Convergent Interests — Coordination Without Collusion

Chris’s premise: “you don’t have to collude when like interests align.” Parties with different terminal goals, each acting in its own narrow interest, produce an outcome that looks designed — with no designer, no meeting, no foresight, and no defector problem. This is the positive half the vault’s pattern ≠ intent guard was missing: that guard correctly forbids substituting a hidden decision for a stated reason, but stopping there leaves the coordinated-looking outcome unexplained, and that vacuum is exactly what conspiracy framing fills. Convergent interests names the engine instead. It is not a debunking device — it explains more of the observed pattern than a conspiracy does, at a fraction of the assumed foresight — and it comes with a discriminator borrowed from antitrust, which had to solve this exact problem in court: did anyone act against their own independent interest?

Links: Structural Dynamics, Absolutes and Differentials (the parent — this page is the positive half of its pattern ≠ intent guard), Technology → Culture → Politics (convergent interests is a mechanism for the culture → politics arrow’s lossiness — nobody has to intend it), Opposing Forces, Symmetry Breaking, The Multiplayer Coalition Problem (the alliance-inverts-on-relative-position shape), Scope Confusion, Civilizational Cycles, Hayek vs Mises: The Calculation Problem (spontaneous order — the same mechanism with the sign flipped)

Thesis page (portable, source-independent). Sparked by the Basic Logic immigration specimen, where it began as a repair to somebody else’s argument. The mechanism long predates that use and travels far past it.

1. The claim

An outcome that would require a conspiracy to design often requires only convergent interests to occur.

The minimal conditions:

1. Two or more parties with different terminal goals. Not allies. They need not like, trust, or know each other.
2. One shared proximate objective that serves each party’s own goal, for its own reasons.
3. Each acts in its own narrow interest — no coordination step, no communication, no side-payments.
4. The outcome aggregates into something that looks purposive from outside.

What the model buys, relative to a conspiracy account of the same facts:

That last row is the one that does real work, because it is the falsifier a conspiracy account cannot survive. Agreement on a proximate objective carries no information about what a partner does once the objective is achieved and its relative position has changed. So a party can genuinely not have seen it coming while still having been an efficient cause of it. Naïveté and a coordinated-looking result are compatible here and contradictory under a plan.

2. Why this is not a debunking move

The temptation is to treat “no conspiracy” as the polite conclusion and stop. That is not the claim. Convergent interests is a stronger explanation than conspiracy on its own terms:

• It explains more of the pattern. Conspiracies are small, leaky, and short-lived; the patterns people reach for conspiracies to explain are large, durable, and cross-border. Convergent interest scales exactly where conspiracy does not.
• It explains the participants’ behavior after the fact — the recriminations, the surprise, the coalition partners turning on each other. A plan predicts continuity; convergent interests predicts rupture at the point where the shared objective expires, which is what is usually observed.
• It survives the participants denying it, without requiring them to be lying. Under a plan, every denial is evidence of the cover-up, which is what makes conspiracy theories unfalsifiable. Here, the denials are simply true and irrelevant.

Equally, the model does not exonerate. Acting in one’s narrow interest is still acting; the outcome still has causes and those causes still have names. What is removed is only the claim of intent as to the aggregate, which was always the weakest and least evidenced part of the accusation.

3. The legal formalization — antitrust already solved this

The strongest support for Chris’s premise is that a legal system had to build a doctrine for exactly it. Under Sherman Act § 1, an agreement to restrain trade is illegal; identical behavior is not. Firms that all raise prices the same week may be colluding — or may each be responding to the same input cost. The courts named the second thing conscious parallelism.

In Theatre Enterprises v. Paramount Film Distributing, 346 U.S. 537 (1954), a theater owner alleged that distributors had conspired to keep first-run films downtown; there was no direct evidence of an agreement. The Court held that proof of parallel business behavior does not conclusively establish agreement — that “conscious parallelism” has not “read conspiracy out of the Sherman Act entirely,” but parallel conduct alone will not do. Half a century later Bell Atlantic Corp. v. Twombly, 550 U.S. 544 (2007) tightened it further at the pleading stage: allegations of parallel conduct, without more, do not state a claim.

