The book makes one big claim — that ideas get distorted as they travel from reality out to a general audience — and to explain how, it builds up a handful of specific terms. This page is where those terms live, in plain language, so you never have to take one on faith. If a chapter loses you on a word, come here, then go back. That’s what it’s for.
Entries are grouped by where each term enters the argument rather than alphabetically, so reading top to bottom roughly follows the book’s own path; use your browser’s find for a direct lookup. Each entry links to the chapter or note that works the term through in depth. To see how the terms connect — which chapters cover which, where each claim gets its source backing, what’s still open — try the interactive ontology map.
The pipeline
Transport. The work of re-encoding an idea for the next stage of its journey — compressing a finding into a news story, a story into a meme. Transport is lossy: every re-encoding strips some information, and what gets stripped is usually the decoding instructions (methodology, qualifications, scope) rather than the bare claim itself. See transport-vs-selection.
Selection. A separate mechanism operating alongside transport. At every stage, a gate decides which ideas pass through at all, using local criteria — what’s publishable, newsworthy, shareable, clickable. In modern digital media, selection is the dominant force shaping what spreads. See Chapter 5b.
Selection gate. The choice mechanism at each pipeline stage. Each gate has tunable criteria — they are institutional choices, not laws of nature, which means they can be designed differently.
The pipeline. The sequence of stages an idea moves through from reality to a general audience: raw observation → formal finding → news → public discussion → meme. The book treats this as the unit of analysis. See Chapter 1.
Memetic fitness. The grab-bag of things a selection gate actually rewards: emotional charge, identity reinforcement, status, “tells you what to do,” novelty, controversy. None of these is coordinated, and none of them cares about whether the thing is true. See transport-vs-selection.
Option space. The menu of forms an idea could take at a given stage. Transport generates the candidate variants, the medium sets which are even possible, and selection only picks among them — it never adds to the menu. That’s why the medium, which writes the menu, is the higher-leverage and less visible lever. See medium-and-manipulation.
Manufactured content. Content injected into the pipeline at later stages without ever originating in measurement — propaganda, AI-generated text, agenda-driven framing. The pipeline’s gates cannot reliably distinguish manufactured from measured content, which is what makes the distinction load-bearing.
The medium
The medium (in this book’s sense). Not the substrate (paper, airwaves, fiber) but the gate-criteria the substrate enforces. Different media reward different content; changing the medium changes what can win, what gets rewarded, and what receivers come to want. See medium-and-manipulation.
Technology vs. medium (Postman). The technology is the machine or tool itself; the medium is the whole environment of habits, rewards, and expectations that grows up around how the technology gets used. This book’s “medium” always means the second. The same technology can host very different media — which is why you can’t blame the hardware for what is really a pattern of use. See Chapter 6.
Medium shapes want. Different media cultivate different appetites in their audiences over time. A medium that rewards shallow, fast engagement trains people to want shallow, fast content; one that rewards sustained attention does the opposite. McLuhan’s “the medium is the message,” restated at the level of what audiences come to want. See medium-and-manipulation.
The receiver side
Preconditions. The mental equipment a receiver must already hold in order to decode an idea correctly — vocabulary, frameworks, contextual knowledge. An idea is “complex” in the book’s sense when its preconditions are large or rare.
Transferable preconditions. Field-general skills — statistical literacy, careful reading, the habit of asking for evidence — that help across many topics at once. The book’s first guess at what makes someone a good bridge between fields; later refined into versatile expertise (depth plus those habits, not the habits alone). See bridge-nodes-and-versatile-expertise.
The three layers of a message. Hofstadter’s model, adapted: every message has a frame layer (this-is-a-message), an outer layer (how to decode), and an inner layer (the content). The book’s load-bearing move: the outer message is what it otherwise calls preconditions, and it lives in the medium. See three-layer-message.
