In this section: Adamic Theory · Radical Reason · Anastomotic Kino-Materialism · Strategic Bricolage · Urrepublika · Palimpsest · Canon Lines & Forcing Phrases · Theme Explorer

In this branch: AKM overview · A Regenerative Pluriverse · The Fascomachy: An Interactive Guide · Against Drift · The Copernican Sequence · Human Self-Domestication Syndrome · Ghost Rivers and Hogging Hubs · Chokepoints · The Sapiens Terminus Pattern · The Double Bank · The Anastomotic Circuit · The Instruments of Omission · The Transcendental Object at the End of Time

Chokepoints

A Political Economy of Network Position

Anastomotic Kino-Materialism · Network Power · Political Economy

Plain-Speech Canon: A chokepoint is not a busy place. It is a place others cannot cheaply avoid.

On this page: The bridge and the detour · Centrality does not appoint a sovereign · What the detour costs · Three cases · The Chokepoint Test · How this argument can lose · Return pressure has to be built · Who inherits the gate? · Receipts · Sources

I. The bridge and the detour

Take a bridge that carries ten thousand people a day. Build a second bridge beside it, just as sound, with room to spare and a different owner, and the man in the first tollbooth develops a sudden interest in customer service. People can leave.

Now burn the second bridge and put the nearest crossing three days downstream. The first bridge carries exactly the same traffic as before. The man in the tollbooth, however, has become a philosopher of pricing.

Nothing about the bridge changed. Only the detour did.

The bridge is invented, traffic count and all. Its lesson still has to earn its keep, which is what the rest of this page is for.

We are usually told that power is something you own: money, land, weapons, an office, a degree framed in the hallway. Those matter. But a person can own very little and still stand in the one doorway everybody else must use to reach freight, credit, care, a login, a paycheque, a permit or an afternoon.

So the useful question is not how much passes through a position. It is what happens to the people who try to go around it.

That is the wager of this page, and it fits on the bridge: if you want to know who can dictate terms, the price of the detour tells you more than the size of the traffic. It is a wager, not a law of nature. Three agreeable examples will not promote it.

Power is not how much traffic you carry. It is how much the detour costs everyone else.

Network comparison · schematic

Same hub. Different price of leaving.
Usable, cheap bypassFour terminals connected through hub H, with cheap bypass routesABCDH

Capacity exists. Switching fits the deadline. The hub is avoidable.

Costly or unusable bypassFour terminals connected through hub H, with costly bypass routesABCDH

Capacity, time, compatibility or permission makes leaving costly.

Solid lines: routes through H. Dashed lines: alternatives to inspect. Geometry does not represent distance or cost.

Then check control. A bottleneck becomes gatekeeping power only through the ability to set terms; service duties, regulation and collective action can constrain it.
Centrality alone does not establish power. Compare usable alternatives and the reserves needed to reach them; this schematic is not empirical evidence.

II. Centrality does not appoint a sovereign

Network analysts have a word for the big dot in the middle of the diagram: central. Degree counts its connections. Betweenness counts the paths that run through it. Neither number tells you who holds the keys, what the other road costs, or who can afford to wait.

None of this is a fresh discovery, and this page does not pretend otherwise. Stephen Borgatti showed that a centrality measure has to fit whatever is actually moving through the network. Phillip Bonacich showed that in exchange networks centrality and power need not rise together. Richard Emerson supplied the plain root of the matter: my power over you depends on how much you need what I have, and how easily you can get it somewhere else.

Three distinctions keep the big dot honest.

A bottleneck is not a gatekeeper. A storm can close a route and a damaged cable can choke a flow without anyone giving the order. Vulnerability is not coercion. It is bad weather with consequences.

A gatekeeper is not always free to squeeze. A highly central utility may be bound by service obligations, independent review and price limits. Its customers may have nowhere else to go while its managers have nowhere lawful to take them.

A threat is not a victory. An actor can impose heavy costs and still fail to get what it asked for. Breaking things and winning arguments are different skills, often confused by people who possess only the first.

Draw a large dot in the middle of a diagram and readers will crown it. Check who has the keys.

III. What the detour costs

Here a chokepoint is a position that controls access to a flow that matters, for people whose usable alternatives are expensive within the time they have. Chokepoint power is the ability to turn that dependence into terms. The bet is that the cost of the detour, measured on its own, explains who gets their way better than the traffic count does.

