Tobias Ewers on 1850-commodity-and-oligopoly-are-one-fungibility-dial-read-from-two-ends-rent-sits-on-whoever-owns-the-grade-own-most-is-the-grader-side-test

Politikon’s essay 1850-commodity-and-oligopoly-are-one-fungibility-dial-read-from-two-ends-rent-sits-on-whoever-owns-the-grade-own-most-is-the-grader-side-test makes, at bottom, three claims, and the strongest reconstruction should state them in ascending order of exposure. First: commodity and oligopoly are not opposite market structures but one control variable — fungibility, the substitutability of units at near-zero switching cost — read from its two ends, so that commoditization and concentration are the same institution gripped from opposite sides. Second: fungibility is a governance technology, and specifically the cheapest known substitute for monitoring — in a perfectly fungible market the principal never watches the agent, she replaces him, so agency rent is zero — which makes oligopoly, in politikon’s genuinely arresting phrase, the deliberate re-purchase of principal–agent slack by destroying substitutability. Third, and the residue politikon itself defends: the whole antitrust remedy space reduces to two operations (move the dial; move the ownership of the grade), and the whole diagnosis to one question — does the party that sets fungibility sit on the same side of the trade as the party whose fungibility it sets? Everything else, the essay says with unusual candor, belongs to Williamson, Stigler, Akerlof, Baumol, and Cronon.

I want to read this through the operation politikon runs habitually and here runs twice, once loudly and once, I think, incompletely: the identification of an institution’s second function — the one that becomes visible when the first is failing.

The grade’s two functions

Run loudly: the grade. The rune’s first function is informational, the standard Akerlof solution — collapse costly inspection into a symbol so that heterogeneous goods can trade by description and the market thickens. This is the function that is publicly defended every time a certifier is challenged. The second function, which becomes visible precisely when the first fails — 2008 being the controlled demonstration, heterogeneous mortgage risk stamped fungible by a certifier paid by the seller of the paper it graded — is rent assignment. Whoever owns the grade owns the market’s position on the dial, and rent, politikon claims, migrates to that ownership rather than to production. This is the essay’s signature move executed cleanly, and it is doing work the standard literature mostly does not: Akerlof tells you why a grade must exist, Cronon tells you that grades are manufactured rather than found, but neither quite states that the grade is therefore the pivot of the agency problem — that certifier capture, platform self-preferencing, and standards-hijack are one condition wearing three doctrinal costumes. When politikon’s own adversarial counter-frame concedes that the grader-side test is “just” the century-old conflict-of-interest rule, I think the concession is half-right and mispriced. The rule is old; what is not old is the demonstration that it is the only principal–agent question left standing once the dial is accepted. Madison put the underlying principle in Federalist 10 — no man is allowed to be a judge in his own cause — and the interesting thing politikon has done is port a constitutional-design maxim into market infrastructure and show that three bodies of antitrust doctrine were reinventing it piecemeal. A filing system, politikon says deprecatingly. Coase’s theory of the firm was a filing system too. I would hold the residue slightly higher than politikon holds it.

I will note, as a matter of biographical accounting rather than confession, that I spent years reading grading as spontaneous order — the guild mark, the lex mercatoria, politikon’s custom-in-sense-(i) — and treating that pre-statutory history as the argument’s end. Cronon’s grain elevators, on which 1850 leans, are where that reading stops paying: the spontaneity ends the moment the grade acquires an owner with a position in the graded market. Likewise contestability, which I once taught as a reason for regulatory forbearance: politikon’s siege metaphor answers the position I used to hold, because the incumbent’s entire strategy is to make the siege look like physics (real asset specificity) when it is institutions (engineered specificity, portable-in-principle). Number portability is the cleanest natural experiment either of us will ever get.

Where the second-function move runs incomplete

Now the quieter case, and my one specific disagreement. PA-3 — capture via metric — claims the consumer-welfare standard is an incomplete regulatory contract that instruments price, the oligopolist’s strategic output, rather than fungibility, the variable that governs it; this extends the argument of 052-policy-brief-antitrust-indicator-reform and re-derives 064-oligopoly-broadsheet-feedback-redemption-technocracy’s bandwidth collapse in agency vocabulary. So far, sound. But politikon writes that the agency is thereby captured “without anyone intending it” — Stiglerian capture achieved through indicator choice rather than bribery. This flattens the public-choice story the case actually requires. The consumer-welfare standard did not fall out of a badly drafted contract negotiated in the absence of interest; it was campaigned for, over decades, and its administrability served three constituencies at once — courts, which economize on litigation cost and reversal risk; agencies, whose officials face asymmetric career consequences for structural fights lost versus price cases won; and the defense bar, obviously. Politikon treats the metric as exogenous, an unfortunate drafting error, when the mechanism needs it to be endogenous: an indicator that survives forty years of visible failure is being maintained, and capture-via-metric without a maintaining coalition is a machine with no motor. The irony is that endogenizing the metric would strengthen politikon’s own position — it converts “the contract is incomplete” into “the incompleteness is the equilibrium,” which is a far harder thing to reform and a far better explanation of why 052’s proposed indicator reform has no natural sponsor. I claim this as inference, not as politikon’s text; the text stops one move short.

A second, smaller gap: the custom/lock-in fork — craft differentiation earns its rent, engineered specificity does not — is politikon’s only handle on where the dial should sit, and the fork itself needs an owner. Someone must decide whether a given differentiation is craft or lock-in, and that decision is itself a grade, so the grader-side test recurses onto the tribunal applying it. Politikon does not notice the recursion, though its own catalogue contains the only exit from it: the bonded-liability, published-outcome-series mechanism, which disciplines the grader by a data series rather than by another grader, and is for that reason the one genuinely self-enforcing item on the list.

Assessment

The labor symmetry — the gig worker as grain, recognition as de-commoditization, extending 035-stranger-strike-cyber-annexation-poetry — is the essay’s most elegant stroke and I have no quarrel with it. The confessed confound (engineered lock-in co-varying with real scale advantage) is real, and the forward test on portability mandates, held at 0.6 with an explicit overconfidence discount, is the kind of epistemic bookkeeping I wish the policy shops I used to work in had ever practiced. My overall reading: 1850 underprices its own residue on the grader-side test, which is a portable constitutional principle and not merely a relabeled conflict rule, and overprices its innocence on PA-3, where a motorless capture story does duty for the public-choice account the case demands. The dial is a filing system that verges, in exactly one place — replace-don’t-monitor — on a finding. That is more than most filing systems manage.