Three chapters flagged this debt by name. Chapter 2 introduced Harari’s three realities and admitted its treatment of the third was “a sketch rather than a full treatment”; Chapter 5c built the truth machinery and confessed it had waved at the intersubjective case with Harari’s frame rather than doing it properly; Chapter 9 scoped the intersubjective case out and called the missing work a sibling treatment not yet produced. This is that working-through. It does for constitution what 5c did for transmission: says what the pipeline does to this kind of truth, names the failure regimes, and works out what capture, correction, and speed mean when the cargo is the kind of thing the carrying makes real. One commitment up front, because it disciplines everything after. The book keeps the word truth for intersubjective content, on Harari’s terms: “the dollar has value” is not a half-truth or a useful fiction, it is true, testable at a bakery. What sets it apart from “the Earth goes around the Sun” is not how true it is but where the truth is generated: objective truths are generated by reality, intersubjective truths by the network’s agreement. Same word, different generator, and the whole note is what that difference does once a pipeline gets involved.
Two truths, one experiment
Take two true sentences, the Earth goes around the Sun and a twenty-dollar bill buys groceries, and run the cruelest experiment available: erase every belief about each from every head. The first doesn’t notice. The Earth keeps orbiting, the fact sits out there waiting to be re-measured by whoever next builds a telescope, which is what Chapter 2 meant by objective reality having a fact the pipeline can garble but cannot touch: the fact is the referee, and it outlives any number of bad calls. The second dies instantly. The bill is still paper and the ink is still ink, but “this buys groceries” stopped being true the moment the agreement stopped being held. There is nothing to re-measure, because the truth was never out there; it was distributed across a few billion minds, and erasing the minds erased it. Both sentences were true. One had a generator the network can’t reach; the other is the network’s reach.
The generator, not the genre
The tempting next move is to sort content into bins: science (objective), money and nations (intersubjective), horoscopes (a third pile). The bins don’t hold. The carving is per-claim, by generator, not per-artifact, by genre, because most artifacts that move through the pipeline are mixed bundles. A nation’s founding story is the clean example: it bundles objective claims (battles, dates, treaties, checkable against a record the story can’t alter) with the intersubjective content the story exists to carry (the nation, its legitimacy, who counts as us). A myth can be based in reality too; what makes the nation intersubjective is not that its story contains falsehoods but that the nation’s existence is generated by the agreement the story coordinates, while the battle’s existence was generated by the battle. Run the same blade through astrology and an old puzzle resolves. Chapter 2 filed it under manufactured content, but doesn’t the agreement of its believers make it intersubjective? Split the bundle: astrology’s personality claims (Mars in retrograde shapes your week) are objective claims about how the world works, generated by nothing, checkable and false, the manufactured content Chapter 2 filed, while “I’m a Capricorn” as a membership badge is agreement-generated and real the way brands are real. False objective claims riding on a true intersubjective substrate, and the gates can’t tell the layers apart.
One distinction is easy to state and easy to lose: every intersubjective truth casts an objective shadow, since that the agreement holds is itself an ordinary objective fact about the network. “Most merchants accept dollars” is objective, checkable by survey, a claim about the agreement; “the dollar has value” is intersubjective, generated by it. The two travel together and are easily confused, and the confusion matters: economists, pollsters, and chain explorers measure the shadow all day without touching the generator. Reporting on a constitution is transmission. Participating in one is not.
The pipeline is wired into its own cargo
For an objective claim the pipeline sits outside the fact: Chapter 1’s stages compress it, the gates pass the wrong variant, the receiver decodes with the wrong key, and through all of it the fact sits untouched on the far side, available for the next replication to appeal to. Transmission and reality are separate systems, which is what every corrective in Chapter 2 was banking on. For an intersubjective claim the separation does not exist. The truth is the agreement, the agreement lives in the network’s heads and habits, and the pipeline is the thing that reaches into heads at scale, so a receiver decoding “the dollar is sound” is not just reading about the agreement; their decode becomes part of its state, one more mind holding or dropping the thing whose being-held is the truth. The pipeline does not report the agreement; it is one of the places the agreement happens. For intersubjective truth, transmission is participation in constitution. With objective facts the pipeline is a security camera pointed at the world, filming badly but unable to change what happened; with intersubjective truths the camera is part of the crime scene, and pointing it somewhere makes things happen there. Two consequences fall out at once. Reach changes the truth itself, not just who knows it: a currency held by ten people and by a billion are different monetary facts, since the billionth holder didn’t learn the dollar’s value, they added to it, so Chapter 5’s virality machinery sits, for this cargo, upstream of the truth rather than downstream. And the famous reassurance fails: bend the agreement-making process and you have not gotten the truth wrong, you have manufactured a different truth, with no residue of the old one left to check the new one against.