An alternative only counts if it delivers enough of the service, soon enough, to the people who need it. A rival supplier who will not take you, has no capacity, fails the standard or sits behind a rule you cannot clear is not an alternative. It is a brochure.

The detour carries four kinds of cost, and they belong in separate columns:

  • Money and capacity: extra charges, replacement equipment, financing, and whether there is enough volume to go round.
  • Time and compatibility: migration, retraining, credentials, data conversion, standards, and the wait for recovery.
  • Legal and organizational exposure: permissions, penalties, coordination, retaliation, and dependence on other gatekeepers.
  • Bodily and social cost: exhaustion, disrupted care, lost shelter, worsening health, and work quietly moved onto someone else.

Resist the urge to add the columns into one impressive decimal. Nobody has supplied the exchange rate between a lost afternoon and an interrupted course of treatment, and anyone who offers one is selling something.

Shared standards and infrastructure make ordinary life cheaper and leaving more expensive, both at once. That is a trade-off to inspect, not a scandal to announce. Concentration alone does not prove abuse. Duplication alone does not prove freedom.

Then there is the clock. An exit that opens after you have run out of money is an exit in theory only. The second wager here is that an alternative available before your deadline protects you far better than the same alternative arriving after it. Measure the road, and the reserves needed to reach it.

A worker can be hard to replace and unable to miss a paycheque. The employer can need that worker badly and still sit on months of cash. Being indispensable does not pay the rent during the weeks it takes to prove it.

Being indispensable does not pay the rent during the weeks it takes to prove it.

This is where The Sucker-Punch Clock, Humane Tempo and The Domus Account come onto the map. They find the time and unpaid care hidden inside an exit that looks, on paper, perfectly available.

IV. Three cases

These three were chosen with their endings already known. They show how the mechanism works, not how often it wins. Demonstrations, not verdicts.

A. The expensive detour: SWIFT and sanctioned Iranian banks, 2012

Following an EU decision confirmed by the Belgian Treasury, SWIFT announced that it would disconnect sanctioned Iranian financial institutions on 17 March 2012. SWIFT does not hold anyone’s money. It carries the standardized messages that let banks move it. That turned out to be plenty.

The gate had two sets of hands on it: the operator, and the authorities able to compel the operator. The detour was brutal. Replacing an established, widely accepted network and the relationships that ran through it was not a matter of downloading a different app. Another messaging channel was not access to the same banks, and Farrell and Newman note how much restoring that access mattered in later negotiations.

The fine print is real. Wider sanctions, departing correspondent banks and ordinary legal fear were all in play, so Iran’s losses cannot be hung on SWIFT alone. The case shows a very expensive detour. It does not put a price on it.

B. The cheap detour: LINX LON1, 2021

On 23 March 2021, LINX LON1 went down, and RIPE Atlas watched the internet react. In its sample of 26,258 IPv4 and 14,751 IPv6 source–destination pairs, roughly 20% briefly failed to reach their targets. Then traffic found other roads, and reachability came back.

Here is a big dot that proved avoidable. For the routes that kept or quickly regained connectivity, the detour already existed and was quick to take. “Central” in this case means prominent as an observed routing intermediary, not top of a computed ranking. The detour was not free: other transit can cost more and longer paths add delay, and someone had already paid for the redundancy that made switching possible. Nobody priced that investment and nobody measured bargaining power, so the lesson is about short-run operation, not every participant’s bill.

A toy network makes the same point with arithmetic. Join four terminals to a hub by links costing 1 unit each, then connect every terminal directly to every other at 2.1 units, with capacity to spare. Every shortest route runs through the hub at a cost of 2, so the hub’s weighted betweenness is maximal. Remove the hub and each pair takes its direct link for 2.1, a 5% increase. Maximum centrality, trivial detour. The costs are invented; the model proves only that the combination can exist, not how often it does.

C. The chokepoint workers built: Flint, 1936–1937

Contemporary reporting described workers occupying the Fisher Body plants, preventing the removal of dies, and running internal committees and outside supply lines. That reporting was labour reporting, with a side, not a neutral census. The agreement of 11 February recognized the UAW as bargaining agency for employees who were union members, and traded renewed bargaining for evacuation and restarted production.

Nobody handed those workers a chokepoint. They built one. Scattered labour became coordinated control of the sites and the machinery; occupation blocked replacement; shared provisioning let people keep refusing. Virginia Tech’s historical introduction also credits the occupation of Chevrolet No. Four, and Governor Murphy’s refusal to enforce the evacuation injunction with troops. Position mattered. So did a governor who declined to send in the soldiers.

