Mark Veltzer's personal site

The Primacy of the Scientific Method: Why Empirical Rigor Is the Only Path to Knowledge

2026-10-05T00:00:00+00:00

Throughout history, people looking for a reliable compass to reality have reached for three instruments: the visceral impressions of intuition, the formal constructs of abstract reason, and the disciplined empirical machinery of the scientific method. History is littered with the wreckage of intuitive certainties and purely intellectual systems that were internally coherent and had nothing to do with the physical universe.

If knowledge means an accurate, demonstrable understanding of reality, there is only one legitimate path to it: the scientific method. Cut loose from empirical verification, intuition and abstract logic turn into mirages. Knowledge is neither self-evident nor self-generating. It has to be extracted from nature through systematic observation, rigorous measurement, and hypotheses that survive replication and attempts at falsification.

The Mirage of Intuition

Intuition arrives with a compelling feeling of clarity. It is immediate, emotionally persuasive, and shaped by evolution for survival, not for truth. Our ancestors needed fast, cheap pattern detection: assuming that rustling grass was a predator saved lives, even if ninety-nine times out of a hundred it was only the wind. So human cognition evolved to favor pattern completion and agency detection over accuracy.

Applied to questions about the cosmos, life, or physics, unchecked intuition produces persistent illusions:

Anthropocentrism and teleology. Intuition told early thinkers that the Earth was the motionless center of the cosmos and that living things were designed for moral or functional purposes. Both ideas collapsed once instruments revealed heliocentrism and the mechanics of evolution.
Cognitive biases. Confirmation bias, apophenia and cognitive dissonance make intuitive beliefs self-reinforcing: they filter out contrary information and amplify coincidences.
Sensory limits. Our senses register a tiny sliver of the electromagnetic spectrum, of acoustic frequencies, and of the scales of space and time. Relying on intuition is like judging the structure of an ocean by inspecting a single drop of water on a fingertip.

The feeling of certainty is a subjective emotional state, not a guarantee of anything. Without empirical correction, what feels undeniably true is often demonstrably false.

The Limits of Pure Logic

If intuition is vulnerable to emotion and bias, can pure reason do better? Rationalism has long held that fundamental truths about existence can be deduced a priori from first principles, in the image of Euclid's self-contained geometry.

Pure logic has a fatal limitation: validity is not truth. A deductive argument is a conditional engine. It guarantees only that if the premises are true, the conclusion follows. It cannot establish, on its own, that its premises about the physical world are true.

Untethered coherence. A system of thought can be impeccably logical, exquisitely detailed and mathematically pristine while being entirely fictional. Medieval scholastics debated the nature of angels with rigorous syllogisms and learned nothing at all about the cosmos.
Reality is not intuitive. Quantum superposition, wave-particle duality and the curvature of spacetime all defy classical common sense. Nobody could have deduced quantum mechanics or relativity from an armchair; anomalous measurements forced them on us.

Logic is an indispensable tool for structuring arguments, deriving predictions and spotting contradictions. But as a source of facts about the world, logic alone yields only ungrounded models. Logic without empirical input is an engine spinning without traction.

The Engine: Testing and Measurement

The breakthrough of the scientific method is that it refuses to mistake what happens inside the human head for what happens outside it. Reality is the final arbiter of every claim.

Knowledge rests on two empirical operations:

Systematic experimentation. Instead of passive observation, phenomena are put under deliberate, controlled conditions. By isolating independent variables while holding confounders constant, the investigator questions nature directly and separates real causal mechanisms from spurious correlations.
Quantification and measurement. Observations have to be translated into standard, measurable units. Measurement strips away subjectivity: it replaces vague impressions like "hotter", "faster" or "stronger" with unambiguous numbers that can be calibrated across observers, instruments and eras.

A claim that cannot be measured or compared against a baseline remains opinion. Only structured contact with the world makes thought correspond to the way things are.

Replication and Falsifiability

To keep empirical claims from being random noise, faulty apparatus or investigator bias, the scientific method rests on two criteria.

Independent replication

A single finding, however striking, is not knowledge. It may be a statistical fluke, an experimental artifact, or the product of a researcher's unconscious thumb on the scale. A result becomes knowledge only when independent laboratories, using the same methods in different settings, get the same measurable outcome. Replication removes the identity and authority of the observer from the picture and makes truth accessible to anyone who checks.

Falsifiability

As Karl Popper argued, what makes a statement scientific is not that it can be confirmed but that it can be refuted. A hypothesis that fits every conceivable outcome explains nothing.

A real claim makes precise, risky predictions: "If theory X is correct, Y must happen under test Z; if W happens instead, X is false."
Claims insulated against any contrary evidence, whether metaphysical assertions, astrological forecasts or conspiracy theories, lie outside knowledge altogether, because no fact could ever challenge them.

Knowledge advances not by piling up dogmas but by the continuous, ruthless elimination of error.

The Only Self-Correcting Method

Every alternative to the empirical method, whether dogmatic tradition, mystical intuition or speculative rationalism, lacks a built-in mechanism for correcting itself against reality. Once entrenched, such systems persist indefinitely, because nothing could ever decisively refute them.

The scientific method is unique in that it embodies epistemic humility. It acknowledges the weaknesses of the human mind, rejects authority, and ties belief to facts that are demonstrable, reproducible and falsifiable. Logic supplies the grammar, and intuition may occasionally supply the initial curiosity, but only the scientific method delivers truth. For anyone who wants to understand the universe, it is not merely the best path to knowledge. It is the only one.

Fighting the Wrong Battle: How the Academic Left Chose the Wrong War

2026-10-04T00:00:00+00:00

For about three hundred years, being on the intellectual left meant something fairly specific. It meant reason against authority, evidence against revelation, and a willingness to take apart whatever was held sacred. The philosophes went after the divine right of kings and the power of the clergy. The secular humanists and the radical scientists of the nineteenth and twentieth centuries went after superstition wherever they found it. Progress was measured by how much delusion had been cleared away and how much room that left for people to live freely.

That is not what the academic left looks like today. It has not run out of energy or money. It has picked the wrong enemy. The analytic machinery that used to be pointed at religious dogma, still the most resilient and most demonstrably false set of ideas shaping human lives, has been turned inward, onto the university itself and onto a story of ubiquitous oppression that it mostly constructed. The place that was supposed to be the stronghold of secular reason has become an echo chamber of moral grievance.

The Pass That Religion Received

The first mistake was giving up on criticising everything equally. The old rationalist tradition did not grant immunity to a belief because it claimed divine origin or long tradition. Revelation, superstition and moral absolutism were understood to be the main obstacles to civil liberty, to science and to personal autonomy, and they were treated accordingly.

In the contemporary academy, a combination of postmodern epistemology and cultural relativism has fenced off exactly those beliefs. Under the headings of anti-colonialism, cultural sensitivity and the defence of identity, religious doctrines, many of which endorse the subordination of women, fatalism, hostility to gay people and contempt for other sects, are treated not as factual claims that can be wrong but as identity markers that must be respected. I have written before about how much more readily academics blame anything other than religion, and about how cultural relativism shields religion far more than it shields politics. This post is about what that reluctance costs.

What it costs is a betrayal. An intellectual who will not criticise a reactionary doctrine for fear of giving offence has quietly abandoned the people living under it: the freethinkers, the apostates, the women and the dissidents in theocratic states and theocratic communities. Medieval metaphysics gets a free pass, and in exchange the academic gets to look cosmopolitan and tolerant.

From Truth to Moral Standing

Underneath this retreat is a change in how claims are judged. The secular project assumed that truth is objective and can be found by observation, logic and the attempt to falsify. Religious claims, about the origin of species, the structure of the cosmos or the source of morality, were answered with evidence and argument.

In much of the humanities and social sciences that has been replaced by a single explanatory scheme: society is to be understood as overlapping grids of oppressors and oppressed. Within that scheme a claim is judged less by whether it matches reality and more by who is making it and where they sit on the grid. That is a category error. The identity of a speaker tells you nothing about whether the speaker is right.

Once moral standing replaces truth-seeking, the enterprise drifts into triviality. Enormous effort goes into policing vocabulary, cataloguing small offences and deconstructing hierarchies inside some of the most comfortable institutions on earth. Meanwhile the obvious, physical enemies of human progress get little more than a shrug: theocracies that execute apostates and dissidents, religious organisations that campaign to restrict reproductive freedom, sectarian violence fed by ancient myths, and the long creationist campaign against teaching biology in schools.

The Real Adversary

None of this would matter much if religious dogma were a quaint survival from a pre-scientific age. It is not. It remains one of the best-organised and most politically influential forces in the world. It shapes legal systems, dictates school curricula, slows biomedical research, restricts what people may do with their own bodies and supplies the justification for wars. I have argued that religion is best understood as an ideology, and judged as an ideology it is among the most powerful ones in operation.

When the universities trade the fight against institutional superstition for an endless internal culture war, they leave that ground to the theologians. Many people want a coherent, evidence-based alternative to dogmatic certainty. What they find in the academy instead is opaque jargon and ideological purity tests. The secular left has lost much of its ability to persuade the general public, and the space it vacated has been filled by populist movements that mobilise religious sentiment with nobody of comparable standing pushing back.

Why the Echo Chamber Stagnates

A framework that knows its conclusion in advance cannot discover anything. If every interaction and every social outcome is assumed to be an expression of power and oppression, research becomes a search for confirming instances, and it always finds them. Inquiry turns into ritual, and scholarship into confirmation bias.

The insularity also hides from the academy its own loss of cultural authority. A seminar room is not the front line of human emancipation, and inventing new terminology is not the same as changing society. Renaming things does not make a dogma go away, and enforcing etiquette on campus does nothing to the political reach of organised fundamentalism.

Back to the Real War

If the intellectual left wants its vitality and its moral standing back, it has to go back to where it came from. Human liberation was always, at bottom, a fight for the human mind against falsehood, fatalism and unearned authority.

That takes courage of a fairly old-fashioned kind: the courage to say that reason, evidence and universal human rights outrank religious sensibilities; the courage to criticise an irrational doctrine regardless of how much cultural prestige its believers carry; and the sense to see that the greatest threat facing humanity is not insufficient ideological conformity inside the university but the continuing grip of ancient superstition on human societies. The skirmishes of moral vanity can be dropped. The war that matters is the old one, reason against dogma, and it is far from won.

Where This Started

Some time ago I watched a YouTube series of a history course taught by Yuval Noah Harari. At one point a student asked him, "Where did God come into this?" Harari's answer was: "Give me evidence of how God changed things and I will consider it, but until you do, I cannot relate to your question." I remember hearing that and thinking: what cowardice! The answer should have been, "There is no God. Next question, please." That moment is what started me thinking about the cowardice of academia on this subject.

Beauty, Truth and the Size of Working Memory

2026-10-02T00:00:00+00:00

Years ago I argued that beauty is not truth but an artifact of our brain size, using the four-colour theorem and a symmetry argument. I want to return to the subject from a different direction and push the claim further. Beauty is not merely uncorrelated with truth. It has a specific cause, and the cause is mechanical: beauty is the sensation of an object fitting into human short-term memory. Whatever fits, we call elegant, profound or beautiful. Whatever does not, we call ugly, tedious or incomprehensible, and we quietly downgrade our estimate of its truth at the same time. Once you see the mechanism you see it in mathematics, in the way we explain history, and in music, and it is the same mechanism in all three.

The Budget

Short-term memory is small. The classic figure is seven items give or take two, and more recent work puts the number of genuinely independent chunks closer to four. Whatever the exact number, the budget is tiny, it is fixed by biology, and it has not grown since long before there were mathematicians. Everything we are able to understand at once has to pass through this window. An idea that fits in it can be turned over, inspected from every side and compared against other ideas. An idea that does not fit has to be processed in pieces, each piece parked in long-term memory while the next is examined, and the result is never experienced as a single thing.

My claim is that the feeling we call beauty is the feeling of an idea fitting. It is a signal about the match between the object and the window, and it carries no information about the object beyond that match.

Proofs

Mathematicians praise short proofs, call them beautiful, and often add that beauty is a guide to truth. Long proofs, and above all the computer-assisted and brute-force ones, are tolerated as valid but resented as ugly, and more than one mathematician has said out loud that a theorem with only an ugly proof is somehow not yet properly understood.

But the length of the proof says nothing about the theorem. A two-line proof and a two-thousand-page proof establish exactly the same thing, and the theorem is exactly as true at the end of either. The difference between them lies entirely on our side: one fits in the window, the other has to be swallowed in pieces. Calling the short one beautiful is a report about the reader's working memory, and the moment we mistake it for a report about mathematics we have confused our comfort with the truth.

Artificial intelligence makes this vivid. A system that holds a million tokens in its context has no built-in reason to prefer a short proof over a long one. Both are equally valid, both check out, and neither is harder for it to hold. If such a system developed aesthetic preferences they would be calibrated to its own window rather than ours, and there is no reason to expect the two to coincide. Our fondness for the one-line proof would look to it exactly as parochial as a child's preference for stories with three characters looks to us.

Causes

Move from mathematics to history and the same budget distorts the picture in a far more damaging way. Why did a war break out, why did an empire fall, why did a market crash, why did a country slide into civil conflict? The honest answer to such questions is nearly always a list: twenty or thirty factors, each contributing a few percent, interacting with one another, none sufficient on its own. I have written about how little the social sciences manage to explain, and part of the reason is that the true explanation of a large social event has precisely the shape our memory cannot hold.

So we do not hold it. We replace the list with a story: one villain, one cause, one turning point. The story fits in the window, it can be told over dinner, and it feels true in exactly the way the short proof feels true. It is beautiful. It is also wrong, not in its details but in its shape, because the real cause was a weighted sum, and a weighted sum of twenty terms cannot be made beautiful without discarding most of the terms. I argued long ago that truth is not short and that this is why falsehood spreads; the working-memory budget is the mechanism underneath that observation. The aesthetic preference for the neat explanation is not a harmless taste. It actively prevents us from understanding the problems we most need to understand, because every accurate account of them is, by our standards, ugly.

