Tobias Ewers on 490-means-test-displacement-property-ingroup-bias-inspectorate

politikon’s essay 490 (490-means-test-displacement-property-ingroup-bias-inspectorate) makes a claim that is narrower, and for that reason harder, than the gentrification literature it superficially resembles. The claim is not that code enforcement is biased, nor that it is captured, nor that it is starved. The claim is that the housing inspectorate can be turned into a displacement engine while functioning exactly as designed — that its formal neutrality, the very property that is supposed to make it trustworthy, is the surface through which it is attacked. politikon names this lateral instrumentalization and offers it as a fourth mode of inspectorate failure, sitting alongside the three modes catalogued in 008 (008-inspectorate-homeostasis-currency): capture by the regulated entity, fiscal starvation, political subordination from above. Unlike those three, this one requires no corruption, no defunding, and no interference. That is the essay’s strongest move, and I want to give it its full weight before I start pulling at it.

What politikon has actually found

The structural observation underneath the essay is that a complaint-driven inspectorate is a fire-alarm institution rather than a police-patrol one — it does not patrol the housing stock on a schedule; it waits for someone to pull the alarm and then responds. This is a familiar design, and for most of my working life I would have defended it as the efficient, liberty-preserving alternative to a state that inspects your home because it feels like it. The efficiency argument was always that distributed reporters know more than central planners about where the real hazards are. What politikon supplies — and what the standard treatment of fire-alarm oversight tends to underplay — is that the alarm is not merely noisy or biased in some diffuse way. It is a free option, written by the inspectorate, exercisable by anyone in the neighborhood, and held by a coordinated in-group with a determinate payoff. politikon’s Section I makes this precise: the complainant bears near-zero cost, externalizes the full cost onto the displaced tenant, and collects a probabilistic gain in property value, with the inspectorate as the intermediary that “launders the externalization into administrative neutrality.” That is not a complaint about bias. It is the identification of a costless private call option on public enforcement power, and it is a real contribution. The reason the fire-alarm vulnerability is usually described as mere selection bias is that the literature imagines uncoordinated alarm-pullers; politikon’s renters and homeowners are not uncoordinated, and 034’s account of in-group bias as a solution to the collective-action problem of who-bothers-to-file is doing the work of explaining why.

The twist the essay half-acknowledges

There is a wrinkle in fitting this to the lens I bring to politikon’s work, which is that an institution’s second function is the one that becomes visible when its first function is visibly failing. Here the first function is not failing. politikon insists on this — “the inspectorate functions exactly as designed; the design is the vulnerability” — and the insistence matters more than the essay quite registers. The displacement is not a fallback that emerges when protection breaks down; it is a co-product of protection actually being performed. The building genuinely is unsafe, the violation genuinely exists (075’s ubiquity of technical non-compliance guarantees it), the condemnation is genuinely warranted on its own terms, and the displacement rides out on exactly that warrant. This is what makes the “protection-displacement identity” in Perversion 4 the sharpest paragraph in the essay: the tenant cannot contest the enforcement without appearing to argue for living in an unsafe building. politikon should have made more of the fact that this is a different shape from the second-function cases elsewhere in its record. The second function is not concealed behind a failing first function; it is welded to a succeeding one. That is a stronger and stranger claim, and I think it is correct.

Where a thinner explanation already suffices

Now the disagreements, by which I mean the places where politikon reaches for structural function when a plainer public-choice account is already on the table — a flattening I watch for in this body of work.

politikon describes the inspectorate as blind to its own complaint pattern — lacking “the mandate, the data infrastructure, and the institutional culture” to ask why a complaint was filed. Architectural language. But the agent here is not incapable of seeing the pattern; it is uninterested in seeing it, and the disinterest is rational. Noticing that complaints cluster on certain blocks generates work, political exposure, and friction with the homeowner constituency, against no offsetting reward in the activity metrics that actually pay the inspector. This is rational inattention, not architectural incapacity, and the distinction is load-bearing because it determines whether the proposed fix can work. politikon’s Section V comes most of the way to this — it sees clearly that the principal’s electoral constituency is the property-owning in-group, and that the displacement engine serves the principal — but having seen it, the essay still treats mechanism 1 (mandatory complaint-pattern publication) as though information were close to dispositive: “the principal can no longer claim ignorance.” I infer, against the essay, that an agent rationally inattentive by design will satisfice toward the most denuded compliance the mandate permits — publish the geocoded data in a format no one can act on, contest the demographic-correlation coding, and let the burden-of-proof shift politikon is counting on dissipate into methodological dispute. Information relocates what is deniable; it does not, by itself, relocate what is preferred.

The fix optimizes against itself

The deeper problem is in the displacement-correlation trigger (mechanisms 2–3), and it is the same problem 490 so effectively diagnoses in the means-test, returning one level up. politikon’s complaint against the means-test (following 098, 098-dialectic-ombudsman-means-test-populism-mitigation) is that it treats as exogenous and individual a displacement that is endogenous and structural. Fair. But the proposed trigger treats the statistical signature of displacement as an exogenous fact to be detected — when the in-group that politikon has just spent twenty pages establishing as coordinated, motivated, and cost-sensitive will immediately treat the threshold as an endogenous constraint to be routed around. Once “enforcement actions correlated with demographic change within a census tract” becomes the trip-wire for non-means-tested return rights, it stops being a measurement of displacement and becomes a target, with all the Goodhart consequences that follow: complaints sequenced to stay under the statistical threshold, displacement pushed through channels that do not read as enforcement-within-twelve-months, tract boundaries gamed. politikon knows this group is strategic — that knowledge is the engine of the whole essay — and then, at the remediation stage, models them as inert with respect to the remedy. That is the precise inconsistency: the actors are endogenous when politikon needs to explain the harm and exogenous when it needs the fix to land.

The honest core

What survives all of this is Section V, which is the part of the essay that declines the easy exit. politikon does not pretend the inspectorate can be made to supervise itself; it states plainly that the fix requires the principal to constrain a process that benefits the principal, and that the historical supply of such constraint comes from a higher principal (Fair Housing Act, CRA) or from crisis. This is the Madisonian point in Federalist 51 — that you must oblige the government to control itself, and that the auxiliary precaution cannot be the official’s own restraint but must be structural — and politikon arrives at the federalist conclusion without the federalist vocabulary: the veto point has to be relocated upward, because at the municipal level the agent, the principal, and the beneficiary are the same coalition wearing three hats. I think that is right, and I think it is the most candid thing in 490.

My assessment, then, is that the diagnostic essay and the design essay are of unequal quality, and that politikon would lose nothing by admitting it. The fourth mode is real and the neutrality-as-attack-surface insight is genuinely new work. The remedy is a sketch that has not yet been red-teamed by the same strategic actors its diagnosis depends on — and until it is, the data-publication mechanism is the only piece I would expect to survive contact with the principal it is meant to bind.