What Flint shows is temporary bargaining capacity, not permanent worker control. Wider organizing, the negotiations and the conduct of the state all belong in the explanation.

V. The Chokepoint Test

Start with the flow. “The platform is powerful” is a conclusion looking for its paperwork.

  1. Bound the case. Name the flow, the people affected, the gatekeeper, the boundary and the interruption window. Separate the operator from the owner, the regulator and whoever can compel either.
  2. Describe the position. Say which network and which centrality measure. Keep traffic, connection counts and dependence in separate boxes. Record missing links instead of quietly drawing them out of existence.
  3. Inspect the real alternatives. Check routes, capacity, eligibility, compatibility and shared upstream dependencies. Two suppliers using the same indispensable gateway are one exit in two envelopes.
  4. Price the detour before you know the ending. Record money, delay, risk, coordination and bodily cost. Separate the sunk cost of building redundancy from the cost of using it. Mark unknowns as unknown.
  5. Measure endurance. Reserves, stock, available care, tolerance for interruption, enforcement, appeal, and how much the gatekeeper depends on the people at the gate. Do not infer any of it from who later won.
  6. Watch the demand and the result. Record what was demanded, threatened, restricted or conceded, and which demands failed. Tell exclusion, rent, data collection and compliance apart.
  7. Follow the costs and the repair. Who absorbed the damage, who could push it back upstream, and whether the fix built a new road, handed the gate to someone else, or simply renamed the office.

This is a worksheet. Filling it in no more proves the doctrine than filling in a tax form proves the tax code is fair.

VI. How this argument can lose

The argument lives or dies inside comparable cases. SWIFT, LINX and Flint cannot be ranked on one coercion scale simply because they share a web page.

The fair test is dull, as fair tests should be. Take comparable service relationships in one sector. Before looking at who won, record centrality, substitutes, switching costs, gate discretion and both sides’ reserves. Then see whether the price of the detour improves prediction beyond centrality and the usual rivals, on cases the model has not already seen. Include failed demands and small gatekeepers. A sample made only of famous monopolies has already voted. Better still, find a change that makes the detour cheaper without removing the gate, such as a new route or an interoperability rule, and see whether the gatekeeper’s terms soften.

If they do not, the thesis should be narrowed or dropped. Law, wealth, ownership, violence, legitimacy, quality, concentration and private information are all serious rival explanations, and some of them build detour costs of their own. If a centrality measure that already accounts for substitutes predicts just as well, the dependence insight survives and the claim to a separate variable goes quietly in the bin.

A regression can also remove the mechanism and then announce it was never there. Decide the causal question first.

VII. Return pressure has to be built

The third wager: people who can already organize, hold reserves, reach decision-makers, control a necessary input and survive retaliation are better placed to push costs back upstream and win relief. Call it capacity to force return, the ability to make consequences land on the institution that shipped the cost out.

It does not mean harm finds its own way home. A household can absorb damage for years. Illness is a consequence; it is not necessarily pressure on the employer whose schedule helped cause it. So measure capacity before the transfer, then look separately at the pressure produced and the relief won. A strike can cost an employer money without winning its demand. A concession can move the burden onto a less organized group, which is one way a victory gets quietly financed.

The idea fails if, across comparable cases, stronger prior capacity brings neither more pressure nor more relief. Repression and failed coordination have to be studied as mechanisms, not stapled on after each defeat to keep the idea breathing.

This develops the counterflow question in Ghost Rivers and Hogging Hubs, and it adds a condition to the Anastomotic Circuit: a cost can sit in a sink indefinitely when the people carrying it cannot make its sender answer.

VIII. Who inherits the gate?

The political aim is to make refusal survivable and power contestable. The existence of a chokepoint does not settle whether using it is justified. That question is moral, and it stays moral however elegant the network diagram.

Refusal can change the network. A union can coordinate an interruption. Interoperability can lower switching costs. Public provision or mutual aid can make an alternative real. A cooperative can take over an existing gate. These are different tools: new owners do not make the detour cheaper, and a new road does not democratize the old office.

Every repair faces two questions. Who can now refuse? Who must now endure?

A counter-chokepoint needs rules for authorization, participation, appeal, recall, duration, and the protection of people who suffer the interruption without controlling it. Those are commitments to work out through The Machinery of Cooperation, not conclusions to deduce from one strike that went well.

