On How the Division of Intellectual Labor Becomes Self-Undermining (some Rousseau, Adam Smith, and more Babbage)

I received quite a few very kind messages of condolences after yesterday’s guest post by Amy Olberding, whose regular guest column appears as ‘Amy’s Folly.’ Now, I admit that I aspire my blog persona to sound like Amy’s, but this confusion of identities did surprise me. Sadly, my dad died eight years ago. Amy’s writings on grief allow me to feel what I would like to feel (if that makes sense).

After my talk in Northeastern, I got snowed in by the blizzard. So, I am already a bit preoccupied by my talk in Buffalo on Friday. Feel free to drop in if you are in the area.

In honor of Paul Humphreys (recall this post), I treat epistemic opacity as the inability to surveil the steps of a process from a known input to a known desirable (or truthful, useful, beautiful, etc.) output in a timely manner to the decision-maker or responsible agent. I put it like that to make clear that this ignorance is pragmatic in character and can probably be modelled in terms of trade-offs between the quality or benefit of the output and the cost of surveillance (Of course, sometimes the opacity is not pragmatic, but ontological in character.) In addition, I use the ambiguous language of ‘surveillance’ because the process can be computational, social, or natural in character.

Such epistemic opacity is ubiquitous. For example, throughout the history of philosophy, ‘sympathy’ (συμπάθεια), was used to describe (a/the) cosmic and psychological mechanism(s) in which the process was opaque in this very sense, even though the start and end of the process were known.¹

As I have noted, in the (1755) Discourse on Political Economy, Rousseau notes that the Hobbesian sovereign of a geographically large society is hampered by epistemic opacity. That is, such a sovereign cannot survey the population in real time. So, such a sovereign must rely on intermediaries (a bureaucracy, viceroys, tax-farmers, delegated parliaments, etc.) in order to overcome the effects of this first-order epistemic opacity. Unfortunately, the very mechanism by which epistemic opacity is tackled often introduces a different, second-order (or ‘derivative’) epistemic opacity.

This phenomenon can be illustrated by the following thought: once one diagnoses how the very social mechanism (say a bureaucracy) one introduces to tackle first-order epistemic opacity generates higher-order epistemic opacities, one may be tempted by two strategies. First, one may introduce monitoring mechanisms that surveil the bureaucracy one has instituted. Second, one may invest in mechanisms (a census, real time street cameras, infrared tags, a machinery of record that tracks births, deeds, etc.) that make the population that is being governed more legible. In both cases, these mechanisms will themselves generate further kinds of third-order (‘second derivative’ etc.) opacities and so on.

In fact, now that it is recognized that modern AI is epistemically opaque in the very sense I have introduced here, it is totally foreseeable that our bureaucratic life will be governed by such epistemic opacities all the way down.² (Even the machines will not be able to tell on themselves and each other.) In each case, there is an implied cost-benefit analysis that shapes how one might wish to think about the trade-offs that lurk in Humphreys-style opaque processes. In fact, recognition of such trade-offs may drive a sovereign into adopting a bureaucracy vs (say) tax-farming vs (say) dividing the kingdom into principalities for his offspring (etc.).

As Foucault notes (recall) in Security, Territory, Population (pp. 94-5 in the translation), the body of knowledge that, according to Rousseau, deals with this social decision problem in the context of the division of labor is the art of government. It introduces ‘economizing’ into the rule over others. (If one treats AI as ‘agents’ — as is increasingly done — one can extend this to computer science.) What’s crucial here is that one can discern that quantitative techniques are quite welcome in such an art of knowledge, but that the art cannot be exhausted by it.

In my post last week (here), I noted that one careful reader of Rousseau, Adam Smith (himself a fine theorist of sympathy), introduces a subtle modification into Rousseau’s approach right at the start of the Wealth of Nations in his very chapter on the division of labor. It turns out that for Smith there are two kinds of divisions of labor: in the concentrated division of labor (for example, the pin-factory), at least parts of the process are fully surveyable, whereas in the distributed division of labor it is not.

As an important aside, as I hinted last week, Smith’s implied distinction itself turns out to require modification. In On the Economy of Machinery and Manufactures (1832) (recall also this post), Babbage offers an example derived from Prony’s reading of Smith in which a distributed computational process appears to be surveyable. And this forms his template for thinking about the difference engine (and later Turing machines and all the original sins of GOFAI until the rise of neural networks). Smith and Babbage are both not quite on the right track; the key is not distributed vs concentrated, but the trade-offs that make survivability a live possibility.