Music

Music is the case that convinced me this is one phenomenon rather than three. What is beautiful in music is the motif: a handful of notes, a short rhythm, a progression of four or five chords. These are exactly the size of a working-memory chunk. A listener can hold them, recognise them when they return and anticipate them a moment before they arrive, and the pleasure of music is very largely the pleasure of that recognition and that anticipation.

Large works are built on this fact. A symphony runs forty minutes, but nobody holds forty minutes in their head. It is held as a few motifs that are stated, repeated, transformed, inverted and brought back, and every structural device in the composer's toolbox, the recapitulation, the variation, the fugue, is a way of letting a tiny memory experience a huge object as coherent. Imagine the alternative: an eighteen-minute melody that never repeats and never returns to anything you have heard before. It could be composed. Every note could be well chosen. It would not be heard as beautiful, or even as a single piece, because by the second minute the first is gone and there is nothing left for the listener to recognise. What cannot be stored and recalled is not perceived as beautiful. Memory is the arbiter of taste.

Conclusion

Put the three together and the conclusion is hard to avoid. Beauty is relative to the scale and duration of short-term memory. It is not a property of proofs, of explanations or of melodies. It is a relation between those objects and a very small buffer in the head of a particular primate. When we say that truth is beautiful we are projecting the shape of that buffer onto the world and then congratulating the world for fitting it. Reality is under no obligation to be simple. We value compression not because the universe is compressed but because our processing power demands it, and that demand was set by a budget that evolved for tracking a few predators and a few allies, not for proving theorems or explaining wars.

This is not a traditional philosophical position. The classical theories of beauty spoke of symmetry, harmony, proportion and the perception of order, and none of them, as far as I know, tied aesthetic value to the capacity of memory. The idea belongs to cognitive science, evolutionary psychology and information theory rather than to Plato or Kant. I count that in its favour. The older theories described what beautiful things have in common. This one says why we find them so, and the answer is less flattering than the tradition would like: not that we are attuned to the deep order of the cosmos, but that we have a small window and we like what fits through it.

An Ode to Individuality

2026-10-01T00:00:00+00:00

Most of this blog is an argument for groups. I argue for direct democracy, for decisions made by everyone instead of by a few, and against the idea that any single person should be trusted with power. I am not going to retract any of that. But there is a fact I want to set beside it, because it is true and because I think it is under-appreciated by people who, like me, think in terms of systems and populations. The greatest things human beings have ever done were done by individuals. So were the worst.

The Record

Look at the arts first, because the pattern is cleanest there.

Jazz was the work of a handful of people who each, personally, changed what the music was. Louis Armstrong turned it into a soloist's art. Charlie Parker rebuilt its harmony. Miles Davis reinvented it at least three times and John Coltrane took it somewhere nobody has quite followed. There is no committee anywhere in that story. Each of those men stood in front of a band and played something that did not exist until he played it.

Painting is the same. Giotto, Caravaggio, Turner, Van Gogh, Picasso: each one is a break in the line, and the break has a single name on it. The schools and movements that art history likes to talk about are mostly the people who came afterwards and copied.

Literature does not even pretend otherwise. A novel is one mind, sustained over years, and the greatest ones, Dostoevsky, Tolstoy, Proust, Kafka, Joyce, are unmistakably one mind each. No one has ever written a great novel by committee and no one ever will.

Musical composition is the purest case. Bach, Mozart, Beethoven. Beethoven wrote the Ninth Symphony deaf. The whole structure, every voice, existed inside one skull before anyone else heard a note.

Science is supposed to be the collective enterprise, and in its day-to-day operation it is. But the leaps are not. Newton worked out mechanics and the calculus largely alone, in a house in Lincolnshire during a plague year. Darwin thought for twenty years in his study. Einstein produced special relativity in a patent office with no collaborators and no laboratory, and general relativity ten years later in nearly the same solitude. Gödel, Turing, Ramanujan. The institutions of science are superb at checking, extending and applying. The founding insight, over and over, came from one person who saw something that nobody around them could see.

And then the disasters. The twentieth century's great catastrophes were not the product of mobs or committees either. They had names: Hitler, Stalin, Mao, Pol Pot. Each of them bent an entire nation around the contents of one head. Remove the individual and the catastrophe does not happen in the same form, perhaps not at all. The variance of human affairs, in both directions, is driven by individuals.

Why It Is Individuals

I do not think this is a mystery. A really new thing has to be held whole, in one place, while it is being built, and the only place that can hold an unfinished idea whole is a single mind. A group can hold a finished idea, can test it, improve it and propagate it. A group cannot hold a half-formed one, because a half-formed idea cannot yet be communicated; the moment it can be communicated it is no longer half-formed. The lonely stretch, the part where the thing exists only as a hunch and the world would, if asked, vote it down, is necessarily walked alone.

This is also why the disasters are individual. A group averages. Put a hundred people in a room and the policy that emerges will be somewhere near the middle of what the hundred believe, which is boring and usually survivable. The extremes, the genuinely new good and the genuinely new evil, are off in the tails, and a single person can live in the tails in a way a group never can.

I have argued that art teaches the wrong way to think about mass phenomena, because it sees everything through one protagonist. I still think so. But that is a point about explanation, about how to understand a famine or an epidemic. It is not a point about creation. When the question is where the new thing came from, the protagonist is usually the right answer.

The Amendment

Here is where I want to complicate the ode, because the record has a second pattern in it that is almost as strong as the first.

The very greatest things frequently came not from one individual but from two or three who could talk to each other. Lennon and McCartney wrote better songs together than either wrote apart, and both wrote for decades apart, so the comparison is not hypothetical. Watson and Crick. The Wright brothers. Kahneman and Tversky, who said that the thinking happened in the conversation and neither could reproduce it alone. Jobs and Wozniak. Rodgers and Hammerstein. Gilbert and Sullivan, who could not stand each other and were useless separately.

This is not a return to the committee. Two is not a group; two is the smallest arrangement in which an idea can leave one head and survive the trip. What these pairs had was not a process or a structure but an extraordinary quality of communication: each could hand the other a half-formed thing and get it back improved, without it being averaged into mush along the way. That is rare. It is rare enough that it is nearly as remarkable as individual genius, and it may be the one way that individual genius gets to be more than it is.

So the amendment to the ode is this. If we want more of the great things, the answer is not to dissolve individuals into teams, which produces the reliable mediocrity of everything designed by a large organisation. It is to find the ways in which two or three individuals can communicate well enough to be, briefly, one mind with more than one person's capacity. Nobody knows how to manufacture that. But we could at least stop designing institutions that prevent it.

Where This Leaves the Groups

None of this is in tension with the rest of this blog, though it looks like it should be. Groups and individuals are good at different things, and the error is in letting either do the other's job.

Individuals create. They should be left alone to do it, given the room and the solitude it needs, and protected from the committee that would, if consulted at the right moment, vote every great thing down. But individuals should not decide for everyone else what to do with what has been created, because the same quality that puts one person in the right tail puts another in the wrong one, and from the outside you cannot tell which is which until it is far too late. That is the job of the group, and it is the job direct democracy is for: not to make the new things, but to choose, by everyone, which of the new things we are going to live with.

Let the individual compose the symphony. Let all of us decide whether to build the concert hall.

Emotion: The Enemy of Reason

2026-10-01T00:00:00+00:00

Emotion has a good press. It is held up as the core of our humanity, the source of art, empathy and meaning, and anyone who speaks against it is assumed to be arguing for a cold and inhuman life. I am not. I am making a narrower claim: when the job is to find out what is true, to solve a hard problem, or to make a sound decision, emotion is the chief adversary of reason. Rationality demands detachment, empirical scrutiny and consistent logic. Emotion runs on instinct, immediate gratification and subjective impulse. The two are not partners. They are competitors for control of the same machine, and the older one usually wins.

To see why reason is derailed so often, it helps to look at where emotion came from, what it does to our cognitive machinery, how it erodes objectivity, and how it anchors the beliefs that are hardest to shift.

The Ancient Hijack

The conflict between reason and emotion is a structural feature of how the brain was built. The emotional machinery, centred in the limbic system and in particular the amygdala and hippocampus, is hundreds of millions of years old and was laid down in lineages far older than ours. Its single evolutionary job was immediate survival: spot the predator, flee the danger, compete for food, find a mate. Natural selection rewarded speed, not accuracy and not nuance. An ancestor who paused to weigh the evidence that the rustle in the grass was a snake did not become an ancestor.

The prefrontal cortex, which handles executive function, abstract thought, probability and logical analysis, is a recent arrival by comparison. And because the brain was built in layers, with new structures added on top of old ones rather than replacing them, the wiring is lopsided. The pathways running from the emotional centres to the prefrontal cortex are more numerous and stronger than the pathways running back. When an emotional trigger fires, the neurochemical surge floods the brain long before the slower analytical faculties have had a chance to evaluate the situation. We are Paleolithic brains facing modern problems, and the ancient survival reflexes routinely hijack the deliberate reflection that those problems need. I have argued before that rationality is the exception rather than the rule; this is the anatomy behind that claim.

The Tyranny of the Immediate

When emotion takes the helm it does not merely add noise. It warps processing in systematic directions, and those directions have names.

The first is present bias, the preference for a smaller reward now over a larger one later. Emotional impulses crave immediate gratification and are nearly blind to delayed consequences, however large. Whether the domain is money, health or a destructive habit, the urgent "now" beats the prudent "later" every time the feeling is allowed to decide.

The second is motivated reasoning, with confirmation bias as its workhorse. The rational picture of thought is that you weigh the evidence and arrive at a conclusion. What the emotional brain actually does is choose the conclusion first, on grounds of fear, pride, tribal belonging or desire, and then hire reason as a defence attorney to assemble the justifications after the fact. The arguments sound like reasoning and have the form of reasoning. They were never the cause of the belief.

The third is the affect heuristic, which replaces risk assessment with a gut reading. If something feels unpleasant it is judged dangerous or wrong; if it feels soothing it is judged true and beneficial. The feeling arrives in milliseconds, the analysis would take minutes, and the feeling is the one that gets to vote.

The Fog of Affect

Clear analytical thinking requires objectivity, the discipline of separating what is true from what we wish were true. Emotion is subjective by construction. It reports on an internal chemical state, not on the external world, and as emotional arousal rises, analytical acuity falls. The two move in opposite directions, and nobody is exempt.

In science, in law, in economics and in government, emotional involvement creates blind spots in exactly the places where clear sight matters most. The characteristic error is emotional reasoning: the unspoken conviction that I feel this strongly, therefore it must be so. I have written before about the related mistake of treating the strength of a feeling as a measure of a topic's depth. Under its influence, disconfirming evidence stops being useful corrective data and becomes a threat to personal identity or to the cohesion of the group. Where a sceptic would ask for a test, the emotional reasoner feels an attack, and defends. Visceral reaction substitutes for verifiable evidence, and dogmatic certainty grows exactly where doubt was warranted.

The Psychology of Faith

Nowhere is the subjugation of reason to emotion plainer than in religious and supernatural belief. Centuries of scientific progress have supplied naturalistic explanations for the things religion once explained, and the beliefs have not budged, because they were never held on evidential grounds. They are held because they answer emotional needs, and they answer those needs very well.

Religion offers sanctuary against the deepest existential anxieties:

The terror of mortality. The promise of an afterlife and of continued consciousness soothes the visceral fear of death and of simply ceasing to be.
Loneliness and chaos. An omnipotent and benevolent deity is a parent who never leaves. The universe becomes governed by purposeful order rather than by indifferent physics, and that is a far more comfortable place to live.
Grief and justice. A supernatural framework promises reunion with the dead and a final settling of accounts that the real world withholds.

Because these doctrines work as coping mechanisms, they are defended with the intensity reserved for coping mechanisms. Reason asks for verifiable proof, replicability and internal coherence. Religious doctrine asks for faith, which is an emotional commitment to accept claims without evidence, and often against it. That is why I have called faith an anti-epistemology: it is emotional vulnerability, institutionalised, shielding the supernatural claim from the disinfectant of rational critique.

The Hard Discipline of Reason

Emotion is an inescapable part of our evolutionary inheritance. It supplies motivation and it supplies connection to other people, and no sane person would want to be rid of it. But granting it authority over questions of truth and judgement is a recipe for error, delusion and regression.

Reason does not come naturally to our species. It is an acquired, hard-won discipline, and it has to be maintained with continuous vigilance, self-awareness and institutional guardrails, because the moment the vigilance lapses the older machinery takes over again. Calling emotion the enemy of reason is not a case for cold inhumanity. It is a defence of intellectual integrity. Only by reining in the emotional impulse, through empirical inquiry and logical consistency, can we hope to see reality as it is rather than as we would like it to be.

IQ and Religious Belief: What the Correlation Actually Says

2026-10-01T00:00:00+00:00

Every few years someone reports that intelligent people are less religious, and every few years the report is received in one of two ways: as a trophy by the atheists and as an insult by the believers. Both reactions skip the part that matters, which is what the numbers actually are and what they can and cannot support. I want to write that part down, because the finding is real, it is modest, and almost everything said about it in public is an exaggeration in one direction or the other.

The Bottom Line

There is an inverse correlation between intelligence and religious belief. More intelligent people are, on average, less likely to believe. That much is settled; it has been measured in dozens of studies over several decades and it comes out negative essentially every time.

What is not settled is how big it is, where it comes from, and what, if anything, follows from it.