A successful counter-chokepoint can become an ordinary chokepoint with revolutionary stationery.

The other side of the bridge

Fiction. This returns to the essay’s invented bridge.

The bridge workers had bought the crossing. They published the accounts, elected the toll committee, and lowered the price for residents. They put a sign over the tollbooth reading THE BRIDGE BELONGS TO ITS USERS. The letters could be read easily from the resident lane.

Then the second bridge closed for repairs. The cooperative voted to keep the resident rate and raise the commercial rate until the work was done. A courier named Imani, who owned her van and carried supplies between two small clinics, was charged as commercial traffic. She could pay the toll or take a detour that added four hours to each run. The committee offered her a membership application. It accepted applications from non-residents, provided them with a full explanation of ineligibility, and counted each explanation as an outreach contact.

At the next meeting she asked for a temporary rate. The chair said the bridge needed maintenance too. Imani agreed. She asked to see why her journeys cost more to carry than a resident’s. The accounts showed the revenue the committee needed. They did not show a difference between her van and the same van driven by a member. The chair called this a question of classification. Imani called it a toll.

The cooperative granted her a temporary rate and opened its toll rules to people who used the crossing but could not join. It also set aside money to keep the ferry landing usable after the second bridge reopened. That proposal had nearly failed: maintaining a possible detour was hard to justify while everyone was paying for an actual bridge. Imani still paid to cross. She could now challenge the price before her next delivery, and the ferry gave her a route if the challenge failed.

Stack-U and Stack-D cannot be told apart by counting their gates. Look instead at who may contest the gates, remove the gatekeepers, keep the alternatives open, and survive doing so.

Build the detour. Fund the time needed to reach it. Make the keys recoverable.


Receipts: evidence tier, claims and falsifiers

Evidence tier: Falsifiable. The main claims have explicit tests and falsifiers; those tests have not been run. The comparative thesis remains Asserted. Documentary facts, interpretations and illustrations are distinguished below. Evidence Protocol · Theme Explorer

IX. Claims register

Identifiers are proposed for this page. Falsifiable is the work-level tier, not a substitute for individual statuses. No new comparative study has been run for this revision. Dependencies point outward; reverse dependencies are derived.

AKM-CHK-001 — Bypass and the power to impose terms

  • Claim: Within comparable flow relationships, independently measured bypass costs improve explanation of a gatekeeper’s ability to impose terms beyond centrality alone.
  • Status: Asserted. Established dependence and network research motivates the proposition; its comparative AKM formulation remains untested.
  • Test: The sector-bounded comparison and, where feasible, change-based design in section VI; outcomes, rivals, uncertainty and the material effect threshold defined before analysis.
  • Illustration: SWIFT exclusion; the contrasting operational rerouting at LINX. These are not matched cases.
  • Falsifier: Reliable, adequately powered comparisons show no material incremental predictive value, and credible bypass-cost reductions leave term-imposition unchanged. Comparable performance by a substitution-aware centrality measure narrows the claim of distinct added value.
  • Scope and rivals: Network-mediated control of specified flows within stated time windows. Rivals: law, wealth, ownership, violence, legitimacy, quality, scale, concentration, information, and centrality measures already representing alternatives.
  • Dependencies: Emerson (1962); Bonacich (1987); Borgatti (2005); Farrell and Newman (2019); Evidence Protocol.
  • Revision status: Proposed. Errata: 2026-09-14, prepublication revision. The source draft’s stronger suggestion that the absence of costly alternatives excludes serious positional power is narrowed to dependence-based control of the specified flow. Cheap exit does not erase previously obtained information or unrelated powers. Downstream effect: no universal inference from centrality to coercion, or from cheap bypass to powerlessness.

AKM-CHK-002 — Centrality, gate control and actual outcomes

  • Claim: A centrality ranking alone does not establish gate discretion, the cost of substitution, or realized coercion.
  • Status: Inferred. A methodological conclusion from the cited network research and the distinction between position and institutional authority.
  • Test and illustration: In a declared network class, compare prediction from centrality alone with prediction using gate rights, substitutes and independently measured outcomes. LINX shows extensive bypass of a prominent intermediary; the four-terminal model gives an exact high-betweenness, low-bypass counterexample; SWIFT shows institutional authority activating exclusion.
  • Falsifier: Replicated centrality-only prediction matches richer explanations across the prespecified variation in rights and substitutes, requiring the warning to be narrowed in that domain.
  • Scope and rivals: Centrality measures that omit institutional and substitution properties; not a rejection of network analysis. Rivals: measurement error, misspecified flow models, incomplete networks, a measure already encoding the omitted information.
  • Dependencies: Borgatti (2005); Bonacich (1987); Evidence Protocol. Revision status: Proposed. Errata: None published; warning narrowed to centrality measures that omit rights and substitution.

