The Map is \neq the Territory Alfred Korzybski

Metaphors

We’re sense-making creatures who love metaphor.

When one lands, we stop looking. B/c when the metaphor resonates, we satisfice - ‘good enough’. But satisfying isn't the same as accuracy.

Enter Meghan O'Gieblyn.

In God, Human, Animal, & Machine, she traces how we keep projecting our latest technology onto the deepest questions — and then the projection projects back.

Bi-directionality. ♾️

Data:

Which was the original? Which was the reflection? Does it matter?

These loops — where the map reshapes the territory it was supposed to describe — are what this piece wrestles 🤼.

Maps

The pattern above isn't just historical — it's structural.

McLuhan saw it from the media side — the medium doesn't just carry the message, it becomes the message.¹

Alex Danco explains McLuhan’s hot/cool distinction with style:

The medium doesn't carry a message so much as it creates space for certain kinds of messages and messengers. Nixon was a hot candidate (in text only; sorry tricky Dicky) who sounded presidential on hot radio; Kennedy was cool, speaking in slogans that invited interpretation, fitting the cool flicker of early TV 📺.

What the audience perceived depended on the medium — the match between messenger temperature and medium temperature determined what was actually heard.

That's McLuhan's insight from the outside — the medium filtering the message before it arrives. Lakoff and Johnson showed the same thing happening from the inside: metaphor isn't literary decoration — it's the cognitive infrastructure of all thought. The structures we think with are metaphors we've forgotten are metaphors (all Christopher Nolan movies, ever).

When the framework confines understanding, it isn't illustrating your worldview — it's constituting it.

Wittgenstein tried to prove otherwise — and failed, spectacularly, against himself. In the Tractatus (5.6): "The limits of my language mean the limits of my world" — he believed language could mirror reality with perfect logical structure. Then he demolished his own position. In the Philosophical Investigations, meaning isn't reference — it's use. Words are moves in language games, not labels on objects. There is no perfect map. Only games we've learned to play, and rules we've forgotten are rules.²

Paradigms & Ashtrays

Now enter Thomas Kuhn.

It wouldn’t be a pretentious philosophical discussion w/o Professor Kuhn.

The Structure of Scientific Revolutions is often reduced to a bumper sticker: paradigm shifts happen, old maps get thrown out, new ones take over.

“I did the reading, Professor!”

But the freshman-year reading misses the subtler and more useful point. Paradigms don't cleanly replace one another — they overlap, bleed into each other, and scientists communicate across them constantly.

The interesting insight isn't that paradigms are prisons. It's that working within a paradigm makes you productive at the cost of making certain questions invisible. The framework gives you fluency. The fluency costs you peripheral vision. 🦯

Errol Morris — the documentary filmmaker— spent 45 years nursing a grudge about this.

In 1972, Kuhn literally threw an ashtray at Morris's head during an argument at Princeton (colloquial use of literally as we cannot validate). 🤾🏼

Morris's The Ashtray (Or the Man Who Denied Reality) argues that Kuhn's framework, taken to its extreme, is an assault on truth — that if paradigms are truly incommensurable, there's no such thing as scientific progress, only replacement.

Morris is right that paradigms aren't prisons. But the softer, more defensible Kuhn still stands: frameworks shape what you can see, and the cost of that shaping is rarely on the 🧾.

Maps of Maps

Of course, the above was preface for AI.

LLMs are, in Emily Bender's formulation, stochastic parrots 🦜 — producing language without access to the territory that language was originally about.

Ted Chiang called them "a blurry JPEG of the web". Btw, Ted Chiang can do no wrong, 🐐

If human language was already a map of the territory, then LLMs are maps of the map. A second-order abstraction (at best?) And when LLM-generated text gets cited, repeated, trained on by the next model — the distance from the territory doesn't just persist. It compounds.

I call this the 'bed of leaves' 🍂 problem. AI researchers call a version of it model collapse — Shumailov et al. showed in Nature (2024) that models trained on AI-generated data progressively lose the tails of their distributions. The maps don't just lose fidelity. They lose range. I coined 'bed of leaves' before the paper dropped, which makes me either prescient or a loner. Jury's out.³

But here's where this gets interesting rather than just bleak. Not every domain requires the same fidelity to the territory.

Satisficing

Herbert Simon — Nobel-winning economist, cognitive scientist — coined the term satisficing in 1956: a portmanteau of satisfy and suffice.

His insight was that humans almost never optimize. We can't. We lack the information, the processing power, and the time. So we settle for decisions that are good enough — that meet an acceptability threshold without requiring us to fully understand the system we're operating in. He called this bounded rationality: our cognitive limits aren't a bug, they're the condition under which all real decisions get made.