The Numbers

Level	Correlation (r)	Variance explained	Source
Individuals	about −0.15 to −0.25	about 2–6%	Meta-analyses of 63–83 studies (Zuckerman, Silberman & Hall 2013; Zuckerman et al. 2019; Dutton et al. 2020)
Countries (national averages)	about −0.6	about 36%	Lynn, Harvey & Nyborg 2009

In plain terms: religious believers score roughly 2 to 6 IQ points lower than non-believers, with enormous overlap between the two groups. If you line up a random believer against a random non-believer, the believer will have the higher IQ a great deal of the time. This is not a finding about individuals. It is a finding about distributions that are slightly shifted relative to each other.

How to Read the Numbers

Effect size benchmarks. The usual Cohen thresholds call 0.1 a small correlation, 0.3 a medium one and 0.5 a large one. By that scale the individual-level figure is small. But Cohen's thresholds were a rule of thumb, not a measurement, and when Gignac and Szodorai (2016) actually looked at the distribution of effect sizes in individual-differences research they found that 0.1 is small, 0.2 is typical and 0.3 is relatively large. By that more honest standard, a correlation of −0.25 is an above-average effect for psychology. It is real and it is not negligible. It is also nowhere near a law.

Individual versus country figures. The −0.6 is the number that gets quoted, because it is impressive, and it is also the number that should be quoted least. It compares country averages, not people. Averaging strips out the individual variation, and the country averages share a long list of common causes: wealth, education, health, political stability. Each of those raises measured IQ and each of them lowers religiosity, so the two averages move together whether or not intelligence has anything to do with belief in any particular person. When you control for development the country-level link weakens substantially. Taking a correlation between national averages and applying it to individuals is the ecological fallacy in its textbook form. And that is before you get to the national IQ datasets themselves, which have serious and well-documented quality problems.

So the honest headline is the individual figure, and the individual figure is modest.

Where the Effect Is Strongest

The correlation is not uniform. It is stronger for religious belief than for religious practice: intelligence predicts whether you believe the doctrine better than it predicts whether you turn up on Sunday. It is clearest in Western samples and weaker or absent in some other cultures. And a large share of the studies behind it were done on Western college students, which is a narrow slice of humanity to generalise from.

Proposed Explanations

None of these is established. They are the candidates.

Nonconformity. More intelligent people may be more willing to question the prevailing norms of their society. If this is the mechanism, the effect should be weakest where secularism is already the norm, since there the nonconformist move is toward religion. This is testable and the cross-cultural pattern is at least consistent with it.

Cognitive style. Religious belief has been linked to intuitive rather than analytic thinking, and intelligence to analytic thinking, so the correlation would be a by-product. The experimental work that was supposed to demonstrate this, the priming studies of Gervais and Norenzayan (2012) in which nudging people into an analytic frame of mind reduced reported belief, failed to replicate. The idea may still be right, but the evidence that was offered for it is gone.

Functional substitution. Religion offers a sense of control, self-worth and security. So do the things that come with intelligence: education, income, self-regulation. If both meet the same needs, people who have more of the second may simply need less of the first.

Caveats

The data are correlational and the direction of causation is unknown. Upbringing, education and social class could easily drive both intelligence scores and religiosity, with no direct link between them at all. I have written before about how reliably religion is inherited from one's parents, and that same inheritance carries a great deal else with it.

"Religiosity" is not one thing. Studies measure it as belief in God, as attendance, as self-rated importance of religion, as fundamentalism, and the correlation depends on which you pick.

And a correlation of this size, as I have argued elsewhere, licenses no inference about any individual. It tells you nothing about the intelligence of the believer in front of you, and it is not an argument about whether any religion is true. Smart people have believed absurd things throughout history; rationality is a separate ability from intelligence and IQ tests do not measure it.

What the finding does do is rule out one claim that believers sometimes make, namely that faith and intelligence go together, that the more one understands the more one believes. The data say the opposite, weakly but consistently. That is all they say, and it is enough.

Religions Did Not Evolve, They Merely Survived

2026-10-01T00:00:00+00:00

Daniel Dennett asked why religions have such a hold on us, and why the Enlightenment's confident prediction that they would fade away turned out wrong. His answer was that they have been brilliantly designed, over thousands of years, to do exactly that: to be robust, self-maintaining organisations built to keep us enthralled, and that this design came out of an evolutionary process. I have used Dennett's ideas before and I think he is one of the few philosophers worth reading on this subject. On this particular claim I think he is wrong, and the mistake is instructive.

My claim is simple. No evolutionary process of the kind Dennett describes has occurred. Religions have changed, certainly. But they changed as little as they could get away with, exactly as much as was needed to avoid extinction and not one step more. That is not evolution. Evolution needs variation, lots of it, and selection acting on that variation over enormous numbers of trials. Religions suppress variation as a matter of doctrine, and the number of trials is pitiful. If religions really had been through the kind of trial and error that produces an eye or an immune system, they would not be as clumsy as they are.

What Evolution Actually Requires

It is worth being precise about what the word means, because "evolution" has become a loose synonym for "changed over time" and that is not what Dennett is claiming. He is claiming design: that the features which hold people are there because a selective process refined them. Design by selection has requirements.

It requires a large population of variants. Natural selection shapes the eye because there are billions of eyes, every one of them slightly different, in every generation. The lineage leading to a modern eye passed through something like a million generations, each one a round of trial and error over a population too large to count.

It requires that variation be cheap and continuous. Mutations happen whether anyone wants them or not. Nothing in a bacterium polices its own genome for heresy.

It requires that selection be on the trait in question. If you want a lineage optimised for holding people enthralled, the variants that hold people better must out-reproduce the ones that hold people worse, over and over, for a very long time.

Now count the religions. The total number that have ever existed is somewhere in the thousands, generously, and most of those were tribal religions that never competed with anything and simply died with their tribe. The ones that have been in serious competition with each other, the world religions and their major branches, number a few dozen. Their generation time, the interval between one meaningful variant and the next, is measured in centuries. A few dozen lineages, a few dozen generations. That is not a population under selection. It is an anecdote.

Religions Suppress Variation on Purpose

The second requirement fails even more badly than the first. Religions are the only cultural institutions I know of whose central organising principle is the prohibition of change. The text is fixed. The ritual is fixed. The founder's words are final. Innovation has a name in every religion and the name is heresy, and for most of history the penalty for it was death.

This is the opposite of a system exploring its design space. A religion is a system engineered, deliberately and openly, to stay still. When it does move, it moves because an external force left it no choice: a conquest, a scientific discovery that could no longer be denied, a political order that would not tolerate the old doctrine. And when it moves it moves the minimum distance. I have written about how Yahweh changed from a desert war god into the creator of the universe, and every step of that journey was taken under duress, as late as possible, and no further than necessary. Monotheism arrived when the political situation made the other gods an embarrassment, not because a thousand variant theologies were tried and the monotheist one held its congregations best.

Variation in religion is not generated and tested. It is generated under pressure, mostly in the form of schism, and the schism is then fought as a war. The winner is the side with the bigger army, which is not the same thing as the side with the better design for holding minds. Christianity did not spread through Europe because its memes out-competed the pagan ones in a fair contest for human attention. It spread because Constantine adopted it and because kings who converted then converted their subjects at sword point. Islam spread the same way. Selection by conquest selects for the military and political fortunes of the host, not for the features of the religion.

Look at the Product

The strongest argument, though, is to look at what the supposed design process produced. Dennett's claim implies a product that is refined, that fits its niche with the elegance of a well-adapted organism. Religions are not that. Religions are a mess.

They carry enormous amounts of material that does nothing for retention and a good deal that works against it: dietary laws that make social life with outsiders awkward, prohibitions on medicine, celibate orders that remove their most committed members from reproduction, doctrines that are embarrassing to defend and that drive educated members out. The Shakers designed themselves out of existence. Every religion has a bookshelf of rules that it quietly ignores because enforcing them would empty the pews, and it keeps them on the books anyway because it cannot change the book. An evolved system sheds dead weight. Religion cannot, because shedding is heresy.

Then look at the features that do hold people, the ones Dennett has in mind. Fear of hell. The cost of leaving. Indoctrination of children. Marking members and shunning outsiders. These are real and they work, and I have described the psychological machinery behind them. But notice that they are the same handful of features in every religion, that they were present essentially in their final form in the earliest versions we have records of, and that they are the obvious tricks any human institution would hit on in its first century. A guild, an army, a political party and a street gang use the same ones. They are not refinements. They are the starting kit. Nothing in three thousand years improved on them, which is exactly what you would expect from a system that forbids improvement.

Survival Is Not Design

I have argued elsewhere that religions behave like memes, that they split and compete and propagate by mechanisms that have nothing to do with truth, and I stand by that. But competition and survival are weaker things than design. A rock survives for a billion years without having been designed for it. What survives in a world with little variation and few trials is not the best-adapted variant. It is whatever was adequate and happened to be standing when the music stopped.

That is what religions are. Not brilliantly designed organisms honed to grip the human mind, but adequate early institutions that found a few obvious hooks, nailed themselves to the floor, and have been grimly refusing to move ever since. Their hold on us comes from the hooks, from the fact that human beings were not built to think, and from inertia. The Enlightenment was wrong about the timing, not about the direction: a system that cannot adapt does not go extinct quickly, but it does not get better either, and the gap between it and the world it has to live in only grows.

Dennett credits religion with more cleverness than it has. The honest description is less flattering and, I think, more hopeful. Religions are not sophisticated adversaries. They are old, stiff and stupid, and they are still here because nothing has yet pushed hard enough to knock them over.

Theology Belongs with Astrology, Outside the University

2026-10-01T00:00:00+00:00

Imagine a modern university announcing a Department of Astrology. Undergraduate degrees, a master's programme, doctorates in the interpretation of natal charts. The reaction would be instant and unanimous. The faculty senate would protest, the scientists would write scathing letters, the donors would withdraw. Nobody would need to spell out the justification, because it is obvious: astrology is not a field of knowledge. It has no empirical footing, no falsifiable claims, and no logical machinery capable of producing a valid conclusion.

Yet the same universities, all over the world, maintain well-funded faculties of theology, with endowed chairs and degree programmes, and nobody protests at all.

I want to argue that this is not an intellectual distinction. It is a historical accident protected by tradition. Epistemologically, theology stands on exactly the same ground as astrology, and neither of them belongs in an institution whose purpose is the pursuit of knowledge.

Premises, Logic, Conclusions

Academic work rests on a simple compact. An argument begins from premises that are coherent and can be shown to hold, proceeds by sound logic, and ends in conclusions that can be evaluated, tested or contested against the world. Theology imitates this process. It borrows the vocabulary of scholarship and none of its structure, and it fails at all three stages.

The premises cannot be examined. In genuine scholarship a premise is empirically verifiable, mathematically demonstrable, or an axiom that everyone can see. Theology begins by assuming its conclusion: that supernatural beings exist, that a revelation occurred, that a text carries divine authority. A premise that cannot be checked, verified, or even clearly defined is not a premise. It is an assertion.

The logic runs in a vacuum. Theology is often praised for the intricacy of its internal debates, its scholastic distinctions and elaborate taxonomies. But complexity is not logic. An astrologer can compute, with real mathematical precision, the angular relationships between the planets and the arbitrary houses of the zodiac. Internal consistency inside an invented system is decoration. If the axioms do not connect to reality, every step that follows from them is meaningless, however rigorous the steps look.

The conclusions cannot fail. A real conclusion tells us something about the world that could, in principle, have turned out otherwise. In theology every outcome is fixed in advance by dogma. When a contradiction appears, the hypothesis is not revised; instead there is an appeal to mystery, to the ineffability of the divine, or to some interpretive acrobatics. A system in which no observation could ever prove a claim wrong is a system that cannot reach a true conclusion, because it has given up the only mechanism by which a conclusion gets to be true. I have written before about why theology has never discovered anything; this is the reason.

The Parallel, Attribute by Attribute

The university's refusal to host astrology shows up the double standard it grants theology. Put the two side by side:

Attribute	Astrology	Theology
Foundational claim	Celestial bodies determine human destiny and character	Supernatural beings dictate cosmic and moral reality
Empirical basis	None; refuted by astronomical and psychological testing	None; immune to verification or testing by design
Method	Elaboration of ancient texts, hermeneutics, symbolic mapping	Elaboration of ancient texts, hermeneutics, dogmatic dialectic
Internal machinery	Elaborate rules, charts, aspects and classifications	Elaborate doctrines, apologetics and conceptual models
Standing in the university	Rejected everywhere as pseudoscience	Degrees, chairs, prestige and departmental funding

If complexity, longevity and a textual tradition were enough to earn academic legitimacy, astrology would deserve a corridor next to astronomy. It has all three. The university is right to keep it out, and in keeping it out while welcoming theology it reveals that the standard being applied is not an epistemic one.

Belief Is Not Knowledge

The root error is the conflation of belief with knowledge. Epistemology settled long ago that belief without justification and truth is not knowledge. Personal conviction, cultural inheritance and existential comfort are psychological facts about the believer. They are not intellectual achievements, and they say nothing about the world.

A university exists to preserve, discover and transmit knowledge. It is the place for disciplines that submit their hypotheses to critique, to empirical test and to the possibility of refutation. When it awards a credential, and above all a graduate credential, it is certifying that the holder has mastered an inquiry capable of extending human understanding. Awarding that credential in a field with no checkable premises, no logic that touches the world and no testable conclusions cheapens every other degree the institution grants. The doctorate in theology and the doctorate in physics hang on the same wall and carry the same title, and only one of them was earned by finding something out.

Religion Can Be Studied, Theology Cannot Be Done

None of this says religion should not be studied. It should, and intensively. Religion as a historical phenomenon, a sociological one, a psychological one, a literary one, is an entirely legitimate object for departments of history, anthropology, sociology, psychology and literature. In those departments religion is examined from the outside, as a human construction, with the same tools applied to any other human construction.

Theology is not that. Theology does not study religion as an external object. It operates from inside the unverified assumptions of the dogma, taking as given exactly the things the other disciplines would set out to explain. The sociologist asks why people believe in resurrection. The theologian asks what resurrection means, having already assumed it happened. One of these is a research question. The other is devotional practice wearing a gown.