Outer message. The decoding instructions that come (or fail to come) with a message — what tells you how to read the inner message. Most pipeline distortion is loss of the outer message, not the inner one. The claim survives transit; the key it needed doesn’t.
Receiver budget. The fixed cognitive bandwidth any reader has for incorporating new material — a tablespoon of attention-weeks. Trainable through shared vocabulary, education, and accumulated engagement, but always finite. See Chapter 3.
Want. The receiver’s standing appetite for engaging with material. The book argues want is the prime mover behind both transport and selection — receivers engage with what they want, and the gates reward what enough receivers want. See transport-vs-selection.
Complexity ceiling. The most complexity an idea can carry and still cross a given network. The bigger the network — and the more compression and selection pressure it applies — the lower the ceiling. See Chapter 5.
Handle-ability. A rival to “complexity” for what actually decides whether an idea spreads: not how many preconditions it needs, but how easily it plugs into mental models people already have, how much feeling or identity it carries, and whether it tells them what to do. A property of the selection side, not the idea itself. See Chapter 5.
Compressed form / complex form. Two versions of the same idea. The compressed form is cheap to process and easy to pass on (and easy to weaponize); the complex form is expensive — it asks the receiver to load preconditions and work through implications — but it keeps what the compressed form drops. See Chapter 5.
Compression and truth
Compression. What happens to an idea as it traverses lossy transport. Compression strips the outer message (the decoding key), leaving the claim intact but underdetermined.
Interpretive latitude. The range of valid readings a compressed claim supports across a population. Compression doesn’t just shift the expected decode; it widens the variance of decodes — different receivers reach for different keys and arrive at different inner messages. See Chapter 5c.
Truth-value placement (the three regimes). The book’s framework for what happens to truth under compression. The same compressed claim can be preserved truth (the receiver’s decoding key is close to the original), inverted truth (keys oppose — a true claim decoded into a false belief), or orthogonal truth (the claim is decoded as an identity flag rather than a claim at all). Which regime occurs is set by the selection gate. See Chapter 5c.
Three realities (Harari). Three kinds of thing a claim can be about: objective (true regardless of what anyone thinks — atoms, planets), subjective (true only inside one mind — a pain, a dream), and intersubjective (true because enough people agree — money, nations, laws, brands). The pipeline treats them differently: it transmits-or-distorts objective claims, mostly skips subjective ones, and partly creates intersubjective ones — so for the last kind the failure mode is capture of the agreement-making process, not loss in transit. See Chapter 2 and intersubjective-truth.
Intersubjective truth. A truth generated by the network’s agreement rather than by reality. “The twenty in your pocket buys groceries” is really true — but erase the agreement from every head and nothing is left out there to re-measure; the truth was the agreement. The book keeps the word truth for these (Harari’s terms): what differs is the generator, not the truth-ness. The carving is per-claim, not per-artifact — a myth can be based in reality too, and most artifacts are mixed bundles. Because the pipeline reaches into heads at scale, it is wired into the generator: for this cargo, transmission is participation. See intersubjective-truth.
Constitutive regimes. What the three truth-regimes become when the cargo is intersubjective: convergent constitution (the decodes compose into one determinate agreement — most money and law, most of the time), divergent constitution (camps affirm the same words as different agreements — a constitutional crisis), and hollow constitution (the words stay affirmed while the constituting behavior evaporates — everyone still calls it money, nobody holds it). The failure axis is spread-and-substance rather than error: an intersubjective truth can’t be wrong, but it can split or empty. See intersubjective-truth.
Reification. Treating a made thing as a found thing: an intersubjective truth processed as if it were objective — “money just is valuable,” “the market decided,” “the law is the law.” The mirror image of the orthogonal regime’s migration, and not innocent: an agreement that reads as a law of nature has a generator nobody audits, which is exactly the camouflage a captured generator wants. De-reification — saying out loud that this is a decision, not physics — is a political act. See intersubjective-truth.