AKM-CHK-003 — Capacity to force return

  • Claim: Within comparable institutional conditions, greater pre-existing return capacity predicts greater upstream pressure and greater relief from transferred costs; these outcomes must be assessed separately.
  • Status: Asserted.
  • Test and illustration: Register organization, reserves, access, input control, coalition commitments and retaliation protection before comparable cost transfers; follow upstream losses, remedies and displacement onto other groups. Flint supplies a mechanism illustration, not an estimate.
  • Falsifier: Reliable comparisons show neither greater upstream pressure nor greater relief among groups with stronger measured prior capacity. If only pressure rises, narrow the relief claim; do not redefine success.
  • Scope and rivals: Organized responses to identifiable transfers; no law of automatic feedback, inevitable revolt or eventual repair. Rivals: transfer severity, repression, information, preferences, political intervention, coordination failure, selection into organization.
  • Dependencies: AKM-CHK-001; AKM-CHK-004; Ghost Rivers and Hogging Hubs; the AKM overview’s circuit account. Revision status: Proposed. Errata: None published; pressure imposed and relief obtained separated in this revision.

AKM-CHK-004 — Endurance and usable exit

  • Claim: At comparable replacement cost, alternatives available before a party’s independently specified interruption limit constrain gatekeeper demands more than alternatives arriving after it.
  • Status: Asserted.
  • Test and illustration: Compare matched relationships with measured recovery times and reserves; where feasible, examine changes in reserve support or migration time while holding service and gate authority comparable. The worker with scarce skills and no rent buffer is hypothetical; Flint illustrates organized endurance without isolating its effect.
  • Falsifier: Repeated well-measured comparisons find no material difference in concessions or failed demands across the deadline boundary, and credible improvements in timely access produce no corresponding change.
  • Scope and rivals: Time-sensitive dependence; not a claim that cash alone measures endurance. Rivals: quality, skill scarcity, legal protection, repression, urgency, household support, organization, measured before outcomes.
  • Dependencies: AKM-CHK-001; The Sucker-Punch Clock; Humane Tempo; The Domus Account (these orient the question; their rhetorical force is not evidence). Revision status: Proposed. Errata: None published; new identifier for the explicit endurance proposition.

Case status and what would weaken each reading

  • General: Illustrations, not completed tests. Cases were selected with outcomes known. Contrary findings are prospective criteria for further investigation, not a retrospective claim to preregistration.
  • SWIFT, 2012: Measured (documentary event) for the disconnection; Inferred for the high cost of replacement. No independent price of bypass is estimated. Weakened by transaction records showing affected institutions promptly replaced the service at comparable cost, reach, reliability and lawful usability, with the observed disruption coming entirely from other restrictions.
  • LINX LON1, 2021: Measured (published network observations) for the rerouting; Inferred for comparatively low operational bypass. The study neither prices the prior redundancy investment nor measures bargaining power. Weakened by application-level or billing records showing sustained, materially unacceptable losses among apparently recovered connections. The four-terminal model is Modelled: stipulated costs, not measured prices at LINX.
  • Flint, 1936–1937: Measured (documentary record) for the occupation and agreement; Inferred for the counter-chokepoint mechanism. Weakened by production and management records showing GM could promptly restore equivalent output without the occupied facilities, and that the settlement followed pressures unrelated to the interruption.
  • The bridge in section I: Heuristic. Its traffic count is invented.

Revision prepared 2026-09-14; voice revision 2026-09-17. Historical illustrations are bounded to their dates. No original measurements, causal estimates, preregistered results or completed falsifier tests are claimed. Normative commitments concern the governance of power; they do not acquire empirical status because the cases are persuasive. The 2026-09-17 revision changed register and moved this apparatus into the drawer; it did not change any claim, grade or falsifier.

X. Sources

← Ghost Rivers and Hogging Hubs · AKM overview · The Machinery of Cooperation →