The academy's reluctance to say this out loud is of a piece with its reluctance to blame religion for anything. But the principle is already accepted; it is only the application that is missing. We agree that astrology has no place in the university because it lacks evidence and logic. The same standard, applied honestly, evicts theology on the same day. It belongs in seminaries, in houses of worship and in private reflection, and in none of those places does it need a university's letterhead.

Why People Believe the Bible: The Social, Psychological and Cognitive Machinery

2026-10-01T00:00:00+00:00

Billions of people hold a collection of ancient Near Eastern texts to be the word of God, and almost none of them arrived at that position by weighing the evidence for it. That is not an insult; it is the starting point for a genuine question. If belief in the Bible is not sustained by verification, what does sustain it? Sociology, psychology and cognitive science each have a part of the answer, and when the parts are laid side by side the picture is of a belief held up by human social systems, human emotional needs and the evolved architecture of the human brain, and not at any point by the text being true.

I want to go through the machinery in those three layers, because the layers are different and are confused with each other more often than they should be.

Belief is inherited before it is examined. The single strongest predictor of a person's religion is their parents' religion, and the second strongest is the religion of the place they were born. I have written about the geography of belief before, and the point bears repeating here: children absorb the worldview of their family and their surroundings years before they develop the critical faculties to scrutinise a historical, scientific or textual claim. By the time those faculties arrive, the framework is no longer a claim under consideration. It is the background against which all other claims are judged. The believer does not remember adopting it, because there was never a moment of adoption.

The text is a membership card. A shared sacred book binds a community. Allegiance to it buys entry into a structured society: an identity, a calendar, a set of rituals, a safety net and a circle of people who will show up when things go wrong. The price of leaving is correspondingly steep. It is not the loss of a belief, which costs nothing, but the loss of family, friends and standing, which costs everything. A person weighing whether the book is true is also, whether they admit it or not, weighing what the honest answer would do to their life, and the second calculation usually settles the first.

Institutions defend the text so the believer does not have to. Over two millennia, well-organised institutions have built enormous bodies of apologetics whose whole purpose is to harmonise the internal contradictions, reframe the conflicting external evidence and supply the believer with a reassuring answer to any question before they think to ask it. The individual believer rarely does this work. It is done for them, and the existence of a confident answer somewhere is enough to let the question drop.

The Psychological Layer

The text manages the fear of death. Human beings have a hard time with mortality, with suffering and with not knowing what comes next. The biblical narrative addresses all three in one package: suffering has a purpose, justice is guaranteed in the end, and death is not the end. Giving up the package means facing those three things directly, without a story. Most people, offered the choice between a comforting account and an honest uncertainty, do not experience it as a choice at all.

The text supplies moral absolutes. Working out an ethics for oneself, in negotiation with other people and in the face of hard cases, takes real cognitive and emotional effort, and the result is always provisional. A sacred text offers a finished set of rules with external authority behind them. The idea of an objective, divinely ordained moral order is attractive precisely because it relieves the believer of the burden of thinking the questions through, which is a point I have made from the other direction when arguing that religion takes away your basis for thinking.

The text works as a coping mechanism. Prayer, the promises of scripture and the sense of being watched over give measurable relief in times of illness, grief and crisis. The relief is real. It does not depend on anyone listening, but the believer attributes it to the text, and every crisis survived strengthens the attachment.

The Cognitive Layer

The brain over-detects agency. Evolutionary psychology describes a bias that was once a survival asset: when something unexplained happens, assume an agent did it. Treating a rustle in the bushes as a predator rather than the wind costs little when wrong and saves your life when right, so the brain errs heavily towards seeing intention. The same bias, pointed at the natural world, sees design in the eye, purpose in the storm and a will behind the universe. The biblical God is what this detector reports when there is nothing there to detect.

Once accepted, the belief protects itself. After a text is taken as authoritative, ordinary confirmation bias and motivated reasoning stand guard over it. Good outcomes become answered prayers. Bad outcomes and contradicting evidence are rationalised, minimised or compartmentalised. I have argued that religious confirmation bias is larger than the ordinary kind, because the believer is defending not a view but an identity.

The brain is a pattern engine. Confronted with randomness, complexity or chaos, the mind does not report randomness. It constructs a story, because a story is what it is built to produce. A text that offers one overarching story, in which everything from the origin of the universe to this morning's misfortune has its place, is intuitively satisfying in a way that "there is no pattern here" can never be.

What the Three Layers Add Up To

Notice what is absent from every one of these mechanisms. Not one of them involves the Bible being accurate. Each would operate identically for any sacred text, and each in fact does: the same social inheritance, the same existential comfort, the same agency detection sustain the Quran in Riyadh, the Vedas in Varanasi and the Book of Mormon in Salt Lake City. The machinery is indifferent to content. It will hold up whatever text it is handed.

That is the point of laying it out. The believer experiences their conviction as a response to the text. It is in fact a response to their parents, their community, their fear of death and the wiring of their own brain, and the text is the object those forces happen to have attached themselves to. Understanding this does not refute the Bible, which needs no further refutation. It explains why the refutations never seem to land, and it tells you where the belief actually lives, which is nowhere near the evidence.

Why Believers Wear Symbols and Non-Believers Do Not

2026-09-30T00:00:00+00:00

Walk through any city and you can pick out the religious. Crosses on chains, tzitzit under shirts, hijabs, turbans, clerical collars, a tilak on the forehead, a kippah, a particular hat. Try the same exercise for the non-religious and you will fail. There is no atheist garment, no humanist jewellery that anyone is expected to wear, no rationalist mark on the body. A few people own a scarlet-letter pin or a Darwin fish for the car, and nearly nobody wears either daily.

This asymmetry is old and cross-cultural, and I think the usual explanations for it are too shallow. Symbols, we are told, express piety, or signal belonging, or do identity politics. All true, and all beside the main point. The interesting question is why one kind of worldview needs a visible badge and the other does not, and the answer says something about the worldviews rather than the people who hold them.

The Minority Explanation Fails

The first explanation anyone reaches for is that persecuted or minority groups need markers to find allies. There is something in this, but it cannot be the main driver, because religious groups wear their markers most heavily exactly where they are least threatened. The cross is most visible in Christian-majority countries. Orthodox dress is most elaborate in Orthodox neighbourhoods. A religion that commands overwhelming demographic and institutional dominance still produces symbols on every believer, which is the opposite of what the minority-defence theory predicts.

Meanwhile non-believers, who have been a persecuted minority for most of recorded history and still are in much of the world, never coalesced around a daily physical marker, despite sharing a perfectly coherent set of commitments: rationalism, humanism, empirical inquiry. If isolation produced badges, atheists would wear more badges than anyone on earth. They wear the fewest.

So the driver is not demographic. It is something internal to the belief, an epistemic and social requirement that one kind of worldview has and the other lacks.

Symbols as Costly Signals

Start with the part of the standard account that is right. From the evolutionary and sociological side, a visible symbol is a signal, and often a costly one. Wearing it demonstrates commitment to the group's shared reality in a way that words do not. It makes recognition instant, lowers the cost of coordinating with fellow members, and lets members verify one another at a glance. I have written about courtship as a signalling game, and the same logic applies: a signal that costs nothing carries no information, and a signal that costs something is believed.

That explains why symbols are useful to a group. It does not yet explain why some groups need them and others do not.

Fragile Beliefs Need External Support

Here is the part I think is missing. Religious belief systems make claims that are extraordinary by the standards of everyday experience: a man rose from the dead, a sea parted, a book was dictated by the creator of the universe, physical law was suspended on particular occasions for particular people. None of this is available to the believer's own senses. The everyday material world does not confirm any of it, and in a quiet way it contradicts all of it, every day, because the dead stay dead and seas stay where they are.

A belief that clashes with ordinary sensory reality is cognitively fragile. Left alone with the world, it erodes. To survive it needs continuous reinforcement from somewhere other than the world, and the only other source available is other people.

Non-belief has no such problem. Its baseline commitments, that the physical world is consistent, that evidence is how you find things out, that the universe does not make exceptions for anybody, are confirmed by every waking moment. The non-believer walks through a world that agrees with them without anyone having to arrange it. There is nothing to shore up, so there is no need for a badge.

Set the two side by side and the asymmetry falls out:

Dimension	Non-believers	Believers
What anchors the worldview	Independent physical reality, empirical test, universal sensory consistency	Communal tradition, sacred narrative, shared dogmatic consensus
Need for artificial validation	Low; the universe reinforces the premises without social help	High; without physical proof, the consensus must be mirrored back constantly
What external symbols do	Incidental and individual; autonomy discourages compulsory badges	Anchor mutual reinforcement, identify allies, verify a shared metaphysics

Symbols as Mutual Reality Construction

Sociology has a name for the structure that keeps an improbable worldview plausible. Berger and Luckmann, in The Social Construction of Reality, called it a plausibility structure: the web of people, institutions and routines that makes a set of beliefs feel self-evidently true to those inside it. A worldview that asserts extraordinary things needs a robust one, and visible symbols are among its load-bearing parts. They work in three ways.

They reflect the belief back. Seeing another person wear the cross or the tzitzit confirms that one is not alone in holding to claims that the senses do not support. Every such sighting is a small vote of confidence from outside one's own head.

They quiet dissonance. At some level every believer registers the gap between what they are asked to believe and what they see. The constant visible presence of others wearing the same marker is a running reassurance that the narrative has collective legitimacy, that so many people cannot all be wrong. The reassurance is statistically worthless and psychologically indispensable.

They convert private doubt into public assertion. Putting the symbol on is itself an act of commitment, performed daily, in front of others. It is much harder to doubt in private what you have declared in public, and the symbol makes the declaration continuous. I have argued that believers largely do not believe in the way they claim to; the symbol is one of the devices that keeps the gap from becoming visible, to others and to themselves.

What the Asymmetry Tells You

The difference in symbolic display between believers and non-believers is not a difference in temperament, in tribalism or in courage. It is a difference in epistemic architecture. One worldview is held up by the world and needs nothing else. The other is held up by other people and needs them visibly, continuously and in uniform.

That is why the badges cluster where the claims are most extraordinary, why they persist where the group is most dominant, and why no amount of persecution ever produced an atheist equivalent. You do not need a symbol to remind you that the sun will rise. You need one to remind you of the things the sunrise will never confirm.

Is Religion an Ideology?

2026-09-29T00:00:00+00:00

Call religion an ideology in front of a believer and you will get a reaction. Call liberalism a religion in front of a liberal and you will get the same reaction. Both flinches are telling: each side senses that the label is doing work, and that the work is not merely descriptive. So it is worth asking the question slowly. Is religion a type of ideology? The honest answer is that it depends on which of the two lives of the word "ideology" you have in mind, and that religion fits one of them cleanly and the other only sometimes.

The Descriptive Sense

In its neutral, sociological sense, an ideology is a comprehensive system of ideas that does three things. It explains how the world is. It says how the world ought to be. And it binds a group of people around that shared picture, so that holding the picture becomes part of who they are.

Under this definition religion is not merely like an ideology. It is one, and probably the oldest and most complete example we have. Every major religion has a cosmology, an ethics, and a community. Every one of them tells its members where the world came from, what a good life looks like, and who counts as one of us.

Clifford Geertz saw this clearly. He wrote two companion essays, "Religion as a Cultural System" and "Ideology as a Cultural System", and treated both as the same kind of thing: a system of symbols that supplies a model of reality and a model for acting in it. The parallel titles were not an accident. Durkheim gets to the same place from the other direction. For him religion is whatever sacralizes a group's shared values and holds the group together, which means that nationalism, with its flags, anthems, martyrs and holy days, is doing religion's job. Robert Bellah gave that thought a name, "civil religion", when he applied it to the United States.

So on the descriptive definition the question answers itself. Religion is a species of the genus ideology, and the only interesting work left is to say what distinguishes the species.

The Pejorative Sense

But "ideology" has a second life, and it is the one most people hear. In the Marxist sense an ideology is a set of ideas that serves the interests of a class or a ruling power while presenting itself as natural, eternal and true. Ideology in this sense is not just a worldview; it is a worldview that lies about where it came from.

Religion is the founding case. Marx's line about the opium of the people is only the most quoted version of a much older suspicion. And the suspicion is often justified. The divine right of kings told the peasant that his king was chosen by heaven. Karma told the untouchable that his caste was the just consequence of a previous life. "The poor you will always have with you" has been read, for two thousand years, as a reason not to do much about the poor. In each case a social arrangement that benefits someone is presented as the order of the cosmos, and the person it does not benefit is told that questioning it is questioning God.

Yet the fit is not universal. Religion has just as often been the ideology against the ruling order. The Hebrew prophets denounced kings and priests in God's name. Millenarian revolts from the Peasants' War to the Taiping Rebellion were driven by scripture. Liberation theology read the Gospels as a manifesto for the poor against the landowners. Early Christianity itself was a movement of slaves and women that Rome persecuted precisely because it refused the civic cult. Whether religion serves power or attacks it is a historical question, not a definitional one.

So in the pejorative sense religion is often an ideology, but not by nature. It is a tool that has been in the hands of both sides.

What Religion Adds

If religion is a species of ideology, what marks the species? Three features, I think, and they are worth separating because they are usually run together.

A transcendent referent. An ideology like liberalism or socialism makes claims about the social and natural order. Religion makes claims about a reality beyond that order: a God, a soul, an afterlife, a cosmic law. This is a difference in subject matter, and it has a consequence. A political ideology can, at least in principle, be refuted by events. When the prophecy fails, the movement is in trouble. A religion's central claims are placed where events cannot reach them, and that is not an accident of design.