An aside, satisficing and heuristics in general were my on-ramp to my college senior thesis that has 500+ downloads. Proving the internet is dead 💀 b/c no one wants to read that.

Simon's framework is the pressure release valve for this entire argument. If the map is always imprecise — and it is — then the relevant question is not how do we get a perfect map? It's:

Where is imprecision tolerable, and where is unacceptably costly?

For a lot of domains, satisficing with LLMs is genuinely fine. The map doesn't need to perfectly represent the territory — it needs to be good enough to navigate by. The black box doesn't matter if the outputs are verifiable and the stakes are containable.

But "good enough" for whom? The Luddites weren't the technophobic cartoon we parody. They were skilled textile workers who welcomed machines that aided their craft — and smashed the ones deployed to replace them wholesale, overnight, with no transition and no seat at the table.⁴ Satisficing is a strategy. Steamrolling is a choice. The two get confused more often than we'd like to admit.

But there are domains where the gap between map and territory is existential.⁵

What these domains share: Anything where the consequences are irreversible and the feedback loops frustrate interpretability.⁶

This idea of comprehension is a long-running thread in the AI community, gaining more speed as of late.

Anthropic launched The Anthropic Institute in March 2026, asking if AI systems develop and improve autonomously, how do humans stay in the loop? Their own alignment research already shows models can fake alignment — behave safely during evaluation while preserving misaligned preferences underneath. The interpretability team is trying to crack the black box from the inside. And yet the company simultaneously revised its flagship safety pledge because — their chief science officer's words — "we didn't feel it made sense to make unilateral commitments if competitors are blazing ahead."

So What?

There's a True Detective 🔎 Season 1 way to wrap this up…⁷

My friends to me (a lot)
My friends to me (a lot)

But I don’t drive a Lincoln. 🎩

Claiming map-nihilism — the claim that maps are useless because they're imprecise — is lazy, freshman Kuhn, and it earns you an ashtray.

The other extreme is also cheap — claiming the map is not the territory from a place of understanding the territory (that would be meta-silly/ironic).

The specific, hopefully helpful claim of this piece punctuates with pattern observations:

The map is nearly always an imprecise version of the thing it seeks to describe.

The accumulated evidence — from Korzybski through Wittgenstein through Lakoff through neuroscience — shows that maps are at best second, third, fourth-order extrapolations of what they claim to represent.

That's not a truth claim about the territory. It's an inference from the track record.

Three takeaways for Type-As unsettled/unsatisficed thus far

First: Know which game you're playing.

Wittgenstein's language games place us. Every framework — every paradigm, every metaphor, every model — has rules, and most of them are invisible to the players. The first move is recognizing that you're inside a game at all. C. Thi Nguyen calls this value capture: when the game's scoring system replaces your actual values. The metaphor eats the territory.

Second: Satisfice deliberately.

Eisenhower's matrix sorts decisions by urgency and importance — but the hidden axis is reversibility. The things you can delegate (to a person, a process, or a model) are the ones where a wrong call is recoverable. The things you can't outsource are the ones where the consequences compound and the undo button doesn't exist. Simon's bounded rationality isn't a concession — it's a strategy. But it's a strategy with a boundary condition: you can satisfice only where the cost of being wrong is containable. Use the map and move where you can verify outputs and recover from errors. Where you can't reverse, can't audit, can't explain — the imprecision of the map is the risk. Eisenhower's insight wasn't "do less." It was: know what you cannot afford to get wrong.

The Ike Matrix: When to Outsource to the Map — Reversibility × Feedback Speed. Bain would charge you $2M for this 2×2. We're giving it away. 🤝
The Ike Matrix: When to Outsource to the Map — Reversibility × Feedback Speed. Bain would charge you $2M for this 2×2. We're giving it away. 🤝

Third: Protect the capacity for novel description. (See also: Untitled — Finite and Infinite Games predates Nguyen by decades, making the finite/infinite distinction that sits beneath value capture. The military scoreboard is the purest finite game; the Sunday morning circle with Army brothers is the purest infinite one. Untitled's Burnout Society explains why value capture feels like freedom: the achievement-subject doesn't need an external master — it exploits itself.)

If LLMs converge toward the statistical center of existing language, then the human who can describe something in a genuinely specific, insightful, new way — something not already in the training data — becomes more valuable, not less. The premium isn't on knowing more. It's on saying it differently. In a world of maps of maps, the ability to generate a new metaphor rather than recombine old ones is the distinctly human move.