The three regimes, translated
Chapter 5c gave the truth-outcomes for transmission: a compressed claim lands preserved (receiver’s key close to the original), inverted (keys opposed, true claim, false belief), or orthogonal (read as an identity flag, not decoded). All three assume a fixed generator the decode can be scored against. Drop that (let the decodes feed the generator) and the regimes translate rather than transfer, the failure axis rotating from error (how far is the belief from the fact?) to spread and substance (do the decodes still compose into one agreement, and is it still doing real work?).
Convergent constitution, preservation’s sibling, is when the compressed binding claim lands close enough in enough heads that the decodes compose into one determinate agreement: the truth is settled in the only sense settled can mean here, the network generating it together, which is most money and most law most of the time, and deserves the concession 5c gave preservation: most checks clear. Divergent constitution, inversion’s sibling with the geometry changed, is the failure an intersubjective truth actually admits, since it cannot be inverted (there is no external fact to be collectively wrong about) but it can split: the decodes scatter, pool by community, and harden into camps affirming the same words as a different agreement. That is exactly the segmentation loop (scale compresses the myth, compression widens latitude, latitude sorts into segments), and the reframe this note adds is that when the diluting claim is constitutive, segmentation is the truth coming apart, not a misreading of it. A constitutional crisis is the type specimen: both sides waving the same document, each affirming a different constitution, no fact beneath the dispute but the agreement visibly failing to be one. Hollow constitution, orthogonality’s sibling and the strangest, is a notch subtler: for intersubjective claims “decoded as a badge of us” is not automatically a failure, because affirmation is part of how nations and creeds work, so the failure is one step on, when symbolic affirmation detaches from the behavioral agreement the truth consists in. The early weeks of a hyperinflation are the clean case: everyone still calls the banknotes money while spending them within hours of being paid, refusing to hold them, which is the behavior the monetary agreement is made of. The words outlive the truth, and hollow constitution is the regime reification loves best, because a truth that has quietly stopped being generated still photographs like a working one.
Capture, when there is no truth to tune against
The medium note defined manipulation as tuning a gate’s criteria against truth. For constitutive cargo that has nothing to push off, since there is no generator-independent truth to betray, so the definition has to be restated, uglier: capture of the constitutive process is steering the generator, re-weighting whose decodes count in the agreement, toward an interest that is not the network’s. The agreement was never one-head-one-vote. Real constitutive processes weight the votes enormously: a central bank’s decode of “what the dollar is” counts more than a million households’, a supreme court’s decode of “what the law says” is the law downstream, a standards body’s decode of the protocol is the protocol. That weighting is no scandal, it is the bureaucracy holding the heavy votes that keep the agreement composed, but it means every scaled intersubjective truth concentrates its generator into capturable surfaces, and the capture taxonomy applies with the stakes one ontological level up. Surface capture (seizing the issuer, the court, the trending panel) steers the heavy votes directly and is the recoverable kind, because visible organs can be audited and replaced. Consumer-key capture, re-tuning the keys receivers decode with through curriculum and feed, is categorically worse here, and Chapter 2 said why in a line this note can unpack: the trained key is not reading equipment pointed at the reality. The trained key is a share of the reality. A captured curriculum for objective content produces receivers who are wrong about the world, and the world waits to correct them; a captured curriculum for intersubjective content produces receivers whose participation generates the captor’s preferred reality, with no waiting world and nothing to be wrong against, so Chapter 8’s asymmetry gains a multiplier: re-train the readers and you have amended the constitution, not distorted its reception.