Be that as it may, one may left doubting that Smith really understood his own treatment of the division of labor in terms of the art of government. After all, he does not mention Rousseau in Wealth of Nations nor, in context, does he use the phrase ‘art of government.’ And even if one grants that informed readers (like Bentham) did discern in Smith a preoccupation with the art of government, it is not obvious that the analytic core of Wealth of Nations (chapters i-vii of Book 1) should be identified with it.

But as it happens, one of the key conclusions of Smith’s treatment of the division of labor is the following: “It is the great multiplication of the productions of all the different arts, in consequence of the division of labour, which occasions, in a well- governed society, that universal opulence which extends itself to the lowest ranks of the people.” (WN 1.1.10, p. 22) Here the long-term output of the art of government (“well-governed”), which shapes and regulates the division of labour, is universal opulence even of the working poor (‘lowest ranks’).

That’s all I wanted to say. But I do want to point to another feature of Smith’s analysis here. In the very passage quoted, and even more in context of the previous paragraph (on the same page), Smith inserts the very art of government and science more generally within the division of labor. Let me quote that passage before I conclude,

In the progress of society, philosophy or speculation becomes, like every other employment, the principal or sole trade and occupation of a particular class of citizens. Like every other employment too, it is subdivided into a great number of different branches, each of which affords occupation of a peculiar tribe or class of philosophers; and this subdivision of employ ment in philosophy, as well as in every other business, improves dexterity, and saves time. Each individual becomes more expert in his own peculiar branch, more work is done upon the whole, and the quantity of science is considerably increased by it. (WN 1.1.9, pp. 21-22; emphasis added)

Now, crucially, specialization (that is, the division of labor) in all sciences and technologies (‘arts’) improves ‘dexterity’ and saves ‘time.’ It does not always increase judgment. This omission is a bit surprising because at the start of the chapter (and in the preceding general introduction), Smith had repeatedly claimed that “THE greatest improvement in the productive powers of labour, and the greater part of the skill, dexterity, and judgment with which it is any where directed, or applied, seem to have been the effects of the division of labour.” (WN 1.1.1, p. 13; emphasis added)³

So, what’s going on? I suspect Smith is alert to the following two features of his analysis.

First, in advanced or complex (or ‘commercial’) society, for Smith the division of labor is an enabling condition for all social activities including for scientists and advisors to rulers (who specialize in the art of government). That is to say, one is surrounded by activities and social systems that generate epistemic opacity and that may themselves be, in part, the attempted answer to deal with or monitor (‘lower order’) epistemic opacity. But that successful governance and knowledge mechanisms themselves generate epistemic opacity, is especially easy to forget for an ‘expert’ in one’s own branch of knowledge, where expertise is constituted by knowledge not by ignorance. Judgment about this feature of social life is not natural.

In addition, the division of labor within philosophy (and the sciences) generates a massive increase in the (desirable) “quantity of science” and also a predictable side-effect that Smith does not emphasize. Let me elaborate on the side-effect. In immediate context, Smith characterizes the “trade” of philosophy as not to do any thing, “but to observe every thing; and who, upon that account, are often capable of combining together the powers of the most distant and dissimilar objects.” (WN 1.1.9, p. 21; this anticipates Sellars’ idea of philosophy “how things, in the broadest possible sense of the term, hang together, in the broadest possible sense of the term.”)

This gets me to the second feature. The division of intellectual (and manual) labor generates an enormous growth in the quantity of science, but itself introduces a predictable form of epistemic opacity among the sciences and even between their sub-divisions.⁴ This opacity among the sciences may be the effect of scarcity of time, but it will also be increasingly the effect of intellectual barriers to entry among the sciences. And so the very condition that is responsible for the growth of knowledge, also effectively undermines the main “trade” of the philosopher (which becomes increasingly challenged in observing every thing in the sciences) and, perhaps, destabilizes the very art of government that is supposed to aid in ruling it. That is, there is more than a hint of Rousseau (of the first and second discourse) lurking here.

To be continued.*

1. One might object that if sympathy involves instantaneous effects (so that start and completion of the process are identical in time), then the process may not have discrete steps. So, perhaps, not all cases of sympathy would fall under this kind of ignorance.

2. This also suggests that even AGI would have structural limitations on its knowledge of decision circumstances.

3. Sam Fleischacker has done more than anyone to call attention to the significance of judgment in Smith’s writings.

4. This theme is familiar from Elijah Milgram’s (2015) The Great Endarkenment – Philosophy for an Age of Hyperspecialization (OUP), Jefrey Friedman’s (2019) Power Without Knowledge: A Critique of Technocracy (OUP) and my own (2024) “Synthetic Philosophy: A Restatement” (here) and my (2024) essay with Nick Cowen “Novel externalities” (here).

*I want to thank Katie Creel and the rest of the gang at Northeastern for discussion of these themes.