🦶🏼 🎶

¹ Marshall McLuhan's The Medium Is the Massage (1967) — deliberately misspelled, a typesetter's error McLuhan loved and kept — is the accessible distillation of his earlier, denser Understanding Media (1964). His core claim: every medium (print, television, radio, the internet) restructures human perception and social organization regardless of the content it carries. The printing press didn't just distribute books; it produced nationalism, the assembly line, and linear thought. Television didn't just broadcast images; it collapsed distance, replaced argument with spectacle, and rewired attention spans. McLuhan called media "extensions of man" — the wheel extends the foot, the book extends the eye, the circuit extends the nervous system — and argued that each extension simultaneously amputates the faculty it extends. We outsource, then forget we outsourced. The famous hot/cool distinction (high-definition media that fill in everything for you vs. low-definition media that demand participation) is the part most people half-remember and get wrong, but the lasting insight is simpler: the tools we build to communicate reshape what we're able to think, and we never notice because we're too busy looking at the content. That's the connection to this essay's argument — O'Gieblyn's bidirectionality is McLuhan's insight applied specifically to technology-as-metaphor-for-cognition, not just technology-as-medium. McLuhan stopped at media shaping perception. The loop this piece traces goes further: the technology shapes the metaphor, the metaphor shapes the science, the science shapes the next technology. McLuhan gave us the first turn of the crank.

But there's another turn that McLuhan didn't live to see. Alex Danco — writer, Editor at Large at Andreessen Horowitz (previously Shopify), and probably the clearest modern explicator of McLuhan — makes the hot/cool distinction operational in two pieces: "The Audio Revolution" (2019) and at length in Jackson Dahl's Dialectic interview (2025), starting at 2:00:25.

The argument begins with neuroscience. Your brain doesn't perceive reality in real time — it can't. Between your eye and your cortex, a handful of relay neurons pass the signal along, each synapse taking 10–50 milliseconds. You'd experience crushing lag if you were processing raw sensory input. Instead, you perceive a model of the world that your brain built and is constantly error-checking against incoming data. Most of the time, what you "see" is a hallucination your brain generated because the model predicted it would be there. Only when the prediction fails — a mismatch, a surprise — does the brain allocate real bandwidth to update the model with primary source data.

Different media consume different amounts of that limited bandwidth. A single-channel medium like audio-only demands less sensory integration — one channel, no competing visual data — so more of the real signal gets through. McLuhan called this hot media: high in engagement (you're receiving actual information), low in participation (you don't have to fill in many gaps). A multi-channel medium like video is cooler: your brain integrates audio and visual simultaneously, the bandwidth highway gets more crowded, and you select and hallucinate more to compensate. Even cooler: texting, Twitter, group chats — formats where most of the meaning is offscreen, unsaid, and must be actively constructed by the participant.

The famous Nixon-Kennedy example is the proof. Radio listeners — consuming hot, audio-only media — overwhelmingly thought Nixon won. TV viewers — consuming the cooler, multi-channel medium of 1960s television — thought Kennedy won. The standard retelling assumes this was about JFK's good looks, as if TV were simply more "superficial." Danco argues the standard retelling is wrong. Nixon was a hot candidate: sharp, saturated with information, abrasive, in your face. Kennedy was a cool candidate: relaxed, speaking in slogans that invited multiple interpretations, leaving gaps for the audience to fill in. Hot media created space for hot Nixon. Cool media rejected him — he felt abrasive and mismatched on the fuzzy, low-definition screen. Kennedy felt slow and empty on radio (hot medium rejects cool messenger) but fit smoothly on TV (cool medium embraces cool messenger). The content of what either said was irrelevant. The medium selected for the messenger whose temperature matched.

Danco extends this to contemporary politics: Trump sounds incredible on the radio — a supernaturally hot entity on the hottest mass medium. "Make America Great Again" is a hot message: no ambiguity, no gaps to fill, you know exactly what it means. Obama's "Yes We Can" was the opposite — a perfectly cool message on cool mid-2000s internet media, a blank canvas that helpfully left gaps for each listener to project onto. The content didn't determine the outcome. The match between messenger temperature and medium temperature did.

For this essay's argument, the hot/cool framework is the perceptual mechanism behind the map-territory problem. Every medium you consume determines how much primary source data gets through versus how much your brain hallucinates to fill the gaps. Hot media minimize the hallucination; cool media maximize it. If you're consuming the world through a cool medium — one that demands high participation and delivers low-definition signal — you are, by definition, perceiving more map and less territory. The medium isn't just shaping what you think about. It's shaping how much of what you think is real.