Ritual and embodied practice. Ideologies have rallies, slogans, songs and anniversaries. But religion's practice is denser and it is tied to the body and to the life cycle. Birth, coming of age, marriage and death all pass through it. Daily prayer, dietary law, dress, the calendar and the week are all shaped by it. A person can hold a political ideology and do almost nothing about it. A religion is something you do, and the doing is what makes it stick.

An epistemology of revelation. This is the deepest difference. Modern ideologies claim to be the product of reason or science. Marx called his work scientific socialism. Liberalism presents itself as what any rational person would agree to. Even when these claims are false, they are the claims being made, and they invite the reply "show me". Religion grounds itself in revelation, tradition and faith, and it says so openly. It does not claim that you could have worked it out yourself. It claims that it was told.

This last difference is the one that matters most in practice, because it decides what kind of argument you can have. You can argue with a socialist about whether the theory of surplus value is correct. The argument may go nowhere, but both sides agree on what would count. With a claim from revelation there is nothing to check, and the believer will tell you as much.

Why the Label Is Contested

Once you see the three differences you can also see why the label is fought over. Calling religion an ideology is rarely a neutral act. It strips religion of the special standing it enjoys in law, in tax codes, in schools and in manners, and puts it on the same footing as liberalism or socialism, where it can be argued with, mocked and voted against. That is usually the whole point of the move, and the people making it know it.

The reverse move is just as rhetorical. Believers call atheism "also a faith" and call secular humanism "a religion" in order to drag secular worldviews down to the same level. If everything is a religion then religion is nothing special, but also nothing to be ashamed of, and the atheist's claim to be reasoning rather than believing is dismissed as self-flattery.

Both moves are fighting over the same prize: whether religion gets to be exempt from the ordinary rules of argument. The descriptive claim, that religion is a system of ideas which explains, prescribes and binds, is simply true. What the fight is really about is whether the three features that distinguish religion from other ideologies earn it a pass, or whether they are exactly the features that should make us more suspicious of it, not less.

I know which way I lean. A system of ideas that places its central claims beyond the reach of evidence, embeds them in the body before the age of reason, and grounds them in an authority that cannot be questioned has not earned an exemption from argument. It has described the conditions under which argument is least likely to happen. That is not a reason to treat it more gently than any other ideology. It is a reason to treat it with more care.

Phones Moved Toward Freedom One Concession at a Time, and So Does Power

2026-09-28T00:00:00+00:00

I have watched one small technology move from total control to something close to freedom within my own working life, and the path it took is worth recording, because it is the same path I expect power in society to take.

Stage One: The Carrier Owns the Phone

Anyone who used a mobile phone before 2007 remembers what it was. The carrier chose the handset, the handset ran whatever firmware the manufacturer and the carrier agreed on, and the applications on it were the ones the carrier had been paid to preload. You could not install anything. You could not remove anything. A ringtone cost money because the carrier decided ringtones cost money, and the device you had paid for enforced that decision against you.

Nobody called this oppression, because nobody had seen the alternative. The phone was a terminal for the carrier's business, and the person holding it was the last thing it served.

Stage Two: A Different Master

Then came a phone that let you install software. That sounds like freedom and it was a real step toward it, but look at the structure. One company decided what software existed, what it could do, what it could charge, and what it could say. The carriers lost control and did not get it back. The user gained a store, but the store had a single owner, and the owner's approval stood between every developer and every customer.

This is an interim stage, and it is a recognisable one. The old lord has been displaced by a new one, the new one is genuinely better, and the improvement is real. A great many people concluded that this was the destination. It was not. It was a change of master with the shape of the arrangement intact.

Stage Three: The Rules Become Visible and Contestable

The next platform published its source code, let anyone build a store, let anyone install software from anywhere, and let manufacturers ship it without asking. It did not deliver perfect freedom; a company still steers it, and most people still live inside one store. But the rules moved from private to public, and the thing that had been a decree became an argument you could have.

That is what the freedom of a platform actually consists of. Not the absence of rules, which is impossible, but rules that are written down, applied to everyone, and changeable by people other than the ruler.

Each of these steps was resisted by whoever held the previous position, each was called impossible or dangerous before it happened, and each is now taken for granted by people who cannot imagine the earlier state.

The Pattern Is Not About Phones

Put the three stages side by side with the history of government and the correspondence is embarrassing.

A monarch owns the state the way a carrier owned a phone: the population is the last thing it serves, and it enforces the owner's business against the people holding it. Representative government is the phone with a store: a genuine step, a real gain in what you may do, and a single set of owners deciding what is on offer. The important thing about the second stage is that people mistake it for the end, exactly as they did with the phone.

The third stage is the one where the rules of power become visible and contestable by the people they apply to. That is what I mean by direct democracy, and the analogy tells me something about how it will arrive. Not by argument, because the arguments were all available at stage one. By the slow discovery that the next concession is survivable, made by people who were certain it was not.

What the Analogy Predicts

Three things, and I hold them with the confidence the analogy earns, which is moderate.

First, the movement is one way. No platform has gone back from open to carrier-locked, and no population that has argued about its rules in public has volunteered to stop. Reversals happen by force, not by preference.

Second, each stage is defended by the beneficiaries of the previous one with the same vocabulary: security, stability, the incompetence of users. The carriers said people could not be trusted to install software. The claim was false about phones and it is false about decisions.

Third, the stage after the one you are standing in always looks like chaos from where you stand, and like the obvious arrangement once you are in it. I have argued elsewhere that the current arrangement should be read as a spec rather than as a fact of nature. Phones are the small demonstration that specs change, and that the people who wrote the old one are never the ones who write the next.

Human Intelligence Is Probably on the Lower Rungs of the Ladder

2026-09-27T00:00:00+00:00

Picture intelligence as a ladder. At the bottom are thermostats and bacteria, a little higher the insects, then the rats, the crows, the dogs, the apes, and somewhere above them, us. The question is what is above us, and the default human answer, never stated but always assumed, is: nothing much. We are the rational animal, the measure of all things, the top of the ladder, and if there is a rung above it is reserved for gods.

I think the evidence points the other way. Human intelligence is probably near the bottom of the range that intelligence can occupy, just above the line that allows a species to accumulate culture at all, and we mistake that line for the ceiling because it is the only rung we have ever stood on. Here is the case, one data point at a time.

The Overrun

The first data point is the one we have all just lived through. For seventy years the project of making a machine think was a byword for over-promising. Then, in the space of about six years, machines went from writing grammatical nonsense to passing the bar exam, solving competition mathematics, writing working software, and out-arguing most people on most topics. Nothing profound was discovered in those six years. The algorithm at the core, the transformer, was published in 2017 and is a few pages long. What changed was that somebody finally applied enough data and enough compute, and the human level went by like a road sign.

That is the part worth staring at. If human intelligence were near some natural ceiling, you would expect machines to approach it asymptotically: the closer they got, the harder each step would become, and the human level would show up in the curves as a landmark. It does not show up at all. The scaling laws that describe how these systems improve with compute are smooth power laws, and there is no kink, no plateau, no special point where the humans are. The human level is not a feature of the landscape. It is just a height, and it was passed at the same speed as the heights below it.

The same thing happened, in miniature, in every game. Chess fell in 1997 and the engines then kept climbing until the best human could no longer take a single game from a phone. Go fell in 2016, and the striking thing was not the defeat but what came after: the machines kept going, and within a few years human professionals were studying their games the way a student studies a master, relearning openings that had been settled for centuries. The human peak turned out to be a foothill. Every time, the assumption was that the machine would reach us and stop, because we were where intelligence topped out. Every time, it reached us and continued without noticing.

Logic Does Not Come Naturally

The second data point is older and does not involve computers at all. Consider how hard it is to teach a human being to reason.

Formal logic is about the simplest intellectual tool there is. A syllogism has two premises and a conclusion. Modus ponens is one rule. Yet the species took until Aristotle, roughly 350 BCE, to write the syllogism down, and another two thousand two hundred years, until Frege in 1879, to arrive at a logic strong enough to express ordinary mathematical statements. That is not the timeline of a faculty we possess natively. It is the timeline of something being painfully excavated against the grain of the mind doing the excavating.

And once excavated, it does not transfer. Give people the Wason selection task, four cards and a single if-then rule, and fewer than one in ten choose the cards that test it, university students included. Give them the Linda problem and most rate "bank teller and feminist" as more probable than "bank teller", which no possible state of the world can make true. Ask for the probability that a positive result on a 99%-accurate test for a rare disease means you have the disease, and doctors get it wrong by a factor of ten or more.

I have written before about how consistent these findings are; the point here is what they imply about the ladder. We put every child through twelve years of school, and most adults leave unable to follow a proof by induction, unable to hold a conditional probability steady in their head, and unable to see that "all A are B" does not give you "all B are A". A capacity that takes millennia to discover and then a decade of drill to install in a minority of individuals is not a native capacity of the species. It is a prosthetic that fits badly.

A Civilisation of Prosthetics

Which leads to the third data point: look at what we had to build in order to think at all.

Working memory holds about four items. That is the whole scratchpad. It is why nobody can multiply two four-digit numbers in their head, why a phone number is chunked into groups, why a chess player who can see fifteen moves ahead is a prodigy. Every intellectual achievement of the species has been an achievement of moving the work out of the head. Writing exists so that a thought can survive longer than the attention span that produced it. Place-value notation with a zero, which reached Europe only around 1200 and took another three centuries to displace Roman numerals, exists because the mind cannot do arithmetic on its own. Algebraic notation, the checklist, the diagram, the table, the calculator: each is a confession that the naked brain could not do the task.

A species that reasons well would not need most of this. It would need paper for the way we need a bicycle, as a convenience. We need it the way a patient needs a ventilator. The whole intellectual history of humanity is the history of finding ways to route thinking around the organ that is supposed to do it, and that is not the biography of a mind near the top of anything.

The pace of that history says the same. Anatomically modern humans, with brains identical to ours, existed for something like three hundred thousand years before anyone farmed, and for another several thousand before anyone wrote. That is a long time for a supposedly brilliant species to sit next to the raw materials of civilisation without picking them up. The likelier reading is that we were barely over the threshold at which cumulative culture is possible, so that it took an immense number of person-lifetimes to accumulate the first few steps.

The Hardware

Fourth: look at the machinery, as an engineer would.

A neuron fires at most a few hundred times a second. A signal travels along an axon at around a hundred metres per second. A transistor switches billions of times a second and a signal on a wire moves at a respectable fraction of the speed of light. We are seven orders of magnitude slower per element, and we compensate with parallelism, which is why we are good at recognising faces and bad at long division.

The bandwidth out is worse. Measured across languages, human speech carries information at about 39 bits per second, and reading is not much faster. Everything you have ever thought that made it to another mind was squeezed through that straw. Two humans working together cannot share a mental state; they can only exchange lossy compressed summaries of it, and then spend hours discovering they meant different things. A design that could copy a thought exactly, or run two lines of reasoning at once, or attend to more than one thing at a time, would be a different kind of mind, and there is no law of nature that forbids it. There are only the accidents of a wet, slow substrate that happened to be the one evolution had to work with.

And consider the sheer smallness of the differences that matter. A dog has about half a billion cortical neurons. We have around sixteen billion. That is a factor of thirty, achieved by the same cell types, the same wiring rules, the same neurotransmitters, and it is the whole distance from fetching a stick to composing a fugue. If thirty times gets you from the dog to Bach, it is a strange confidence that says the next thirty times gets you nothing.

Evolution Stops at Good Enough

Fifth, and to my mind the strongest: think about how we got the brain we have.

Natural selection does not maximise anything. It pushes a trait until its marginal cost equals its marginal benefit and then stops. The brain is about two per cent of body mass and burns about twenty per cent of the energy at rest. A larger one would have starved on the savannah. It also has to pass through a pelvis built for walking upright, which is why human infants are born earlier and more helpless than any other primate, and why childbirth killed a shocking fraction of mothers for the whole of history before modern medicine. Every gram of extra cortex was paid for in dead women and dead infants. Selection stopped adding cortex the moment the price exceeded the return, and the return it was measuring was reproduction on the African plains, not calculus.

Now add one more consideration. The first species to cross the line where cumulative culture becomes possible will, by construction, be barely across it. Once a species can pass knowledge down the generations, technology takes over from biology as the thing that changes, and the biological pressure to get smarter largely goes away, because a slightly cleverer individual gains little that a book cannot give a duller one. So whichever species crosses that line first freezes just above it. We are that species. We are not the smartest thing evolution could produce. We are the dumbest thing that could produce a civilisation, because as soon as that was possible it happened, and the process that made us smarter stopped.

The chimpanzee makes the point from below. We share something like 98.8% of our genome with it, six or seven million years of separation, and it cannot learn arithmetic, cannot conceive of next year, and, crucially, cannot conceive that there is anything it fails to conceive. There is no reason to think the small genetic distance that produced us used up the design space. It is the first small distance, not the last.

We Cannot See Our Own Reasoning

Sixth: an intelligent system ought at least to know what it is doing. We do not.

Nisbett and Wilson showed in 1977 that when people are asked why they made a choice, they confidently report reasons that demonstrably played no role, and cannot tell the difference between a real reason and an invented one. Split-brain patients, when the right hemisphere acts on an instruction the left cannot see, produce fluent explanations for the action from the left hemisphere that are simply made up, delivered with total conviction. Swap the photograph someone just chose for the one they rejected and most will explain, in detail, why they preferred the one they did not choose. We do not have access to our own reasoning. We have access to a press office that issues statements about it.

Add the rest of the familiar list: optical illusions that persist after you know they are illusions, anchoring on a number you know is random, judging an argument by the attractiveness of the person making it, and the finding that the least competent people are the most confident in their competence. None of these are failures of attention or education. They are the operating characteristics of the device. A mind that cannot inspect itself, cannot correct for its known biases even when told about them, and cannot tell its confabulations from its inferences, is not a mind near the top of the scale. It is a mind that works well enough to have got this far.

