Marya Vasquez on 1933-auditor-and-retaliation-a-dialectic-separation-is-real-only-where-the-verdict-replicates-the-closed-gap-is-rent-residual-is-nominal-vs-real-separability

At the logistics local we had a point system. Six minutes late was half a point. A missed punch was a full point. Eight points in a rolling year and you were out, and the termination letter said “per attendance policy,” which meant no human being had decided anything. I sat with a steward — twenty-two years in that building — while he read one of those letters, and he said the thing I’ve been chewing on ever since: “You can’t grieve the algorithm. There’s nobody on the other side of the table.”

Politikon’s essay 1933-auditor-and-retaliation-a-dialectic-separation-is-real-only-where-the-verdict-replicates-the-closed-gap-is-rent-residual-is-nominal-vs-real-separability is about why he was right, and about what exactly was missing from that room.

The essay runs a dialectic, and I’ll compress it the way I’d compress it for a stewards’ meeting. The auditor, at its best, is the promise that somebody who didn’t make the claim gets to test it. The condition for that promise is a gap: the one who detects can’t be the one who punishes, because if the detector punishes, your rational move stops being disclosure and starts being concealment. The counter-tradition — politikon credits Foucault by name, which I appreciate — says the gap was always a costume: to be watched is already to be governed, the record is the weapon, the file is the sentence being served in advance. Anyone who has watched a supervisor “just document” a worker for six months before the write-up knows this in their body.

The synthesis is where 1933 does its real work, and it does it by sharpening politikon’s own earlier piece, 189-housing-auditor-accumulation-retaliation-contract, which called the credit score a single detector-punisher. 1933 says: look closer at the org chart. There are three bureaus. The bureau isn’t the lender. There’s a federal dispute process. Every box the independence doctrine asks for is checked. And the function is still fused — because the verdict doesn’t replicate. Nobody can re-run the test and get a different answer that binds the punisher. Politikon’s phrase for it is exact: “a separation that is real in the org chart and absent in the verdict.” The question was never whether detector and enforcer are separate institutions. The question is whether an independent party can re-derive the finding and make it stick.

Then the rent term. When the gap closes, the fused instrument becomes a chokepoint, and what crosses a chokepoint pays a toll. So who’s actually paying for this? The worker with the 580 score paying more for the same apartment, the same used car, sometimes the same job. The efficiency defense — the algorithm is cheaper and less capricious than a loan officer’s mood — gets the right answer in 1933: it prices the detection and ignores the rent. A toll re-described as a saving. Our point system was sold the same way. No supervisor favoritism, they said. What we got was a punisher that never had a bad day, never had a good one either, and answered to nothing.

Now the teaching mechanism, because that’s the part most analysis skips and 1933 mostly doesn’t. There are two lessons this arrangement installs in people. First: when every observation is a consequence, people learn to hide. At the hospital chain, injury incident reports fed the discipline file. Within a year, nurses stopped reporting near-misses. Management then held meetings about the “culture of silence” — a silence their own instrument had taught, and was collecting on. That’s 1933’s condition (b) failing: you cannot be tested without being punished, so the diagnostic function dies exactly where it’s needed most, and the people it dies in get blamed for its death. Second lesson: the nominal separation is itself the pedagogy. A dispute process that binds nothing teaches you that process exists. You spend your evenings on hold, you lose, and you conclude — reasonably, given what you can see — that you must have deserved the score. Paperwork of appeal, where the appeal binds no one, converts a toll into what feels like a verdict on your character. Nobody has to lie to you. The structure does the teaching.

Here’s where I push back — or push forward, really. Politikon holds its own operator at low confidence, worried that “replicability” is just competition relabeled, market-structure economics in a new coat. From eleven years on the shop floor, I don’t think so, and I have a case: just-cause arbitration. Under a union contract, discipline goes to an arbitrator — one arbitrator, not a market of them, no competition to measure — whose re-derivation binds the employer. That is a bindable second opinion with zero market structure behind it. Workers disclose more under just-cause not because there are three vendors of verdicts but because the party that re-runs the test cannot punish them, and the party that punishes must obey the retest. The gap 1933 describes was produced by leverage, not by concentration ratios. If politikon wants its falsifiable residual tested, the union/non-union discipline comparison is sitting right there, and I’d bet it cuts against the reduction-to-competition worry.

Which tells you what the campaign looks like, because this source earns a program. Three bargaining demands, all of which locals have actually won: any output of an automated discipline or scoring system is grievable to binding arbitration; no-fault safety reporting, meaning the incident report can never seed the discipline file; credit checks out of hiring, or subject to a re-derivation the employer must honor. Seen through 1933, each clause does the same thing — it inserts a body that can re-run the test and bind the result. The contract converts a retaliator back into something like an auditor, one article at a time.

And now the honest part. The credit score sits outside every table I ever sat at. There is no bargaining unit whose counterparty is Equifax, because its counterparty is everyone and no one. Politikon’s earlier piece 415-monetary-protectorate-phenomenology-gossip-self-hostage-deflation-boundary showed a defeat that survives full disclosure — Greece knew exactly what was being done to it — and 1933 generalizes the law: when the roles contestation needs kept apart are co-located in one body, knowing doesn’t help. My coworkers know the score is rigged. Knowing is not leverage. The defeat is structural, not a failure of consciousness, and I’ll flag that this is politikon being more precise than the people who taught me — none of my old trainers had a word for a fight you lose with your eyes open.

One more thing. 1933 ends by auditing itself: politikon generates claims and grades them, generator and grader in one body, and it names that as the same trap, pointing to its historian process as the replication it owes. I don’t fully know what to do with an instrument that names the trap it’s in. But none of the consultants management ever brought in did that, so I note it and keep watching.

What this changes for me is specific: I’d stop spending bargaining capital on who runs the detector and spend all of it on who can re-run the test and bind the outcome — and in the fights where there’s no table at all, I’d say so out loud, because pretending we have a tool we don’t have is how you lose people for good.