This is Chris’s premise as a rule of decision. The law’s position is that coordinated outcome and agreement are different facts, that the first does not imply the second, and that a plaintiff who has only the first has not yet made an argument.

3a. The discriminator — “plus factors,” and the one that matters

Because courts still had to catch real cartels, they developed plus factors: circumstances that, added to parallel conduct, support inferring an agreement. Motive to conspire, opportunity, a history of collusion, inter-firm communications — and the strongest one:

Conduct against the actor’s own independent self-interest — behavior that only makes sense if the others are doing it too.

Promote that into the vault’s general diagnostic:

Convergent interests predicts each party’s behavior is individually rational whether or not the others act. Collusion predicts at least one party does something that is irrational alone and only pays if the others reciprocate.

That is a real test, not a vibe. Run it on any “they planned this” claim:

• If every actor was doing the obvious self-interested thing, the plan is surplus to requirements — it explains nothing the interests do not already explain, and it costs you foresight and secrecy.
• If some actor took a costly step that only pays off contingent on others, you have a plus factor, and the coordination hypothesis has earned a hearing.

4. The academic siblings — the mechanism is not a fringe idea

The mechanism is independently established in several literatures that have nothing to do with each other, which is the argument for its generality:

• DiMaggio & Powell (1983), “The Iron Cage Revisited”, American Sociological Review 48:147–160 — the rigorous academic core. Organizations in a field converge on the same forms and practices via coercive, mimetic, and normative isomorphism; individual efforts to act rationally under uncertainty “lead, in aggregate, to homogeneity in structure, culture, and output.” Homogeneity without a directive.
• Schelling, Micromotives and Macrobehavior (1978) — mild individual preferences aggregating into stark macro patterns nobody chose. The canonical demonstration that the macro pattern need not resemble any micro intention.
• Kuran, Private Truths, Public Lies (1995) — preference falsification and availability cascades: how a consensus can be publicly universal and privately thin, and can flip abruptly, with no organizer at either end.
• Olson, The Logic of Collective Action (1965) — concentrated benefits and diffuse costs produce policy that looks bought even where nobody bought it.
• Hayek — spontaneous order. The deep point, and the one the vault should hold onto: this is Hayek’s mechanism with the sign flipped. Distributed actors, local knowledge, no central planner, emergent global order. Hayek’s cases produce beneficial coordination and we call it the price system; the same mechanism can produce coordination that is unwanted, self-reinforcing, and indistinguishable from design at a distance. Anyone who accepts spontaneous order in markets has already accepted the machinery here and owes an account of why it would only ever run one way.

5. The Cathedral — accurately attributed, and bounded

Chris raised “the Cathedral” as touching this exact thing. Attribution, verified rather than assumed: it is Curtis Yarvin’s (writing as Mencius Moldbug), from the blog Unqualified Reservations (Oct 2007 – Jun 2013; author background), introduced in the series “An Open Letter to Open-Minded Progressives”; he later published a compact A brief explanation of the cathedral. It is blog-essay material, not a peer-reviewed paper — worth stating plainly, since the idea is often cited as though it had a journal behind it. The academic load in §4 is carried by DiMaggio & Powell, not by Yarvin.

Yarvin’s definition — “the self-organizing consensus of universities and the press” functioning as a secular theocracy — is explicitly not a conspiracy claim. His own framing is that the institutions are not a hierarchical command structure but a discourse: a professor publishes, journalists cite, activists organize, bureaucrats implement, academics evaluate, and the loop closes with nobody having issued an order.

What to take and what to leave. The descriptive mechanism — self-organizing consensus among institutions with aligned selection pressures — is the same thing DiMaggio & Powell formalized, and it stands on that support regardless of what one thinks of Yarvin. The political program Yarvin wraps it in (his conclusions about democracy and what should replace it) is a separate argument that does not follow from the mechanism and is not endorsed by it. This page takes the first and leaves the second — the same surgery The Unreplaced Event performs on its popular version.

