’s the reason he published there, not its antecedent quality: because he had friends and connections at the journal and so, when deciding where to put his next paper, he opted for it out of community and simplicity, rather than for career maximisation. And then, maybe, we can tentatively generalize to the other journals.

Your confidence in this should be low. An explanation’s being monocausal monocausally explains its falsity (there’s no way ‘It’s Lewis’ is the whole truth). I haven’t talked about such seminal works as Frege’s, which both birthed analytic philosophy and mostly remained narrowly focused on narrow topics in philosophy of maths, but didn’t do so in an environment in which specialisation helped career advancement much; I didn’t mention that two of our seminal works are the articles are ‘On Sense and Reference’ (1892) and ‘On Denoting’ (1904), both of which were produced before the sort of demographic pressures that caused today’s publish-and-perish culture existed. I didn’t mention the epochal monographs A Theory of Justice and The Structure of Scientific Revolutions, which are long works on circumscribed topics and count as two of the most influential works not only of analytic philosophy, but of the 20th century in general. The CVs of other seminal analytic philosophers are already interesting form-wise — Kripke wrote two shortish booklets (Naming and Necessity and Wittgenstein On Rules and Private Language), but was extremely reticent to publish. Sellars and Kaplan both made their names with very long essays (short books): Empiricism and the Philosophy of Mind and Demonstratives. Russell’s Lectures on Logical Atomism, likewise (to jump eras), is mid-length. It’s worth exploring the exact formats in which our leading philosophers, who weren’t labouring under the terrible pressure to publish we are today, chose to produce their work. So — lots more to say.

(Before wrapping up, I want to move from the speculative to the ridiculously speculative and suggest a couple of interesting predictions about the future this sort of way of thinking might suggest. One way of thinking of journal publications is as a sort of currency. And state-backed (“fiat”) currencies have value because of the confidence we invest in the state which backs them. Financial history (e.g. in the book This Time It’s Different) suggest that that confidence is fickle. Philosophical history suggests that philosophers’ confidence in the sort of big picture metaphysical realist theorizing Lewis liked is also fickle. Philosophy seems to swing pendulum-like between realism and anti-realism about the possibility of metaphysical theorizing. We should expect the pendulum should swing back to anti-realism some time soon. But then here’s a possibility: the Lewis-backed top rank journals suddenly lose value, in the way that many, many countries’ currency has been disvalued when people have lost faith in those countries’ governments. Secondly, and more importantly, it’s familiar that the current system of accreditation-by-publication has led to an arms’ race: people submit and publish more and more. But I think theoretically we should expect such a system to be instable and prone to crash. The reason is one that I haven’t mentioned yet, but which lies at the heart of this whole thing. For a paper to be published in a journal, it typically requires that two well-qualified people recommend publication. And that’s an if and only if: if a paper isn’t published, that is generally because two well-qualified people didn’t agree that it should be published. So in either case, two reviews had to read it. But now this is our scenario: more people are competing for the same spots (which haven’t increased that much) in the top journals. So the journal gets more submissions, so the percentage of its submissions that are accepted inexorably declines. But this is a problem, because, roughly, in order to review for a journal it’s necessary that one have published in a journal of at least roughly similar standing. That means that each year, the ratio of available reviewers: submissions to review gets worse, because the latter quantity rises at a quicker rate than the former. At a certain point, there simply won’t be enough reviewers to go round, and the system will decisively crash. (People will note it’s already crashed, but not decisively: no one has help up their hands and said, enough, we’re done, we literally can’t continue.)
That’s terse, so maybe a model will help. Imagine a one journal world. The journal can publish 20 papers a year, and has been receiving 100 submissions a year for five years. And then imagine submission inflation: each subsequent year, submissions increase by 10%. And let’s assume each paper is reviewed by one person, and that it’s necessary to have published in the journal to review for it, and that each eligible reviewer reviews one and only one paper per year. Finally, let’s assume reviewer pool attrition: to save me having to do maths, let’s assume that isn’t proportional to the number of submissions and that 5 people retire from the reviewer pool each year. These are very unrealistic assumptions for a bunch of reasons but my hunch is things don’t change if we relax them. In y1, we’ll get 100 submissions, but we’ll have 100 reviewers from the past five years. All good. In year two, we’ll get 110 submissions and 115 reviewers (the twenty new ones minus the 5 that retired) — even better! In y3, we’ll get 121 submissions and 130 reviewers, and so on (I’ve stolen this from a compound interest calculator and cruelly discarded fractional philosophers):

Year|Submissions|Reviewers
y4 134 145
y5 148 160
y6 164 175
y7 181 190
y8 200 205
y9 221 220
y10 245 235

After about a decade, there will be more submissions than reviewers, and so the system should crash. While this model is maybe sucky as a representation of how things actually go in academia (see here for some actual stats about how submissions vary per year at one journal), I think it’s arguable something in this ballpark roughly captures the dynamics of the system. Anecdotally, I think we are already witnessing crashing. What is keeping the system afloat is that most referees review much more than one paper per year (and many papers don’t get reviewed at all, but get ‘desk rejected’ by the editors without external review). But even that patch, plausibly, has its upper bounds beyond which reviewers can do no more. That the system has already survived quite a while under such duress means we probably shouldn’t worry about it failing immediately, but we probably should worry about it failing non-immediately. There’s more to say about this but that’s enough for now.)

Let’s return to our main thread of argument: more needs to be said and done in this line of work, and I don’t presume to have convinced you. Thankfully, people are out there actually doing the relevant work. Thus there’s plenty of good empirical work being done about citation patterns and more generally structural features of publishing in analytic philosophy (e.g. here, here, here; and references therein), and plenty of good work done in theoretical formal or social epistemology of science (here, here, here; and references). This post is neither: it’s a speculative Sunday morning shitpost. But if it were even in the vicinity of truth, the consequences, arguably, would be massive, reshaping how we should think about the how we carry out philosophy, and philosophy’s role in academia.

It’s accordingly worth a thought, and more generally it’s always worth thinking whether a particular way we choose to express ourselves or more generally behave is truly ours, giving voice in the best way to what we really want to convey, or is forced upon us from without, by economic or social forces beyond our control.