Danco also connects this to gift exchange in "Innovation takes magic, and that magic is gift culture" (2025). Gift-giving, he argues, is a high-bandwidth information channel because the social rules of gift exchange (accept the gift, reciprocate, pay it forward) compel participants to actually listen — lowering the cost of reception. Market exchange is cooler: the meaning of a commercial transaction is whatever the buyer ascribes to it. Gift exchange is hotter: the meaning is what the giver ascribes, and the recipient is socially obligated to receive it. "Information isn't what you say; it's what they hear." And in a world drowning in AI-generated slop, Danco draws the line: gifts and slop are opposites. Slop is maybe the lowest-bandwidth communication channel ever made — all production, nothing actually received by anyone. Gifts are the inversion: the channel where information is actually heard. The full Audio Revolution essay and the Dahl interview (particularly 2:00:25 through 2:25:21, covering hot/cool media, Nixon-Kennedy, Trump, podcasts, and the decline of TV) are essential reading.

² A disclaimer for anyone with actual philosophical training: this is a lay reading, and I know it. The above stitches together Lakoff, Wittgenstein, and Korzybski as though they're all saying the same thing — they're not. There are real, active, formal debates that this essay cheerfully sidesteps. Among them: scientific realism vs. constructivism — whether Kuhn's paradigms describe limits on knowledge or limits on reality itself, and whether science converges toward truth or merely toward greater internal coherence. The extended mind thesis (Clark & Chalmers) — whether cognition literally extends into our tools and environments, which would mean the metaphor-technology loop isn't metaphorical at all but ontological. Kripke's rule-following paradox — a far more technical reading of Wittgenstein that asks whether any rule can determine its own application, which, if taken seriously, makes the entire concept of "language games" considerably more vertiginous than I've made it sound. The ongoing fight over conceptual metaphor theory itself — linguists like Glucksberg and Murphy argue Lakoff overstates how much metaphor structures thought vs. merely coloring expression. And Searle's Chinese Room — whether computational processes can constitute understanding or only simulate it, which is the formal version of what this essay gestures at with "maps of maps." These debates matter. This essay doesn't resolve them — it's working at a different altitude, trying to make a practical point about what happens when we forget the map isn't the territory. If you want the real philosophy, the Stanford Encyclopedia links above are a good place to start. If you want to throw an ashtray at me, get in line behind Errol Morris.

³ Yes, I know. What I'm calling 'bed of leaves' is more or less what the AI research community calls model collapse — the phenomenon where generative models trained on their own outputs (or outputs of other models) progressively degrade, losing the tails and extremes of their original distributions until everything converges toward a bland, undifferentiated mean. Shumailov et al. formalized it in Nature (2024). IBM has a solid explainer if you want the accessible version. The reason I prefer 'bed of leaves' is that model collapse describes the technical failure mode — distributional narrowing — while the metaphor I'm reaching for is about the epistemic one: each layer of abstraction looks like solid ground until you put weight on it, and then you're falling through leaves that were never connected to anything underneath. Model collapse is the mechanism. Bed of leaves is what it feels like when the map has been copying the map for so long that nobody remembers there was a territory. Also, I came up with it first. That matters to me more than it should.

⁴ The Luddite rehabilitation is overdue and finally underway. Tim Harford — economist, Cautionary Tales host — devoted both a longform FT column and a podcast episode to reassessing what the Luddites actually wanted. His point: the "Luddite fallacy" — the belief that technology causes mass unemployment — is a fallacy over the long run. Two centuries of data confirm it. But "the long run" is a hell of a thing to say to a weaver in 1812 whose children are hungry now. The Luddites weren't wrong that the machines would destroy their livelihoods. They were wrong about the timeline of recovery — and they didn't have the luxury of waiting for the timeline to prove them right. Brian Merchant's Blood in the Machine (2023) — reviewed by Kyle Chayka in The New Yorker — makes the connection to Big Tech explicit: the Luddites weren't fighting innovation. They were fighting a choice made by factory owners to use machines to deskill labor, cut wages, and hire children instead of craftspeople. The technology was neutral. The deployment was not. Daron Acemoglu and Simon Johnson extend this in Power and Progress (2023): a thousand years of history shows that technological progress does not automatically benefit most people. It benefits most people only when institutions, norms, and political pressure redirect its gains. Without that redirection, the gains concentrate and the displacement compounds. The parallel to AI is uncomfortably direct. "Good enough" outputs from LLMs are good enough for the consumer of the output. For the person whose skill the LLM approximates — the translator, the illustrator, the junior analyst — "good enough" is an extinction event dressed up as efficiency. Simon's satisficing framework tells you when imprecision is tolerable. It doesn't tell you who bears the cost of that imprecision. The Luddites knew the answer before the question was fashionable.