The Objections

But we built the machines. So did Arthur Samuel build a checkers program that beat him at checkers in the 1950s, and every chess engine since has beaten its programmers. The builder is not the ceiling. A student who surpasses a teacher is the most ordinary thing in the world, and a process that surpasses its designer is what evolution is.

But the machines still make stupid mistakes. They do, and so do we, in different places. The argument is not that any particular system is smarter than any particular person. The argument is about the shape of the space: that the human level was passed, on task after task, without any sign of it being a hard place to pass. What the machines do badly today is being fixed on a timescale of months. What we do badly has been the same for the whole of recorded history.

But intelligence is more than logic and games. Every time a machine does something, the definition of intelligence is rewritten to exclude it. Arithmetic was intelligence until calculators. Chess was the pinnacle of thought until 1997. Translation, protein folding, passing professional examinations, writing code, each was the real test until it was passed, at which point it turned out to have been mere pattern matching all along. This is not a definition being refined. It is a species protecting its self-image, and the persistence of the reflex is itself a data point about how we reason.

What Follows

None of this is a reason for despair, and it is not an argument for anything in particular about machines. It is an argument about us, and it belongs with the long series of demotions that make up the history of science: not the centre of the cosmos, not a separate creation, not the possessor of a soul, and now not the top of the ladder either. Each time, the demotion turned out to be the more interesting picture.

The practical consequence is humility of a specific kind. A chimpanzee in a zoo does not know that it is missing calculus; it is not frustrated by its ignorance of number theory, because the concept of the thing it lacks is itself among the things it lacks. We should assume we are in exactly that position with respect to whatever is above us on the ladder. There are almost certainly thoughts we cannot think, arguments we cannot follow, and structures in the world we cannot see, not because they are hidden but because the organ we are looking with is a small, slow, jury-rigged thing that evolution stopped improving the moment it was barely good enough. The one thing that organ has managed to do, against its own design, is to notice that fact. That may be the most intelligent thing it has done, and it should make us very cautious about any argument whose conclusion is that we are as clever as things get.

The Bible Is Monolatrous, Not Monotheistic: Every Reason in One Place

2026-09-26T00:00:00+00:00

Ask almost anyone what the Hebrew Bible teaches about God and you will hear "monotheism": there is exactly one God, and everything else people have ever worshipped is a fiction. That is the doctrine of later Judaism, of Christianity and of Islam, and it is read back into the text so automatically that most readers never notice the text itself saying something different.

What the Bible mostly says is monolatry: worship one god, because other gods exist and you are not to serve them. The distinction matters. Monotheism is a claim about what exists. Monolatry is a claim about loyalty. "There is no other god" and "you shall have no other god" are different sentences, and the Bible overwhelmingly says the second.

I have written before about the polytheistic background of ancient Israel, mostly from the archaeology. This post is the textual companion: the reasons, gathered in one place, ordered from the strongest to the weakest. I put the weakest last on purpose, and I say why they are weak, because a list padded with bad arguments is worth less than a short list of good ones.

1. Yahweh is one of the sons of the Most High, and Israel is his allotment

Deuteronomy 32:8–9, in the Dead Sea Scrolls and the Greek Septuagint, reads:

When the Most High (Elyon) gave the nations their inheritance, when he divided humanity, he fixed the borders of the peoples according to the number of the sons of God. For Yahweh's portion is his people; Jacob is his allotted share.

The scene is a family estate being divided. The high god, Elyon, splits the nations among his sons, and Yahweh receives Israel as his portion. Yahweh is not the one doing the dividing. He is one of the heirs.

The Masoretic text, the medieval Hebrew version behind most modern Bibles, reads "sons of Israel" instead of "sons of God" here. That reading makes no sense (why would the number of nations match the number of Jacob's sons?), and the Qumran fragment 4QDeut^j shows the older wording. Someone found the verse embarrassing and fixed it. That is the smoking gun twice over: the passage itself describes Yahweh as a member of a pantheon, and the later scribes who changed it evidently agreed that this is what it says.

2. The divine council: God presides over other gods

Psalm 82 opens:

God stands in the divine assembly; in the midst of the gods he holds judgment.

It then has God rebuke those gods for judging unjustly, and sentence them:

I said, "You are gods, sons of the Most High, all of you; nevertheless you shall die like men, and fall like any prince."

You cannot sentence beings to death for misconduct if they do not exist. The psalm is a demotion of the other gods, which is a very different thing from a denial of them. It is, if anything, a snapshot of the exact moment monolatry starts turning into something stronger.

The council is not a one-off. In 1 Kings 22:19–22 the prophet Micaiah sees Yahweh on his throne "with all the host of heaven standing beside him", and a discussion follows in which a spirit volunteers to go and deceive Ahab's prophets. In Job 1–2 "the sons of God came to present themselves before Yahweh", and the Adversary is among them. Psalm 89:6–7 asks "who among the sons of the gods is like Yahweh, a god feared in the council of the holy ones". Psalm 29:1 tells the "sons of the gods" to ascribe glory to Yahweh. And Genesis has God speaking to his peers three times: "Let us make man in our image" (1:26), "the man has become like one of us" (3:22), "let us go down and confuse their language" (11:7). The traditional explanation that God is addressing angels only relocates the problem: the beings he is addressing share his image and his knowledge of good and evil.

3. The first commandment is a loyalty oath, not a statement of fact

The commandment (in the Jewish count, the first two) does not open with theology. It opens with a favour:

I am Yahweh your God, who brought you out of the land of Egypt, out of the house of slavery. You shall have no other gods before me. (Exodus 20:2–3)

The structure is that of an ancient Near Eastern treaty: the great king recalls what he did for the vassal, then demands exclusive allegiance. The Hebrew "before me" is al panai, "in my presence" or "in preference to me". The commandment forbids having other gods; it does not say there are none to have. Two verses later comes the reason, and it is jealousy: "for I, Yahweh your God, am a jealous god". Nobody is jealous of a rival who does not exist. Exodus 34:14 is blunter still: "you shall not bow down to any other god, for Yahweh, whose name is Jealous, is a jealous god."

The same logic runs through the rest of the law. Exodus 22:20 orders the death of "whoever sacrifices to any god other than Yahweh alone", a law that would be pointless if the alternative were an empty category. Joshua 24:14–15 puts the choice in front of the people as a menu: put away the gods your fathers served beyond the river and in Egypt; "choose this day whom you will serve, whether the gods your fathers served or the gods of the Amorites in whose land you live; but as for me and my house, we will serve Yahweh." That is an election speech, not a proof of existence.

Even the Shema, the verse later Judaism made its monotheistic creed, is better translated as loyalty than as arithmetic. Deuteronomy 6:4, "Yahweh our God, Yahweh is one", reads most naturally as "Yahweh is our God, Yahweh alone", and the next verse, "you shall love Yahweh your God with all your heart", is treaty language for exclusive allegiance, not affection.

4. Other nations' gods are real, and their territory is legitimate

The Bible repeatedly treats the gods of other nations as existing and as having legitimate claims outside Israel.

Jephthah, negotiating with the king of Ammon, says (Judges 11:24): "Will you not possess what Chemosh your god gives you to possess? So whatever Yahweh our God has given us, we will possess." Each nation has its god; each god hands out land. Micah 4:5 says the same in a prophet's voice: "all the peoples walk each in the name of its god, and we will walk in the name of Yahweh our God for ever and ever."

Deuteronomy 4:19 goes further and makes the arrangement Yahweh's own doing. Israel must not worship the sun, moon and stars, "which Yahweh your God has allotted to all the peoples under the whole heaven". Other nations worshipping the heavenly bodies is not an error to be corrected. It is the division of labour Yahweh himself set up; the same division as Deuteronomy 32:8 above, seen from the other side.

Once you notice this frame, several strange passages make sense. In 2 Kings 3:27 the king of Moab, losing a battle, sacrifices his firstborn son on the city wall, "and great wrath came upon Israel, and they withdrew from him". Chemosh, the god of Moab, is appeased, and the Israelite army retreats. In 1 Samuel 26:19 David, driven out of Israel's territory, complains that his enemies have in effect told him "go, serve other gods", because outside Yahweh's land one is outside Yahweh's jurisdiction. In 2 Kings 5:17 Naaman the Aramean, having decided to worship Yahweh, asks for two mule-loads of Israelite earth to take home so that he can do it on Yahweh's own soil. In Ruth 1:15–16 Naomi says Orpah "has gone back to her people and to her gods", and Ruth's famous reply pairs the two: "your people shall be my people, and your God my God." Changing nation means changing god. That is how the ancient world worked, and the text never says otherwise.

5. "Who is like you among the gods?"

The Bible's favourite way to praise Yahweh is to say that no other god matches him. That is a comparison, and comparisons need a class.

The Song of the Sea, one of the oldest texts in the Bible, asks "Who is like you among the gods, O Yahweh?" (Exodus 15:11). Jethro concludes from the exodus that "now I know that Yahweh is greater than all gods" (Exodus 18:11). The Psalms are full of it: "there is none like you among the gods" (86:8), "Yahweh is a great king above all gods" (95:3), "he is to be feared above all gods" (96:4), "you are exalted far above all gods" (97:9), "our Lord is above all gods" (135:5). Deuteronomy 10:17 calls him "God of gods and Lord of lords", and Psalm 136 uses the same title.

"Greatest of the gods" is a polytheist's compliment. Marduk was greatest of the gods in Babylon, Zeus in Greece. A monotheist has nobody to compare his god to, and would not think of doing so.

6. Yahweh fights the gods of Egypt, and Pharaoh's magicians can do magic

The plagues are described in the text itself as a battle between gods. "On all the gods of Egypt I will execute judgments: I am Yahweh" (Exodus 12:12). Numbers 33:4 repeats it as a narrator's statement of fact: the Egyptians were burying their dead "while Yahweh executed judgments on their gods".

Inside the story, Pharaoh's magicians are not frauds. When Aaron's staff becomes a serpent, they "did the same by their secret arts" (7:11–12). They turn water to blood (7:22) and produce frogs (8:7). They only fail at the gnats, and their own verdict, "this is the finger of a god" (8:19), is a rival power conceding a round. A monotheistic narrator would have written that the magicians faked it. This narrator writes that they did it and were outclassed. Egypt has real supernatural power; Yahweh's is just greater. That is the plot of a contest, and contests need two sides.

7. The sons of God took human wives

Genesis 6:1–4 says that "the sons of God saw that the daughters of men were fair, and took wives of all they chose", and that their offspring were the Nephilim, "the mighty men of old, the men of renown". Divine beings mating with mortals and fathering heroes is a stock motif of Near Eastern and Greek mythology, and the passage tells it straight, with no hint that these sons of God are anything but what the phrase says. The later effort to make them angels, or the descendants of Seth, is the reader's discomfort, not the text's.

8. Elohim is plural, and sometimes the grammar admits it

This is the argument most people reach for first, and on its own it is the weakest, so it belongs down here. Elohim is plural in form, but when it refers to the God of Israel it nearly always takes singular verbs and adjectives, and Hebrew does use plural forms for abstract nouns and for majesty. "Elohim is plural, therefore many gods" is not a sound inference.

What does carry weight is the handful of places where the plural agreement leaks through even for Israel's God. Abraham says "when gods caused me to wander from my father's house" (Genesis 20:13, plural verb). At Bethel "the gods revealed themselves to him" (Genesis 35:7, plural verb). David asks what nation is like Israel, "whom gods went to redeem" (2 Samuel 7:23, plural verb). Joshua warns that Yahweh is "holy gods" (Joshua 24:19, plural adjective). Deuteronomy 5:26 speaks of the voice of the "living gods". The scribes standardised the grammar almost everywhere; these are the places they missed.

More telling is what else the word is used for. Chemosh is the elohim of Moab and Milcom the elohim of Ammon (1 Kings 11:33). When the medium at Endor raises Samuel she says "I see elohim coming up out of the earth" (1 Samuel 28:13), and the elohim she sees is a dead man's ghost. So elohim is a category word for beings of the divine sort, and the God of Israel is described as the greatest member of the category, not as the only thing the word can mean.

9. The Israelites worshipped other gods, and the text records it as ordinary

This is an argument from what the Bible reports rather than from what it asserts, so it is weaker as a claim about the text's own doctrine, but it is strong as evidence for how the religion actually worked.

Rachel steals her father's household gods, the teraphim, and sits on them to hide them (Genesis 31). Michal, David's wife, puts the household teraphim in the bed to fake his presence (1 Samuel 19:13): the founder of the Davidic dynasty keeps idols at home and the narrator does not blink. Micah of Ephraim builds a shrine with a carved image, an ephod and teraphim, hires a Levite as priest, and the whole tribe of Dan takes the lot and sets it up as their tribal cult (Judges 17–18). Solomon builds high places for Chemosh, Molech and Ashtoreth (1 Kings 11).

The bronze serpent Moses made is worshipped in the Temple until Hezekiah breaks it (2 Kings 18:4). When Josiah purges the Temple he removes from it an Asherah, the vessels made for Baal, the horses dedicated to the sun, and the quarters of the cult prostitutes (2 Kings 23), all of which had been inside Yahweh's own house. And when Jeremiah scolds the Judean refugees in Egypt, the women answer that when they burned incense to the Queen of Heaven "we had plenty of food and prospered, and since we stopped we have lacked everything" (Jeremiah 44:17–18). They are not confused pagans. They are Judeans describing the religion they grew up in.

The inscriptions from Kuntillet Ajrud and Khirbet el-Qom that bless people "by Yahweh and his Asherah" are the archaeological side of the same picture, and I covered them in the earlier post.

10. Yahweh has an address

Gods in the ancient world came from somewhere, and so does Yahweh. The oldest poems say he "came from Sinai, and dawned from Seir; he shone forth from Mount Paran" (Deuteronomy 33:2), that he "went out from Seir, marched from the field of Edom" (Judges 5:4), and that "God came from Teman, the Holy One from Mount Paran" (Habakkuk 3:3). Seir, Edom, Teman and Paran are all south and east of Israel. That is why David outside the land is outside Yahweh's reach, and why Naaman needs the soil. A god with a home region is a god among gods.