The fork. What stands in for the corrective when there is no referee: an intersubjective truth cannot be falsified, only renegotiated, abandoned, or forked — schism, secession, dollarization, chain split. A fork doesn’t determine which side was right; it determines how many truths there are now. The design lesson for institutions: keep the fork affordable, so a captured version loses its constituency fast instead of holding it hostage. See intersubjective-truth.
Constitutive volatility. The transport collapse’s second blade. Agreement states used to carry inertia — a bank run moved at the speed of queues on the pavement — and now move at meme speed (a bank run coordinated in group chats), while the institutions that steady agreements still move at committee speed. A stabilizer slower than the thing it stabilizes is a spectator with a title. See intersubjective-truth.
Selection up close
Algorithm as selection engine. A recommendation algorithm isn’t lossy transport — it’s a selection gate with adjustable weights. That makes “who tunes the gate?” a literal, operational question, and turns a few companies’ criteria into everyone’s. See Chapter 5b.
Manipulation surface. What an attacker can exploit. In the pure-transport view it grows with network size; in the fuller view it’s the tunable criteria of the selection gates — the more dangerous version, because the weights can be set on purpose. See Chapter 5b.
Superspreader dynamics. A handful of high-connection nodes drive a wildly disproportionate share of what spreads — a network-structure effect with no offline equivalent at the same scale. About who is positioned to amplify, distinct from what can be exploited. See Chapter 10.
Meme ecosystem competition. Memes don’t spread in isolation; how fast one catches on and dies depends on the others competing for the same attention. Fitness is relative to the current crowd, not a fixed property of the meme. See Chapter 5b.
Neutral drift baseline. The null model for spread — what propagation looks like with no selection at all: random copying plus rich-get-richer. Most variants follow it; selection is what you detect as a departure from it. See Chapter 5b.
Replicator ontology. The assumption that culture comes in discrete, copyable units (memes, like genes). The founding premise of memetics — and the thing its critics attack hardest, arguing cultural transmission is reconstructive rather than copy-based. See Chapter 5b.
False consensus among rational agents. O’Connor and Weatherall’s result: a group of people who each weigh evidence correctly — no stupidity, no gullibility — can still settle together on a false belief. The failure lives in the network’s structure, not in individual error. It undercuts any diagnosis that blames careless receivers. See Chapter 5b.
Credibility weighting. The rational, unavoidable habit of trusting evidence more when you trust its source — it’s how anyone copes with not being able to check everything first-hand. It’s also the seam propaganda exploits: manufacture or buy credibility and the weighting carries false evidence as readily as true. See Chapter 10.
Conformity. Adopting a belief because your peers hold it, rather than because of the evidence for it. It can lock a community onto a belief — true or false — and seal it off from correction. See Chapter 5b.
How people hold beliefs
Epistemic vigilance. The cognitive defenses people use to screen incoming claims for plausibility and source-reliability — “mental immune systems” in the popular phrasing. Mercier sharpens it into open vigilance (below). See Chapter 7.
Open vigilance (Mercier). Mercier’s correction to the “people are gullible” story: we are both open to useful information and on guard against unreliable sources, and the two evolved hand-in-hand. If anything we err on the cautious side — more often failing to trust something we should than swallowing something we shouldn’t. See Chapter 7.
Reflective vs. intuitive belief (Sperber). Two ways to hold a belief. Intuitive beliefs are wired into your other beliefs and actually drive what you do. Reflective beliefs are held at arm’s length — stated, shared as identity-signals, but inert when it comes to action. Much of what spreads online (rumors, conspiracy claims, slogans) is held reflectively, which is why people share things they would never act on. See Chapter 7.
Justification market. Mercier’s reversal of the obvious arrow. The naive story: content shapes belief, belief drives behavior. The reversal: behavior is mostly decided upstream (by identity, group, interest), and people then shop for content that justifies what they were already going to do. So a lot of “mass persuasion” is really suppliers meeting a pre-existing demand for justifications — which changes what any fix has to target. See Chapter 7.
