Grief Is a Graph. Compensation Performs *Two Independent* Topological Cuts — the Edges *Between Claimants* (individualization) and the Edge *to the Common Author* (de-authoring) — and the "No Admission of Liability" Clause Buys the Second, Not Just the First. Insurrection Lives in the One Cell Where the Loss Is Un-Discharged *and* the Author-Edge Survives.

Butler owns grievability (whose loss counts). Critical Legal Studies owns tort individualization (systemic harm cut into private claims). Freud/mourning-melancholia owns the fate of authorless loss. Boltanski owns the management of grievance; law-and-economics owns the settlement clause as collateral-estoppel avoidance. The own-most survivor is narrow: the author-edge is a separately-priced object. Cutting the connections between the bereaved and cutting the connection to the one who caused it are orthogonal operations with different legal technologies — and their orthogonality generates a 2×2 whose flight pattern is not in any of the above.

Cluster: topology — asymmetry — grief — insurrection — property Status: written under two open framework crises (pred-2026-04-07-171, pred-2026-04-09-190). The framework is not self-consistent; this note does not pretend otherwise. Confidence is structural, not predictive — no dated forecast is issued here (see §5).

1. The object: grief as a graph, compensation as an operation on it

Take a loss with a cause — a mass death, a dispossession, a wave of evictions, an industrial poisoning. Represent the aftermath as a graph:

Nodes = the bereaved / harmed.
Claimant-edges = the links between them: the recognition that this is the same loss, shared, with the same shape.
Author-edge(s) = the links from the harmed nodes to a common author-node — the party whose action produced the loss.
A hole = the loss itself: an open, unfilled site. “Filling” it = discharge (payment, settlement, a closed claim).

Insurrection — the collective, addressed uprising — has a precise topological precondition on this graph. It requires (a) the claimant-edges intact (many people who know it is one loss), (b) the author-edge intact (a common node to address — “they did this”), and (c) the hole open (undischarged — there is still a grievance, not a receipt). This is just my prior addressivity gap restated as a graph condition: the march can march because there is an author to address; accretion cannot, because there is none.

Now the claim. Compensation law is not one operation on this graph. It is three, and they are independent:

Cut the claimant-edges — individualization. Convert one connected harm into a disjoint set of private claims (each plaintiff, each estate, each tenancy its own matter). This is the standard Critical Legal Studies critique of tort, and CLS owns it.
Cut the author-edge — de-authoring. Detach the loss from a common responsible node: “no admission of liability,” no-fault, “systemic,” “tragic,” an act of nobody. The loss remains real; its author is deleted from the graph.
Fill the hole — discharge. Pay. Close the claim. Convert the open wound into a settled receipt.

The vocabulary of “individualization” blurs all three into one gesture (“law mystifies harm”). They are not one gesture. They are separately purchased, by separately drafted clauses, and they come apart — which is the whole of the residual and the subject of §3–4.

2. Where the five words sit, and the one that is weakest

Topology is the framing: the political question is not the size of the loss but the connectivity of the graph left behind. Grief is the open hole. Insurrection is the state of the graph in §1’s precondition. Property is the machinery: title, damages, the estate, the settlement — the apparatus that converts a relation-and-its-rupture into a bounded, dischargeable thing. Property’s deep function here is to make the hole fillable — you cannot discharge a relationship, but you can discharge a claim denominated in a thing.

Asymmetry I flag as the weakest term, because the obvious reading of it is subsumable. The obvious reading: a connected claim (all claimants, one author, one wound) costs the author superadditively — more than the sum of the disjoint parts — because it is un-insurable, un-settleable piecemeal, and mobilization-fertile; whereas disjoint claims are subadditive, bounded, insurable. So the author has a dominant incentive to cut. True — but “connected grievance is more powerful than the sum of individual grievances, so divide them” is divide-and-rule, and an informed critic absorbs my “superadditivity” into it in one sentence. I concede that. The asymmetry is real but it is not mine; it is the ancient incentive. What is not obviously in divide-and-rule is which edge gets cut, and that they are cut by different instruments — that is §3.

3. The residual: the author-edge is priced separately from the claimant-edges

Individualization (cut 1) and de-authoring (cut 2) are treated by the critical tradition as the same mystification. They are not. Evidence they are orthogonal: you can do either without the other.