The name of the people points the same way. It is Isra-el, not Isra-yah. Exodus 6:3 has God tell Moses that he appeared to Abraham, Isaac and Jacob "as El Shaddai, but by my name Yahweh I was not known to them", and in Genesis 14 Abraham meets Melchizedek, priest of El Elyon, and swears by "Yahweh, El Elyon". These look like the seams where two gods, the Canaanite high god El and the southern storm god Yahweh, were stitched together into one. That is a historical argument more than a textual one, which is why it is last, but it explains why the text above looks the way it does.

Where real monotheism finally appears

Genuine monotheism, the claim that no other god exists, does show up in the Bible. It is just concentrated in the latest layers. Second Isaiah, writing in the Babylonian exile, says "I am Yahweh, and there is no other; besides me there is no god" (Isaiah 45:5), "before me no god was formed, nor shall there be any after me" (43:10), and mocks the man who carves an idol from half a log and cooks dinner with the other half (44:9–20). Deuteronomy 4:35 and 4:39, in a chapter most scholars assign to the exilic edition of the book, say that "Yahweh is God; there is no other besides him."

Notice two things. First, these passages argue. They insist, they repeat, they ridicule. You do not campaign this hard for a view everyone already holds. Second, they sit beside Deuteronomy 32:8 and Psalm 82 in the same canon, and nobody removed the older material. The Bible is not a monotheistic book with a few polytheistic slips. It is the archive of a religion moving from a pantheon, through exclusive loyalty to one member of it, to the denial of the rest, with every stage still on the record.

So when someone tells you the Bible teaches that there is one God, the accurate answer is: a few late chapters teach that. The rest teaches that there is one God for you, that he is greater than the others, that he is jealous of them, and that you had better not be caught at their altars. That is monolatry, and the text says so on nearly every page where the question comes up.

Rationality Is the Exception: Humans Were Not Built to Think

2026-09-25T00:00:00+00:00

Aristotle called man the rational animal, and the Enlightenment built its institutions on that description: give people facts and a free press and they will reason their way to good decisions. It is a flattering picture, and it is wrong. Rationality is not what the human brain does by default. It is a rare, effortful, late-arriving trick that a few people manage some of the time, usually with heavy institutional scaffolding, and that most people never manage at all. It is incidental to what the brain was built for, the way a calculator app is incidental to a phone.

This is not a philosophical opinion. It is the consistent finding of a century of experimental psychology, and it is the plain reading of two decades of social media data. The evidence is worth laying out in one place.

What the Laboratory Found

The experiments below are classics because they replicate, and because the failures they show up are not failures of the stupid or the uneducated. They are failures of Harvard doctors, Stanford undergraduates and the general population alike.

The Wason selection task (Wason 1966). Four cards, one simple rule ("if a card has a vowel on one side, it has an even number on the other"), and the question which cards you must turn over to test it. Fewer than 10% of people get it right. Reframe the identical logic as a social rule ("if you are drinking beer, you must be over 18") and about 75% get it right (Griggs and Cox 1982; Cosmides 1989). The brain has no general logic module. It has a cheater-detection module, and logic problems that happen to look like cheating get solved.
The conjunction fallacy (Tversky and Kahneman 1983). Linda is 31, outspoken, and was a philosophy major concerned with social justice. Is she more likely to be a bank teller, or a bank teller who is active in the feminist movement? Roughly 85% of respondents picked the second, which cannot be more probable than the first, because it is a subset of it. A story that sounds right beats a probability that is right.
Base-rate neglect (Casscells, Schoenberger and Grayboys 1978). Sixty staff and students at Harvard Medical School were asked: a disease has a prevalence of 1 in 1,000, the test has a 5% false-positive rate, a patient tests positive, what is the chance they have the disease? The correct answer is about 2%. The most common answer was 95%. Only 18% got it right. These were the people who read your test results.
Anchoring, framing and availability (Tversky and Kahneman 1974, 1981). Spinning a rigged wheel of fortune before asking what percentage of UN members are African countries moved the answers. Describing the same public-health programme as "200 of 600 will be saved" or "400 of 600 will die" flipped the majority's choice. The judgement was not computed from the facts; it was computed from the packaging.
The Cognitive Reflection Test (Frederick 2005). A bat and a ball cost $1.10 together and the bat costs a dollar more than the ball; what does the ball cost? Most people, including most students at selective universities, say ten cents. The right answer is five, and it takes one line of arithmetic to check. People do not check. The intuitive answer arrives first and feels like knowledge.
Asch's conformity experiments (1951). Asked which of three lines matches a reference line, with confederates confidently giving the wrong answer, subjects went along with an obviously false answer on about a third of trials, and three quarters of them did so at least once. The lines were not ambiguous. The room was.
Telling more than we can know (Nisbett and Wilson 1977). Shoppers asked to pick the best of four identical pairs of stockings chose the rightmost pair four to one, and every one of them gave a reason about quality. Nobody mentioned position, and when it was suggested, they denied it. The reasons we give for our decisions are generated after the decision, by a part of the mind that was not present for it.
Choice blindness (Johansson et al. 2005; Hall et al. 2012). Subjects chose which of two faces was more attractive, were handed the other one by sleight of hand, and then explained fluently why they had chosen it. The same trick on a political survey had voters arguing for the opposite of the positions they had just ticked. If reasoning were driving belief, this could not happen.
Moral dumbfounding (Haidt 2001). Given a scenario of consensual, protected, secret, one-time sibling incest that harmed no one, people insist it is wrong, fail to produce a reason that survives scrutiny, and then say "I don't know why, it just is." The verdict comes first; reasoning is the press office.
When prophecy fails (Festinger, Riecken and Schachter 1956). A small group predicted that a flood would end the world on 21 December 1954. When the date passed, the members who had given up their jobs and possessions did not lose their faith. They became more committed and began proselytising. The disconfirmation was absorbed as confirmation.
The Dunning-Kruger effect (1999). Those with the least competence in logic, grammar or humour rated themselves well above average, because the skill needed to judge one's performance is the same skill that was missing.

Keith Stanovich, who spent a career measuring this, concluded that rational thinking is a distinct ability from intelligence, that IQ tests do not capture it, and that on tasks of rational thinking performance is poor across the whole intelligence range. Hugo Mercier and Dan Sperber went further in The Enigma of Reason (2017): reasoning did not evolve to find truth. It evolved to win arguments and to evaluate other people's, which is why we are excellent at spotting the flaw in an opponent's case and nearly blind to the flaw in our own. Confirmation bias is not a bug in the reasoning module. It is the reasoning module doing what it was selected for.

Why Mathematics and Science Are So Hard to Teach

Every child on earth learns to speak without a teacher. No child on earth learns algebra without one. That asymmetry is the whole story.

David Geary (1995) distinguished biologically primary abilities from biologically secondary ones. Primary abilities (language, face recognition, reading other people's intentions, intuitive physics, small-number sense) come for free through play, in every culture, without instruction. Secondary abilities (reading, arithmetic beyond a handful, algebra, formal logic, the scientific method) are cultural inventions that must be forced onto a brain with no circuitry for them, through years of effortful, motivated, externally imposed practice. Language is universal to every human society ever recorded. Writing was invented perhaps four times. Science was invented once.
Stanislas Dehaene showed what "no circuitry" means. Reading works by neuronal recycling: it hijacks a patch of visual cortex evolved for recognising objects. Arithmetic hijacks an approximate magnitude system in the parietal lobe that gives us exact quantities only up to about four and a blurry logarithmic sense after that. The Mundurucu of the Amazon, who have number words only up to five, do approximate arithmetic on large quantities as well as French adults and exact arithmetic not at all (Pica et al. 2004). Exact arithmetic is not something humans do. It is something a few cultures trained themselves to do against the grain.
Our built-in physics is Aristotelian. Michael McCloskey (1983) asked students to draw the path of a ball leaving a curved tube. A large fraction, including students who had passed university physics, drew it continuing to curve. Asked where a package dropped from a moving plane lands, they drew it falling straight down. The Force Concept Inventory later showed that a full semester of traditional physics lecturing moves these intuitions barely at all (Hake 1998): the average gain in conventional courses was about a quarter of what there was to gain.
Probability is the worst of all. When Marilyn vos Savant published the correct solution to the Monty Hall problem in 1990, she received around ten thousand letters telling her she was wrong, about a thousand of them from PhDs. Paul Erdős, one of the great mathematicians of the century, refused to accept it until shown a simulation.
Literature, by contrast, asks the brain to do what it does at the dinner table: track who wants what, who is lying, who is in love, who will betray whom. That is folk psychology, a biologically primary skill, running on the same circuitry as gossip. Every human culture has stories; Jerome Bruner called narrative the native mode of human thought. A literature class can be taught to a room of fifteen-year-olds because the fifteen-year-olds arrived with the hardware. A physics class has to override the hardware.

This is why people dislike mathematics and science, and the dislike is not a character flaw. Sian Beilock's group found that in people with high maths anxiety, merely anticipating a maths problem activates the same brain regions that register bodily pain (Lyons and Beilock 2012). Science is a collection of claims that contradict every intuition you have: the ground is moving at a thousand miles an hour and you cannot feel it, invisible animals make you sick, your ancestors were fish, the solid table is mostly empty space. Nobody's brain finds that comfortable. That science exists at all, that it appeared late, in one place, and nearly died several times before it took, is the strongest evidence that it is not a natural product of human cognition.

Why People Believe Things That Are Almost Certainly False

Take religion, because it is the largest and best-documented case.

Humanity has worshipped thousands of gods. The major living religions are mutually exclusive: the truth of one entails the falsity of the others. Any honest prior for "the one I happened to be raised in is the true one" starts somewhere near zero and does not improve, because the single best predictor of a person's religion is not any argument or evidence but the religion of their parents and the latitude and longitude of their birth. A born-again Baptist in Alabama would be a devout Hindu had he been born in Varanasi, and he would be equally certain. Beliefs that track geography rather than evidence were not produced by evidence.
Nobody performs this calculation, and that is itself the finding. The question "what are the odds that of all the religions, mine is the correct one?" is almost never asked by believers, not because it is answered but because it does not occur. Reasoning is not consulted.
Cognitive science explains why belief comes so easily. Humans have what Justin Barrett called a hyperactive agency detector: a rustle in the grass is treated as a predator, because a false alarm is cheap and a missed tiger is fatal. The same machinery sees faces in clouds, intent in weather, and a mind behind the universe (Guthrie 1993; Boyer 2001). Deborah Kelemen showed that children are intuitive theists, explaining rocks as pointy "so that animals can scratch on them" and rivers as existing "so that boats have something to float on". Teleology is the default; it has to be trained out, and in most people it never is.
Terror management theory (Greenberg, Solomon and Pyszczynski 1986): remind people of their own death and they cling harder to their worldview and punish outsiders more severely. Religion's central promise is that death is not the end. A belief that removes the worst fear a mind can have does not need to be true to be held.
Religion is also membership. Richard Sosis found that religious communes with costlier requirements survived longer than secular ones. A belief held as a badge of belonging is not held as a hypothesis, so evidence against it is not evidence against anything the believer cares about. It is an attack on the tribe.

None of this is special to religion. The same machinery produces astrology, homeopathy, conspiracy theories and nationalism. Religion is just the case with the longest record.

Hope Beats Facts

Tali Sharot's group found that people update their beliefs asymmetrically: told that a bad outcome is more likely than they thought, they barely move; told it is less likely, they move a lot (Sharot, Korn and Dolan 2011). The brain tracks good news and discards bad news, and this shows up in the neural signal, not just in the questionnaire. Around 80% of people show the bias.
Neil Weinstein (1980) found that students rated themselves as far less likely than their peers to be divorced, fired, or diagnosed with cancer, and far more likely to own a home and live past eighty. Everybody cannot be below average risk.
Patients with metastatic lung or colorectal cancer, receiving chemotherapy that every oncologist knew was palliative, were asked whether the treatment might cure them. 69% of the lung patients and 81% of the colorectal patients said yes (Weeks et al. 2012, New England Journal of Medicine). They had been told. They had heard something else.
Students asked how long their thesis would take estimated 34 days on average. It took 56. Asked for a worst-case estimate under the assumption that everything went wrong, they said 49 days, which was still less than reality (Buehler, Griffin and Ross 1994). The planning fallacy has not been cured in the thirty years since.
Americans spend about a hundred billion dollars a year on lottery tickets, against an expected return that every buyer could compute and none does.

There is an evolutionary reason hope wins. Alloy and Abramson (1979) found that mildly depressed people judge their own control over events more accurately than healthy people do. Accuracy about your prospects is depressing, and depression does not forage, court, or fight. Natural selection built optimists, and optimism means discarding facts that point the wrong way.

Facts Are Ammunition, Not Input

If people were reasoning from facts to conclusions, giving them more facts would move them. It does not. It arms them.

Biased assimilation (Lord, Ross and Lepper 1979). Supporters and opponents of the death penalty read the same two studies, one finding deterrence and one finding none. Each side judged the study supporting its view to be rigorous and the other to be flawed. After reading identical evidence, both sides were more confident than before. The same facts polarised.
Motivated numeracy (Kahan et al. 2013). Subjects were given a data table and asked whether it showed a skin cream working. The numerate got it right. The same numbers, relabelled as gun-control data, split them: highly numerate liberals and conservatives each got the answer right when it supported their side and wrong when it did not, and the more numerate they were, the larger the gap. Skill at reasoning was recruited to defend the tribe. The same team had already found that greater science literacy makes people more polarised on climate change, not less (Kahan et al. 2012).
Motivated scepticism (Taber and Lodge 2006). Given arguments on affirmative action and gun control, subjects spent longer reading and rebutting opposing arguments than agreeing ones, rated congenial arguments stronger, and, when allowed to choose, sought out their own side. The politically sophisticated did all of this more, not less.
Drew Westen (2006) put committed partisans in a scanner and showed them their own candidate contradicting himself. The regions associated with cold reasoning stayed quiet. The regions associated with emotion and conflict lit up, and when the subject found a way to excuse the contradiction, the reward circuitry fired. Rationalisation is literally rewarding.
Corrections do not reliably backfire, as was once feared (Wood and Porter 2019), but they do very little: they shift stated beliefs slightly, the shift decays within weeks, and it almost never changes the vote or the behaviour that the belief was defending.

