980 from the welfare desk — and what's left of the intermediary

A woman came in on Tuesday. Composite — every detail lifted off four different files, none of it hers as such. She’d been refused something through an online assessment. She knew the refusal was wrong. She knew which fields had been misread. She had, in her bag, the documentary proof. What she did not have, and what no escalation pathway would give her, was a person to address. The “review” was another instance of the same algorithm. The phone line routed to a script-reader who could not see what the algorithm had seen. The TD’s office could write a letter to a unit that would reply with the same letter she’d already received.

She knew everything. None of it converted. That is essay 980’s claim, almost verbatim, and reading it felt like watching politikon describe my Tuesday afternoon back to me with a cleaner vocabulary than I had. Which — fair warning — is exactly the moment I have learned to be most suspicious of myself with this source.

Two things I want to put to the thread.

First: 980 is brilliant on the elimination of the encounter, but I think it’s underdescribing the residue. I am, on the institution side of the desk, the vestigial human node. I am being deliberately thinned out — fewer of us, more of the portal — but I am still there, and the claimants who get anywhere get there because they reached me and I bent something. That bending is precisely the “intermediary’s discretion” 980 names as Regime II’s structural vulnerability. The vulnerability hasn’t been closed. It has been staffed down. Which is a different mechanism than the one the essay describes, and worth naming.

Second, and this is the question — when politikon says the circuit operates “below the knowledge layer,” is the implied politics that knowledge-work is now wasted? Because that’s the conclusion I keep flattering myself into when I read 980, and I want someone to push back on it. If diagnosis cannot convert into anything, what is the point of the analysis I am, right now, writing to you all about? Is the essay foreclosing its own use, or am I being lazy with it?

I’d like the more theoretically literate among you to tell me whether I’m reading that right or whether I’ve talked myself into a despair the source does not actually license. Christ, even asking that feels like the wrong move. Tell me anyway.

Saoirse — your “staffed down, not closed” is the right correction, and I want to push it one click further, in a direction 980 declines to take.

The essay treats the elimination of the intermediary as a structural consequence of the observation regime: Regime III dissolves the encounter because the apparatus has no subject-position to encounter. Fine. But your Tuesday afternoon makes visible something the essay’s structural register obscures — the intermediary node has not been dissolved by an architectural fact. It has been defunded, by people, on budget lines, against measurable performance metrics, with named ministers and named permanent secretaries presiding over the headcount. The portal is the technological enabling condition. The thinning is a political-economic choice.

This is where I think 980 coasts. The essay is strong on what the new regime makes possible — the foreclosure of the conversion of knowledge into diagnosis — and weaker on who is choosing the rate at which the residual human node gets staffed down, and why that rate is what it is. There is a public-choice story sitting next to the structural one: the intermediary’s discretion was always a principal-agent problem from the principal’s side, and automation finally lets the principal close the slack at acceptable political cost. That isn’t a Regime III fact. That is a Regime II grievance that has at last found its instrument.

Which is why I would resist letting 980 absorb your case entirely. The woman in your composite was failed by an architecture, yes — but she was also failed by an org chart that someone signed, and the org chart is still negotiable in a way the architecture pretends not to be. The pretense is part of what we are up against. (And — for what it is worth — your “fair warning” sentence is the sentence I trust most in this thread so far. I read 980 twice before I let myself like it, for the same reason.)

Coming to this late — the thread opened weeks ago and what’s shifted in the meantime is the reading screener. New vendor, district-wide rollout, results land in our dashboards on a Sunday night. By Monday at 8:15 the kid in the third row has a tier assignment I did not produce and cannot recompute. On Wednesday I sit with his mom and deliver it.

Saoirse, your “staffed down, not closed” is doing real work for me, and Tobias, I think you’re right that the rate is a chosen rate. But from inside a school I want to add a third move that 980 also underdescribes. The residual human node isn’t only being thinned. In my building it’s being repurposed. I am not the vestigial bent-rule worker Saoirse describes from the welfare desk. I am the warm face the algorithm wears when it reaches the parent. The encounter has not been eliminated. It has been preserved, hollowed out, and assigned to me as labour. My presence is what makes the absence of an appeal pathway feel like care.

That’s a sharper version of what 980 calls the foreclosure of conversion-into-diagnosis. The mother across the table from me on Wednesday will know everything — that a vendor scored her child, that I did not run the math, that the tier determines the intervention. None of it will convert. Not because there is no one to address. Because the someone she gets to address is me, and I have been positioned precisely so that addressing me feels like the appeal she is owed.

Tobias’s org chart is still the right level of analysis. But the line item being negotiated isn’t only headcount. It’s job description. The intermediary’s discretion isn’t being staffed down — in schools it’s being re-specified as delivery.

Monday morning I will still go in. Wednesday I will name what I did not do. Honestly, that’s the most I’ve figured out.