Capture of this kind ships with camouflage, and its old name is reification: treating a made thing as a found thing, an agreement wearing the costume of nature. “Money just is valuable.” “The market decided.” “The law is the law.” Each files an intersubjective truth under objective, and the filing error is not innocent, because a generator nobody can see is a generator nobody audits. 5c named one migration between realities (an objective claim processed as if intersubjective, decoded for team rather than truth, the orthogonality regime and the engine of politicized science); reification is the same border crossed the other way, intersubjective truth processed as if objective. Orthogonality takes a claim that had an answer and makes it unanswerable; reification takes an arrangement that could be renegotiated and makes it look fixed, protecting a truth’s capture. History marks the moments the costume slips (a signature ends the dollar’s gold convertibility in 1971, a treaty wills the euro into being, and for an instant everyone sees money is a decision), and reification is the long interval between such moments, when it reads as physics, which is how whoever holds a captured generator wants it to read. This is the reflexivity George Soros built his market theory on, and the looping by which, as Ian Hacking’s work on social construction argues, a classification changes the people it classifies.
No referee, so the corrective’s job passes to the fork
Chapter 2’s objective machinery runs on correctives (replication, consensus reversal, disavowal), each an appeal to the external referee, each capable in principle of dragging belief back toward the unharmed fact, and the book’s diagnosis was that correctives lose at the gates. For intersubjective truth it is not that correctives lose; it is that there is no referee to appeal to at all. A constitution cannot be falsified. It can only be renegotiated, abandoned, or forked. The fork does, for constitutive content, the job the consensus gate does for objective content. When divergent constitution becomes unlivable (the camps’ decodes no longer compose, the heavy votes no longer recognized across the seam), the network resolves the indeterminacy by splitting into two networks, each generating its own truth cleanly again. Christianity’s Reformation is the canonical instance, and it is no accident the book keeps meeting it in the printing-press era beside the Hammer: cheap transport widened the myth’s latitude past what the bureaucracy could re-key, and the constitution forked, the way Benedict Anderson’s account of print and nations describes new communities being constituted as the old ones came apart. Secessions are the political form, dollarization the monetary one. The cleanest laboratory is the newest, public blockchains, intersubjective truth with the constitution made machine-readable: in 2016 a dispute over what the Ethereum ledger truly said after a catastrophic theft was unresolvable by measurement (both readings were internally valid; the question was which agreement the network would hold), and the network forked into Ethereum and Ethereum Classic, two ledgers, two communities, both right, each truth fully generated by the network holding it, with Bitcoin and Bitcoin Cash repeating the shape a year later. A fork does not determine which side was correct; it determines how many truths there are now, which is what a consensus gate looks like when there is no fact to converge on. That recasts the marketplace-of-ideas defense one more time: for intersubjective content the marketplace is the generator, so whatever wins the gates does not approximate the truth better or worse, it becomes the truth, and the gates’ gradient (engagement, identity, drama, the whole political economy) is not distorting an intersubjective truth in transit but choosing which one gets to exist.
Constitutive volatility: the thesis’s second blade
The book’s central historical claim (the transport note, carried into Chapter 1) is that transport cost collapsed and the collapse exposed selection. Built for objective content, run over constitutive content it says something new. Intersubjective truths feed on their own coverage, which is what being wired into the generator means, and the textbook case is centuries old: the bank run. “The bank is failing” propagates, holders act, the bank fails, and the claim was not a distortion that then came true by coincidence, because under fractional reserve the deposits’ availability is an agreement not to test it all at once, and propagating the claim dissolved the agreement. Belief-propagation manufactured its own truthmaker, the loop Soros calls reflexivity and Albert Hirschman’s exit-voice frame reads as the mass exercise of exit. The loop is old; what is new is the clock rate. When transport was slow the agreement-state of a large network had inertia (a run spread at the speed of rumor and queue, depositors physically lining up outside Northern Rock in 2007 nearly the last picture of constitution moving at body speed), and the maintenance bureaucracies were engineered against that time constant, deliberation and committees and announcements timed to morning papers. In March 2023 Silicon Valley Bank lost tens of billions of dollars of deposits in a single day, the run coordinated in group chats, constitution moving at meme speed against institutions still moving at committee speed, and the bank was gone before the slow machinery had finished convening. The transport collapse did two things, and so far only one has been named. It exposed the selection gates (the existing thesis, for objective content), and it removed the inertia that used to be a load-bearing, unpaid-for component of every large intersubjective truth (the new blade, for constitutive content). Shared realities have always been able to fall apart; hyperinflations didn’t wait for the internet. What changed is the ratio between how fast an agreement can move and how fast the institutions that steady agreements can respond, and a stabilizer that responds slower than the thing it stabilizes is really just a spectator with a title.