Paradox of virality. What spreads is not what people say they want spread. Across the political spectrum, people report not wanting high-arousal negative content to go viral — and it goes viral anyway. The gap points at the selection gate (especially algorithmic amplification), not at receiver preference. See Chapter 7.
Networks at scale
Myth dilution. The loop by which a network’s binding myth loses determinacy as the network grows: scale forces the myth to compress, compression widens its interpretive latitude, latitude sorts the network into segments that affirm the same myth while decoding it differently. See myths-scale-and-bureaucracy.
Bureaucracy (in this book’s sense). Used narrowly throughout the book — by stipulation, not by argument — to mean the institutional repair a scaled network grows to re-supply the decoding key the myth dropped: the preservation function (keeping the un-compressed form alive somewhere) plus the training function (re-installing the decoding key in receivers). Real-world bureaucracies do more (coerce, tax, allocate, record); the book holds those out of scope to keep the preserve-and-retrain mechanism in focus. The honest alternative would be to call the narrow function institutional carriers throughout, but the book inherited the term from Harari’s Nexus and has kept it for continuity. See myths-scale-and-bureaucracy for the mechanism and Chapter 8 for the two-function split.
Network segmentation. The fracturing of a network into sub-groups along the seams of its myth’s interpretive latitude. The result of myth dilution: the network is superficially unified and structurally fragmented at once.
Echo chambers vs. bubbles (C. Thi Nguyen). Bubbles are missing-information structures; echo chambers actively discredit outside information. Small networks tend toward chambers, not bubbles, which changes what “infrastructure for integration” has to do.
Polarization via distrust. O’Connor and Weatherall’s formal result: when agents discount evidence from sources they distrust, communities with any initial difference rationally diverge — each side discounts the other’s findings, the gap widens, the divergence deepens the distrust that drove it. No irrationality or misinformation required. Adding a channel between distrustful communities can accelerate divergence, not reconcile it.
Zollman effect. Increasing connectivity inside a community can reduce the accuracy of its beliefs: densely connected agents see each other’s preliminary, noisy results too quickly and converge prematurely on whichever answer took an early lead. More connection is not monotonically better; there is an optimum.
Inverted-U relationship. A shape that appears repeatedly in the book’s argument — not “more is better” or “less is better” but an optimum density / size / barrier-height with diminishing returns on both sides.
The bridge zone
Bridge zone. The stretch between deep specialists and mass audiences — where journalists, popularizers, influencers, and AI summarizers live. Selection pressure is most intense here, and the reshaping is active (forms built for clicks and approval), not just passive blurring. This is where most of the catastrophic distortion happens. See Chapter 6.
Pseudo-context (Postman). A use invented for information that has no real use left. The information may be perfectly true, but it’s been stripped of the context that connected it to action — so the trivia night, the news quiz, the conversation-filler paper over the gap. Different from manufactured content (which never measured anything): pseudo-context is real information made idle. See Chapter 6.
Political economy of attention
The modality argument. Why selection has an owner and transport doesn’t. Transport loss is a cost — nobody chooses it, so there is nobody to blame and nothing to tune. Selection runs on criteria, and criteria have setters. That asymmetry is what makes the political economy of attention structural rather than accidental: wherever there is a gate, someone owns its dial. See Chapter 10.
Engagement as a captured equilibrium. Engagement-maximization isn’t an outside attack on a platform — it is the platform’s business model doing its normal job. Time-in-app sells ads, so the gates get tuned toward whatever holds attention (identity, threat, outrage). That’s what makes it so stable: there’s no villain to defeat, only an arrangement to dismantle. See Chapter 10.
Out-competition of institutional carriers. Platforms don’t have to censor journals, universities, or depth journalism to hollow them out — they just out-compete them. Engagement-bait costs roughly nothing per view (users make it); depth costs real money per view (someone is paid to make it). Given limited attention, audiences predictably take the free option. This produces a slow institutional collapse with no direct attack. See Chapter 10.