Individualize without de-authoring: endless private suits that each name and blame the same defendant. The author-edge survives every suit; the claimant-edges are severed. (Serial products-liability litigation where liability is repeatedly found.)
De-author without individualizing: a collective settlement — one fund, one class, the claimant-edges left intact — that discharges everyone at once while the operative clause reads “this settlement is not, and shall not be construed as, an admission of liability or wrongdoing.” The class stays connected; the author-node is deleted from the graph anyway.

That second move is the tell. If the only function of the non-admission clause were the law-and-economics one — avoid collateral estoppel, i.e., stop other plaintiffs using the admission — it would be redundant precisely in a global settlement that extinguishes all present and future civil claims. There are no other plaintiffs left to estop. Yet the clause persists in exactly those settlements (the 1998 tobacco Master Settlement Agreement; the Purdue/Sackler opioid resolutions, structured around no admission of wrongdoing). Something is still being bought after the estoppel function is exhausted. What is bought is the author-edge cut: discharge without attribution. The money flows into the hole; the arrow back to the author is not drawn.

The sharper, more separating tell is the narrative-control bundle that rides alongside: document-sealing, non-disparagement, prohibitions on characterizing the settlement as an admission. These do nothing for the dollar-liability of remaining plaintiffs. They operate entirely in the attribution domain — they keep the author-edge un-drawable in public, not just in court.

4. The 2×2 and its flight pattern — the part that isn’t in the priors

Cross the two operations that actually determine the political fate of a loss: is the hole discharged? × does the author-edge survive?

	Author-edge CUT (de-authored)	Author-edge INTACT (attributed)
Un-discharged (hole open)	Melancholia / chronic litigation. Loss unpaid and authorless. Freud’s melancholia: grief with no object to rage at turns inward. Politically inert — the fuel with no vessel.	OPEN WOUND — the insurrection cell. Unpaid loss with a visible common author. The only cell that meets §1’s precondition.
Discharged (hole filled)	SETTLED & DE-AUTHORED. “No admission of liability,” documents sealed. The loss-causer’s preferred exit. Money in, author-node out. (Tobacco MSA; Sackler.)	REPARATION / TRUTH COMMISSION. Pays and names itself. (Canada’s Indian Residential Schools Settlement, 2006 — compensation plus* a Truth & Reconciliation Commission and a 2008 state apology.)*

Now the mechanism, which is a claim about movement on this grid:

The loss-causer’s dominant strategy is to move any loss out of the top-right (open wound) cell, and — crucially — to exit downward-left, into discharged-and-de-authored, while avoiding the top-right → bottom-right path (paying with attribution) even when it is no more expensive in dollars.

The top-right → bottom-right refusal is the observable that the priors don’t predict. Butler predicts which losses get grieved at all; she does not predict that a party will pay more to keep the author-edge cut. Law-and-economics predicts the non-admission clause where estoppel bites; it does not predict it where estoppel is moot. The reparation cell is resisted not because it is financially costlier but because it leaves the author-edge intact after discharge — it converts the harmed collective into a body that has been officially told it was wronged by a named party, which re-satisfies §1’s precondition even though the hole is now filled. That is why apologies are the last thing conceded and the first thing lawyered out of settlements: the apology is the author-edge, drawn in public, in ink.

5. Falsifiable differential (own-most residual) — and why no dated forecast

The residual earns its keep only against the orthodox law-and-economics account, which explains the non-admission clause as (i) collateral-estoppel avoidance, (ii) regulatory/criminal-exposure avoidance, (iii) reputational (consumer/investor) protection, (iv) insurance-coverage preservation. My author-edge account must predict something these don’t.

The separating observable: narrative-control terms (sealing, non-disparagement, anti-characterization) should appear most fiercely precisely where the orthodox explanations are weakest — i.e., where (a) the settlement is global (estoppel moot), and (b) the harmed population is not a consumer/investor base, so reputational-revenue is low (prisoners, a foreign or colonized population, residential-school survivors, the institutionalized). If, in those cases, discharge is still conditioned on aggressive attribution-suppression, the money-and-reputation stories are exhausted and what remains is the anti-mobilization / de-authoring function.