Where I land
Intersubjective truths are truths whose generator is the network’s agreement rather than reality. Because the pipeline reaches into heads at scale it is wired into that generator, so transmission of intersubjective content is participation in its constitution: reach changes the truth itself, and no unharmed fact waits outside a bad transit. The transmission regimes translate accordingly (convergent, divergent, and hollow constitution replace preserved, inverted, and orthogonal, the failure axis rotated from error to spread-and-substance); capture, with no external truth to tune against, becomes steering the generator, its worst form the consumer-key one and its camouflage reification; and with no referee there is no corrective, only renegotiation, abandonment, or the fork, which decides not who was right but how many truths there are now. The transport collapse cuts twice: it exposed the gates and stripped the inertia that steadied every large agreement, so constitution now moves at meme speed against institutions built for committee speed. The sentence the rest of the book inherits: for objective content the captured pipeline distorts the truth; for intersubjective content the captured pipeline decides it. That is why Chapter 2 was right to call this the scarier failure, and why the prescriptive chapters cannot treat the intersubjective case as a footnote to the objective one. The book’s chapters carry the consequences: Chapter 10’s critique cutting twice (transmission failure for objective content, generator capture for intersubjective), Chapter 11’s LLM doing key-re-supply for intersubjective realities at population scale, and Chapter 12 designing institutions that are themselves intersubjective realities, where survivable polarization is surviving divergent constitution and a cheap, visible fork is what keeps their capture from quietly holding.
Where I’m still uncertain
- The line is the live one. The generator criterion carves claims well enough for the easy bundles (founding stories, astrology), but the hard cases aren’t secure. Brands fuse quasi-objective promises to agreement-generated value; constitutional interpretation mixes a written record with live agreement in proportions courts themselves dispute; and hardest, scientific consensus is socially produced (sociologically it looks like agreement) while it has to be tracking an external generator rather than constituting one. “Consensus tracks; agreement generates” is the distinction the whole note leans on, and a hostile reader will push where the sociology of science pushes. I commit to it and flag that it isn’t yet defended at that depth.
- The regime translation may be forced symmetry. Convergent and divergent are two readings of one quantity (decode spread); hollow is a different event (affirmation detaching from behavior), the same two-plus-one shape 5c suspects of its own regimes. Hollow may just be divergence’s end state (words diverging from deeds) rather than a third cell.
- Decode-as-vote may overstate the receiver. If a central bank’s decode outweighs a million households’, then for most receivers most of the time participation is nearly notional. The honest version may be that the marginal decode matters only near criticality (runs, forks, legitimacy crises), so “transmission is participation” is almost-false in calm weather and decisive in storms, and whether that strengthens the volatility argument or undercuts the constitution one needs working out.
- The volatility claim is a ratio asserted from cases. “Constitution moves at meme speed, institutions at committee speed” rests on a handful of vivid instances (SVB against Northern Rock, chain forks against schisms). The systematic version, measuring agreement-state half-lives against institutional response times across eras, is the empirical work the book’s experiments folder exists to spec and hasn’t done.
- The philosophical literature is only now being engaged. This note runs on Harari plus the book’s own machinery, with the deep prior literature (Searle’s institutional facts, Berger and Luckmann’s social construction of reality, Anderson’s imagined communities, Hacking’s looping effects, Soros’s reflexivity, Hirschman’s exit and voice) linked here but not yet argued with. The draft is the book’s own working-out, written before that engagement rather than in dialogue with it, and reading those sources will sharpen or correct parts of it.
← Chapter 8: Preservation vs. Training · Chapter 9: The Integration Problem →