Cost-shifting from producers to consumers. When barriers to entry fall, the cost of quality assessment doesn’t disappear — it transfers from producers (who previously had to invest in overcoming barriers) to consumers, reviewers, and maintainers who have no more time. The diagnosis underneath the modern noise problem.
Creation cheap / review expensive. When making content scales far faster than checking it, noise wins regardless of how many gems are in the flood. (LLMs widen the gap from both ends.) See Chapter 10.
The Orwell/Huxley axis (Postman). Two opposite ways an information environment fails. Orwell: control by inflicted pain — censorship, surveillance, information withheld. Huxley: control by inflicted pleasure — the truth isn’t hidden, it’s drowned in a sea of engaging irrelevance. The book’s diagnosis is squarely Huxley, which is harder to fight because there is nothing to push against. See Chapter 10.
Capture, unified
Capture. The mechanism behind much of the book’s alarm: tuning a selection surface against the very thing it was supposed to serve. The taxonomy sorts every case along three axes — which surface is captured, by whom (an outside adversary, the institution’s own business logic, or both), and how hard it is to recover (set by which surface). See capture-taxonomy.
Self-capture vs. external capture. Two sources of capture. External: an outside actor (a propaganda shop, a state, an ad-tech firm) learns the gate’s criteria and games them — an arms race you can at least fight. Self: the institution’s own business model is the thing tuning the gate against truth — no adversary, just an equilibrium, which is why it’s more stable. The common real case is composite: self-capture lays down a slope that external actors then ride. See capture-taxonomy.
Consumer-key vs. surface capture. Why some captures are far worse than others. Capturing something that installs decoding equipment in people’s heads (what they’re taught, what an AI is trained on and for) damages the people and can’t easily be un-installed. Capturing something that only shapes the surface a person meets (which headlines get shown, how a tool is configured) damages only the surface. The first kind is the dangerous kind. See capture-taxonomy.
AI as a new kind of node
Selection-design surface. Any place where a design choice shapes what content can exist or travel: a gate’s criteria, the menu of possible forms, the training data, the training objective, the deployment settings. Historically these were spread across separate institutions — a newspaper held one, a school another. The point of the term is to make concentration countable. See Chapter 11.
LLM as a collapsed design moment. A language model is the first node in the pipeline to own almost everything at once — the gate, the menu of options, the content generator, the training data, the training objective, the deployment settings, even the price tier — all in one place, owned by one party. A social-media platform owns two or three of those; an LLM owner owns five to seven. That concentration is why the book treats it as the limit case of its political-economy worry. See Chapter 11.
Decompression on demand. Used faithfully, an LLM can take a compressed claim or a dense source and re-expand it — supplying the preconditions a particular reader is missing. This is the first real way out of the fixed receiver-budget: the reader’s hours haven’t grown, but the model pays the decompression cost. Strictly conditional on faithfulness — a hallucinating model supplies confident-but-wrong instructions, which is worse than none. See Chapter 11.
Compressed form becoming authoritative. The inverse failure: at scale, the AI’s summary of a source quietly replaces the source in everyone’s working memory — now carrying the AI’s authority. The original still exists; nobody reads it. Compression turns into substitution rather than abstraction. See Chapter 11.
Three ways to capture an LLM. Corpus capture — rigging what data it learned from. Objective capture — rigging what it was trained to optimize; the worst kind, because its outputs feed the next model’s training data, baking the bias across generations. Deployment capture — changing its behavior at use-time via system prompts and policies; the cheapest and most opaque. See capture-taxonomy.
Substrate custody. Who holds an AI’s formative surfaces — its training data (corpus), what it was optimized for (objective), and how it is configured in use (deployment). An institution that needs a faithful model has to hold all three itself; renting them from a commercial vendor means trusting the vendor’s tuning. Every hopeful LLM scenario in the book is conditional on this. See Chapter 11 and Chapter 12.