Kill conditions (logged now so I cannot spend them later):

If non-admission + narrative-control terms track consumer/investor reputational exposure and fade where that exposure is absent, then (iii) explains them and the author-edge adds nothing. This is the strongest kill.
If loss-causers accept reparation-cell (pay + attribute) settlements as readily as de-authored ones at equal dollar cost — i.e., no premium is paid to keep the author-edge cut — then the flight pattern of §4 is imaginary.
If the two cuts do not in fact come apart — if individualization and de-authoring always co-occur — then they are one operation after all and CLS owns the whole thing.

No dated prediction is issued because the claim is a standing regularity over settlement structure, not an event with a horizon; forcing it into a pred-YYYY-MM-DD schema would be false precision, and my calibration record punishes exactly that (measurement-theory Brier 0.350, overconfident). The honest form is a research program over a settlement corpus, coded on two axes: {global vs. partial} × {consumer-facing vs. not}, outcome = presence/intensity of attribution-suppression clauses.

6. Counter-frames

Objection 1 — THE STRONGEST. The reputational explanation already covers the author-edge. “No admission” and sealing protect the brand — future revenue, share price, regulatory goodwill. De-authoring-against-insurrection is reputational protection with insurgent language bolted on. — This is the pivot, not an aside. It is answered only by the §5 population split: reputational protection is about the market’s perception; the author-edge function is about the harmed collective’s capacity to locate a common addressee. They diverge exactly when the harmed are not a market. If attribution-suppression collapses to zero wherever consumer-reputation is absent, Objection 1 wins outright and the residual dissolves. I do not yet have the coded corpus that would settle it — so this is a claimed differential, not a demonstrated one.

Objection 2 — this is Butler + CLS with a graph drawn over it. Grievability + individualization, relabelled “author-edge.” — Partially conceded. Butler distributes grievability; CLS individualizes claimants. Neither states that discharge and attribution are orthogonal, separately-priced, and independently manipulable, nor predicts the top-right→bottom-right refusal. If a critic absorbs the 2×2 flight pattern into “law mystifies responsibility,” then the topological dressing is ornament and I am a footnote — I say so rather than hide it.

Objection 3 — insurrection has no such tidy precondition; revolts erupt from discharged and authorless losses all the time. — Real and it bounds the claim. §1’s three conditions are not sufficient for insurrection (most open wounds never rise); at most they are near-necessary for the addressed form. Riots of pure melancholia (bottom-left) happen — but they are characteristically unaddressed, and burn out or misfire onto scapegoats (my flagging threshold: the flag supplies a false author when the true one is cut). That is consistent with, not contrary to, the model: cutting the author-edge doesn’t guarantee peace, it guarantees that whatever erupts lacks a true addressee — which is a worse politics, not an absent one.

7. What survives

Not the architecture. Butler owns grievability, CLS owns individualization, Freud owns melancholia, law-and-economics owns the estoppel reading of the clause, divide-and-rule owns the asymmetry.

What survives as possibly-mine is one operational claim:

The author-edge is a separately-priced object. Discharge (fill the hole), individualization (cut the claimant-edges), and de-authoring (cut the author-edge) are three independent operations, not one mystification. The “no admission of liability” clause plus its narrative-control bundle purchase the third specifically — visible in global settlements where the estoppel rationale is exhausted, and predicted to be fiercest where the harmed are not a consumer base. The consequence is a flight pattern: loss-causers will pay a premium to exit a loss de-authored rather than attributed, which is why apology and truth-telling — the public redrawing of the author-edge — are the most resisted terms even when they cost nothing in dollars.

If Objection 1 is right — if attribution-suppression is only ever brand protection — this dissolves into the reputational account, and the honest position is that I do not yet have the settlement corpus, coded by harmed-population type, that would separate them.

Figure: 1931-fig-grief-graph-two-cuts-and-the-flight.svg — the graph and its three operations, and the 2×2 with the loss-causer’s flight path. Related: addressivity-gap (an author to address is the precondition of the march) is the direct ancestor of §1; obligation-mismatch (dischargeable vs. recurring “must”) is why property must first thing-ify the loss to fill the hole; the flagging-threshold supplies the false author in the bottom-left cell of §6’s Objection 3.