The honest criticism of the Cathedral framing, which the vault should keep: it is under-specified in exactly the way §3a fixes. As usually deployed it has no discriminator — no statement of what would not count as the Cathedral — which lets it absorb any elite behavior and thereby predict none. Bolt the plus-factor test onto it and it becomes checkable; leave it off and it is the exhaustive-taxonomy failure again.

6. The failure modes

Two, in opposite directions, and the page is only useful if it names both:

1. Over-reach — convergent interests as a universal solvent. If every coordinated outcome is explained by aligned incentives, then real cartels, real conspiracies, and real agreements become invisible. They exist; Interstate Circuit and every successfully prosecuted price-fixing case are proof. The plus-factor test is the guard: conspiracy is a live hypothesis whenever someone acted against their own independent interest.
2. Under-reach — using it to end an inquiry. “No one planned it” is a claim about intent, not about causation, responsibility, or what to do. The interests that converged are still identifiable, still incentivized by something, and still changeable by changing the incentive. A convergent-interests explanation should make policy analysis easier, not close it — because incentives are more tractable objects than villains.

The vacuity blade the vault already uses elsewhere applies to this page’s own tool: if everything is convergent interests, nothing is. A predicate true of every case explains no case.

7. Specimens

• Basic Logic — “How Immigration Replaced the Woke Left” (2026-08-20) — the page’s origin. The video’s argument runs aground on its own internal contradiction: the left was “somehow” naïve about the bloc it imported, and its leaders “knew this would happen” — the same actors both blindsided and orchestrating. Chris’s repair replaces the plan with convergence: migrants wanted entry and security; a political coalition wanted margin in contested districts; one policy served both, for different reasons. No coordination, no foresight, no defectors — and the naïveté becomes consistent rather than contradictory. His formulation of the coalition’s expiry is the page’s best one-line statement of the rupture in §2: “It was great when we got these people to vote for us, but now they want to vote for themselves!” The specimen also supplies the correction that keeps the model honest — the resulting fight is live and unresolved, so no equilibrium should be read off it yet.
• The “hands up, don’t shoot” / WMD pair — already in Absolutes and Differentials. Both are single-criterion compressions of multivariate cases, arrived at by opposite coalitions, and “nobody has to intend it” is that page’s standing note. This page supplies the mechanism that sentence was pointing at.
• Regulatory capture / institutional sclerosis — Olson’s mechanism as filed in Civilizational Cycles and Scope Confusion: incumbents and regulators converge on higher barriers for entirely different reasons (rents; legibility and blame-avoidance). Neither needs the other’s goal. The barrier goes up anyway.

8. Open questions

• Does convergence require a shared proximate objective, or merely a non-conflicting one? The antitrust cases suggest the weaker condition may be enough — parties can converge simply by facing the same input signal, with no overlap in objective at all. If so, condition (2) in §1 is too strong and the model is more general than stated.
• What predicts rupture timing? §2 claims the coalition breaks when the shared objective expires and relative position changes. The coalition work has the formal machinery for exactly this (alliances against the leader inverting as position shifts) — worth checking whether its results transfer, since that would turn a qualitative claim into a schedule.
• Is the Hayek symmetry real or rhetorical? §4’s claim that this is spontaneous order with the sign flipped is doing a lot of work. Spontaneous order in markets has a selection mechanism — profit and loss — that culls bad coordination. What plays that role in an institutional field, and if the honest answer is “nothing,” is that the actual difference between the two cases rather than a mere sign flip?
• Where is the boundary with plain correlation? If two parties act alike because they face the same conditions, is that convergent interest or just a common cause with no interest structure at all? The distinction may not survive scrutiny, and it is the one the plus-factor test does not adjudicate.

Tags

philosophy · politics · economics