LLM as capability extender. The hopeful case from Chapter 11, made concrete in Chapter 12: an institution that owns its model’s data, objective, and deployment can use it to extend preserve-and-retrain — decompression on demand against its own archive, tutoring at scale, drafts that real experts then verify. Technically doable, economically marginal, and politically blocked by the same forces that captured social media. See Chapter 12.
The bridge / prescription
Bridge node. A person or institution that successfully moves complex truth between networks that don’t share preconditions. The book argues bridge nodes are not generalists but versatile experts. See bridge-nodes-and-versatile-expertise.
Versatile expertise. Deep specialist expertise paired with metacognitive flexibility — the trained habit of looking for structural analogies, suspending one’s primary paradigm, treating one’s own frame as one option rather than the substrate of reality. The bridge-node mechanism: depth is what gets the bridge admitted as a peer on either side; flexibility is what lets it translate across.
The curse of expertise. The four cognitive failure modes of deep specialization without paired flexibility: paradigm lock-in, overconfidence in cross-application, perceptual filtering, conceptual rigidity. Together they are the cognitive substrate of polarization-via-distrust — same mechanism described from the inside of one head rather than across a network.
The abyss. The possibility space of a sufficiently complex field — every move, question, or direction it could go — when that space is so vast no one, expert or machine, can navigate or even fully perceive it. It takes real competence in a field just to see that its abyss is there; novices mistake the field for something finite. The experiential complement of the curse of expertise: the curse is depth that hasn’t yet seen the abyss, the bridge node is depth that has. Distinct from optionality, which is the approachability of a finite, masterable artifact — a software library is not an abyss, it is a map of one. The abyss is why compression exists at all. See the-abyss.
Preservation vs. training. What the book’s narrow “bureaucracy” splits into. Preservation keeps the full, un-compressed form alive somewhere (archives, journals, specs). Training re-installs in people the ability to decode at that resolution (teaching, mentorship). They work as a pump — preservation holds the pressure, training releases it into the network — so either one alone fails. See Chapter 8.
Capture asymmetry. When these institutions get captured, the training side does more damage and heals more slowly than the preservation side. A captured archive holds a tilted copy you can re-check against other sources; captured training tilts the readers themselves, and a re-tuned readership reads even good new evidence through the bad key. So training is the half to defend first. See Chapter 8.
Differential decay rates. Preservation institutions rot slowly — an archive can coast for decades before anyone notices. Training institutions rot within a generation — stop producing skilled people and the supply runs out in twenty or thirty years. Another reason training is the more time-sensitive half. See Chapter 8.
Curation layer. Institutional structures that preserve the gatekeeping function (separating signal from noise) while removing the access barrier (who gets to participate). Wikipedia editorial, Stack Overflow reputation, preprint-plus-peer-review hybrids. The infrastructure side of the integration prescription. See the-democratization-paradox.
Survivable polarization. A design rule for bridge institutions: build them to work even when trust is partly broken, instead of needing trust to be whole. In practice: decisions a skeptic can audit, reputation that survives partial distrust, real experts inside (not presiding over) the institution, and several independent paths to the same answer. See Chapter 12.
Funding decoupling. The structural counter to out-competition: an institution can’t survive if its revenue depends on winning the free-engagement market — it’ll lose to zero-cost noise. The survivable shapes fund it from outside that market: endowments, public funding, member dues, public-goods mechanisms. See Chapter 12.
Survival, not victory. The reframe at the heart of Part IV. Integration institutions don’t need to beat the engagement machine or clear the attention market — they need to keep doing preserve-and-retrain alongside it without being eroded. Trying to win (by being more engaging) is exactly how they get captured. See Chapter 12.
Work of generations. The closing reframe: institutions of integration are civilization-scale investments on the timescale of universities, journals, and legal systems — not product launches. The captured equilibrium built itself in two decades; what holds against it has to be funded and protected over at least as long. See Chapter 12.