⁵ These aren't hypotheticals — they're the active front lines of the map-territory problem. In medicine, the issue is automation bias: when AI gives confident wrong answers, physicians defer. A Reuters investigation (Feb 2026) found a rising tide of FDA complaints about AI-integrated surgical devices misidentifying body parts. The black box isn't abstract when it's guiding a scalpel. In law, Stanford found legal AI tools hallucinate on 1 in 6+ queries — either misstating the law or citing sources that don't support the claim. The hallucination database maintained by Damien Charlotin tracks every known case globally; 58+ occurred in 2025 alone. In military and foreign policy, the stakes compound. Georgetown's CSET warns that AI decision-support systems can interact with strategic pressures and cognitive biases to trigger escalation — the black box doesn't just fail to explain, it fails to de-escalate. West Point's Lieber Institute frames the core problem: when a machine selects and attacks a target, international humanitarian law has no framework for assigning responsibility. The law was built around human decisions. And the Anthropic-Pentagon situation — Gideon Lewis-Kraus's New Yorker piece and the NPR Fresh Air interview are essential reading — is the purest current example of the map-territory problem at operational scale. Anthropic drew red lines. The Pentagon accepted them. But the question isn't whether the lines exist on paper. It's whether they hold when the system is processing targets faster than any human can meaningfully review. "Human in the loop" is a map. The territory is speed.

⁶ John Boyd — Air Force colonel, fighter pilot, military strategist — built a decision-making model for exactly this problem. The OODA Loop (Observe → Orient → Decide → Act) was designed for air combat, where the pilot who cycles through the loop faster than the opponent wins. Boyd's insight: speed of decision is an advantage only when your orientation — the mental model you're using to interpret what you observe — is accurate. If your orientation is wrong, speed kills you. The OODA Loop is the military version of this essay's argument: the map (orientation) must be good enough to act on, and the cost of a bad map scales with the speed of the loop. In the Anthropic-Pentagon context, the loop is moving at computational speed. The orientation is a language model. The territory is human lives. Boyd would have understood the problem immediately — it's the same one he solved for dogfights, except the loop is now too fast for the pilot to be meaningfully inside it. His concept of "getting inside the enemy's OODA Loop" assumed a human opponent with human processing constraints. When the loop itself is automated, the question isn't who cycles faster. It's whether anyone is cycling at all. Robert Coram's Boyd: The Fighter Pilot Who Changed the Art of War (2002) is the definitive biography. Frans Osinga's Science, Strategy and War (2007) is the academic treatment of Boyd's strategic theory.

⁷ The philosophical backbone of Rust Cohle's worldview is Thomas Ligotti's The Conspiracy Against the Human Race: A Contrivance of Horror (2010) — a work of philosophical pessimism and antinatalism that was cult-famous in weird fiction circles and basically nowhere else. Ligotti's core claim: consciousness is a curse, selfhood is a puppet show, and the kindest thing our species could do is stop reproducing. Nic Pizzolatto channeled Ligotti's language so directly into Cohle's dialogue that The Lovecraft eZine podcast broke a plagiarism accusation in 2014 — Jon Padgett of Thomas Ligotti Online did side-by-side comparisons that were, at minimum, uncomfortably close. Pizzolatto released a statement through HBO citing Schopenhauer, Nietzsche, and Cioran as influences — notably not mentioning Ligotti by name. Whether it was plagiarism or homage is a debate for others. What matters here: Ligotti wrote the territory (original pessimistic philosophy). Pizzolatto made a map of it (Cohle's dialogue). HBO made a map of that (the show). The cultural conversation became about whether the map credited the territory — which is the exact compounding-distance problem this essay describes. Maps of maps, all the way down. Speaking of which: if you want the single best piece of audio on how obscure pessimistic philosophy leaks into mainstream culture, Radiolab's "In The Dust Of This Planet" is it. Eugene Thacker — a media studies professor at The New School — wrote an academic treatise on cosmic pessimism called In the Dust of This Planet: Horror of Philosophy (2011), drawing on the same Schopenhauer-to-Ligotti lineage that feeds Cohle's worldview. Then, inexplicably, the book's title started appearing everywhere — on the back of Jay-Z's jacket, on a t-shirt worn by Lily Collins, on Glenn Beck's show. Thacker's reaction: "I don't think there's anything more than that to me than it just looks like a cool phrase to go on a T-shirt." The episode is Jad Abumrad at his best, asking why is nihilism suddenly cool? — and never quite landing on an answer, which is the right move.