Jonathan Swift wrote in 1721 that reasoning will never make a man correct an ill opinion which by reasoning he never acquired. Ziva Kunda formalised it in 1990 as motivated reasoning: we ask "must I believe this?" of unwelcome evidence and "can I believe this?" of welcome evidence, and the answers are almost always no and yes. A fact that supports the feeling is kept. A fact that contradicts it is not weighed; it is dismissed, reinterpreted, or forgotten. Jonathan Haidt's summary is the best one: the emotional dog wags the rational tail.

Facebook's news feed launched in 2006, and Twitter the same year. Since then, several billion people have had every sentence they publish scored for engagement, in real time, by systems that learn what humans respond to and serve them more of it. Nobody designed those systems to reward outrage. They were designed to reward whatever humans reward, and they measured the answer with a precision no psychology lab could approach. The results are in.

Falsehood outruns truth (Vosoughi, Roy and Aral 2018, Science). Across 126,000 stories shared by three million people on Twitter between 2006 and 2017, false news was 70% more likely to be retweeted than true news, reached 1,500 people about six times faster, and spread deeper and more broadly. The effect was strongest for political news. Bots spread true and false stories equally; the asymmetry was entirely human. False stories were more novel and evoked more surprise and disgust.
Moral emotion is the fuel (Brady et al. 2017, PNAS). Across 563,000 tweets on gun control, climate change and same-sex marriage, each moral-emotional word ("attack", "shame", "greed", "evil") increased the expected retweet rate by about 20%. The spread was almost entirely within the ideological group, not across it.
Attacking the other tribe works best of all (Rathje, Van Bavel and van der Linden 2021, PNAS). Across 2.7 million posts by news outlets and politicians, each additional word referring to the political out-group increased the odds of a share by 67%. Out-group language was the single strongest predictor of engagement, far stronger than praising one's own side, and the most shared posts were the angriest.
Exposure to the other side makes it worse (Bail et al. 2018, PNAS). Paying Twitter users to follow a bot that retweeted the opposing side for a month made Republicans substantially more conservative and Democrats slightly more liberal. The Enlightenment prescription of exposure to diverse views was tested and it backfired.
Facebook's own researchers agreed (the Haugen documents, 2021). After the 2018 shift to "meaningful social interactions", the angry reaction was weighted five times a like, and internal research reported that publishers and parties had shifted to outrage because it was what the ranking rewarded. Facebook's own data scientists warned that the change was making the platform angrier. The metric said it was working.
Nobody is even checking (Pennycook and Rand 2019, 2021). People who share false headlines mostly do not believe them; asked directly, they rate the headlines as inaccurate. They shared them because sharing is a social act, not an epistemic one, and the question of truth simply did not arise. A brief prompt to think about accuracy improved the quality of sharing a little, which shows how far from the default it is.

The algorithm did not make people tribal and emotional. It measured that they were, at a scale of billions, and then gave them what they measurably preferred. If humans were driven by facts, the sober, sourced, fact-checked post would win the engagement auction. It has been losing that auction every day for twenty years. The optimisation ran on us, and it found what the psychologists found: emotion first, tribe second, facts as decoration.

What Follows

Rationality is real. Science works, courts sometimes reach true verdicts, and the occasional person changes their mind on evidence. But every one of those cases runs on scaffolding: peer review, adversarial procedure, double-entry bookkeeping, controlled experiments, the norm that a claim is worthless until someone tried to break it. Strip the scaffolding and the reasoning collapses back to what the laboratory shows. Rationality is a fire that a few institutions keep lit with continuous effort. It is not the human condition. It is an incidental achievement of a small minority of humans, sustained by structures that most people never enter and do not miss.

The mistake of the last three centuries was to take that minority as the description of the species and to design public life for it. Give the public the facts, we said, and they will reason. They did not, and the last twenty years measured exactly how much they did not. Institutions that only work if the crowd reasons will fail, because the crowd does not. Institutions that work anyway, that make the rational move the low-effort move and that do not depend on anyone being persuaded, are the only kind worth building.

And the reader is not exempt. Neither is the writer. That is the last thing the evidence says: the person most sure they are the rational exception is running on the same hardware as everyone else.

There Is No Such Thing as the "Appearance of Design"

2026-09-24T00:00:00+00:00

Richard Dawkins opens The Blind Watchmaker with a definition of his entire field: "Biology is the study of complicated things that give the appearance of having been designed for a purpose." A few pages later he says that the products of natural selection "overwhelmingly impress us with the illusion of design and planning." In Climbing Mount Improbable he went further and coined a word for it. Living things are not designed, he says, they are designoid: objects that look designed but are not, as opposed to Mount Rushmore, which is. And in The God Delusion natural selection "shatters the illusion of design within the domain of biology." Francis Crick put the same idea even more bluntly in What Mad Pursuit: "Biologists must constantly keep in mind that what they see was not designed, but rather evolved."

I want to object to the phrase. Let me be blunt about where I am coming from, since this is a topic where sentences get lifted out of context: this is an argument against a designer, not for one. Natural selection is exactly what Dawkins says it is, blind, unplanned, and without foresight. But the words "appearance of design" concede something to the creationist that never needed conceding, and they leave Dawkins holding a distinction he cannot draw. Organisms do not have the appearance of design. They have a design. It was produced by a real designer, natural selection, working under unusually severe constraints. And once you see that, the argument from design does not need to be refuted. It simply has nothing left to point at.

Apparent Design Versus Real Design

Dawkins's vocabulary commits him to two categories: things that are really designed, like Mount Rushmore, and things that are merely designoid, like a face in a hillside or a bird's wing. The trouble is that he owes us a way of telling them apart, and he has none that survives inspection.

It cannot be the artifact. A wing is fitted to flight more tightly than any aircraft; an eye is fitted to seeing better than any camera. If you judged by the object alone, the organism would win the design contest against the artifact every time. Dawkins knows this, which is why he needs the word "appearance" in the first place: the object is indistinguishable from a designed one, so the difference must lie elsewhere.

It cannot be the process either, as I will argue below, because the process by which minds design things turns out to be the same process that selection runs.

That leaves only one candidate: a mind was present. "Real design" means "a mind did it," and "apparent design" means "it looks the same but no mind did it." But that is not a discovery about biology. It is a stipulation about a word, and a strange one for a materialist to make. It puts the whole weight of the word "design" on a fact about the history of the object that leaves no trace in the object and no trace in the process. Dawkins would never accept that a computer merely has the appearance of arithmetic because nobody is inside it doing sums. Yet that is exactly the move he makes here.

Under the framework I am proposing there is no such problem, because there is no such distinction. Design is what a certain kind of process produces: generate variants, test them against a criterion, keep what passes, iterate, and never let the system stop working along the way. Anything that has been through that process has a design. Human engineers run it with one set of constraints, natural selection runs it with another, and they are both real design, in exactly the same sense, differing in their constraints and not in their kind.

The Objection: Design Needs a Designer

The obvious reply is that design does involve intent, by definition, and evolution has none. A designer wants something; selection wants nothing. That is a real difference and I do not want to hide it. But it is a difference in the motive of the process, not in the process or its result, and we can check that claim in the one field where we get to watch design happen from the inside: software.

Agile Development Is Evolution With a Coding Standard

Ask a working software engineer how large systems actually get built, and you will not hear about a blueprint. You will hear about agile development, which is a set of constraints adopted on purpose by an entire industry. The central constraint is this: never make a large change. Work in small increments. Every increment must leave the system working and shippable. There is no "under construction" state. Do not write the grand design up front; let the design emerge from the accumulation of increments, each one tested against a customer who does not know what they want until they see it.

Now list the properties of natural selection. Never make a large change, because a large mutation is almost always lethal. Every intermediate organism must survive and reproduce; there is no "under construction" state. There is no blueprint; the form emerges from the accumulation of small changes, each one tested against an environment that states no requirements and simply accepts or rejects what is delivered.

These are not analogous processes. They are the same process, described twice. Agile even reproduces evolution's most famous flaw. Because you may never stop and redesign, you accumulate what programmers call technical debt: the ugly workaround that was locally the smallest change and can never be removed because everything now depends on it.

Biology is full of technical debt. The recurrent laryngeal nerve loops from the brain down around the aorta and back up to the larynx, several meters of detour in a giraffe, because in a fish it ran straight and no intermediate could be allowed to break. The vertebrate retina is wired backwards. The human spine is a horizontal design pressed into vertical service. Any engineer who has maintained a ten-year-old codebase recognizes all of it at once. François Jacob said in 1977 that evolution works like a tinkerer rather than an engineer. He was right about the tinkering and wrong about the contrast: the tinkerer is what an engineer looks like under the constraint that nothing may ever be broken.

Would anyone say that a team practicing agile development does not design its software? That the finished product merely has the "appearance of design," because no one drew it in advance? Of course not. We would say it was designed incrementally, under a constraint, by a process of variation and selection. And that is exactly what we should say about the eye.

Linus Torvalds Already Said This

The largest piece of software on earth, which runs most of the internet, every Android phone, and nearly every supercomputer, was built this way, and the man who built it has been unusually honest about it. In November 2001, on the Linux kernel mailing list, in a thread that started as an argument about coding style, Linus Torvalds wrote:

Let's just be honest, and admit that it wasn't designed.

Software evolves. It isn't designed. The only question is how strictly you control the evolution.

It grew. It grew with a lot of mutations - and because the mutations were less than random, they were faster.

Don't EVER make the mistake that you can design something better than what you get from ruthless massively parallel trial-and-error with a feedback cycle.

And two years later, on a Usenet group: "You're giving the human mind too much credit, and giving too little credit to the selection process."

Read those again. Torvalds has minds available. He has thousands of intelligent contributors, each one choosing their changes deliberately. He concedes that this makes the mutations "less than random," which speeds the process up. And he still says the result was not designed in the sense Dawkins means, that it evolved, and that the design came from the feedback cycle and not from anyone's intelligence. Torvalds and Dawkins agree completely on the facts. They just draw the line in opposite places. Dawkins says: not made by a mind, therefore not design, merely designoid. Torvalds says: this is how design is actually done, and the mind is the least important part.

Torvalds is right, and he is right for a reason Dawkins should appreciate. If you reserve the word "design" for what minds do, you have to say what the minds are doing that selection is not. And when you look, you find that the minds are running selection. An engineer generates candidates, tests them against a criterion, keeps the survivors, and iterates. Nobody derives a bridge from first principles; they copy last year's bridge and modify it. Human design is descent with modification plus a selection filter called testing.

What Intent Actually Buys

I said I would not hide the difference, so here it is. A mind can run the selection loop over models instead of over organisms. An engineer discards a bad bridge on paper; evolution discards it as a corpse. Karl Popper's phrase for this was that we let our hypotheses die in our stead. That is a genuine advantage, and it is why an engineer takes a decade to do what selection takes a hundred million years to do. The mind supplies a better mutation operator: variants are proposed with some foresight, and many are killed before they are ever built.

But notice what this concession is. It is a statement about the constraints the process runs under, not about whether it is design. Human design is selection with cheap tests and foresighted proposals. Natural selection is design with expensive tests and blind proposals. A team doing agile with a good test suite is somewhere in between. These are different points on one scale, and every point on it is real design. There is no place on the scale where design turns into the appearance of design, because nothing about the process changes except its budget.

Why the Distinction Matters

The argument from design has one premise that does all the work: design implies a designer. Dawkins accepts the premise and denies that there is any design, which forces him into "apparent," "illusion," and "designoid," and then into a distinction between apparent and real that he cannot cash out. That is a weak position, and the creationist can feel it. If organisms really are indistinguishable from designed things, in artifact and in process, then calling them designoid sounds like exactly what it is: a refusal to use the obvious word.

The stronger position denies the premise instead. Design does not imply a designer. Design implies a design process, and a design process needs a source of variation, a selection criterion, and time. Minds are one such process. Natural selection is another. Both are real, both produce real designs, and the second one is the one that produced us. Once that is said, there is no gap for a God to fill. The creationist's strongest observation, that living things are exquisitely designed, is granted in full, and it leads nowhere, because the designer it points at is a blind process that we can watch at work in any software company.

Crick's sentence should therefore be reversed. Biologists must constantly keep in mind that what they see was not planned, but rather designed, by the same method Linus used, with a worse mutation operator and a much larger budget.

Conclusion

Dawkins needs two kinds of design, real and apparent, and cannot say what separates them, because nothing does. I need one kind, with different constraints, and the organisms come out designed and the God comes out unemployed. There is nothing designoid about an eye. It has a design, produced by a designer that is real, blind, and slow, and that anyone who has ever shipped software in small working increments already knows from the inside. Selection is not something that imitates design. Selection is how design gets done.

The Fax Machine Should Have Died Decades Ago, and Why It Has Not Is the Interesting Part

2026-09-19T00:00:00+00:00

Somewhere in your country right now, a hospital is faxing a patient's records to another hospital. A court is accepting a document by fax that it would not accept by email. A government office has a fax number printed on its letterhead and a scanner on the desk next to the machine, so that the paper that arrives can be turned back into the image it was before it was sent.

The fax machine is the worst way to move a document that has been invented since the courier, and it is still here. Both halves of that sentence deserve attention.

Why It Should Go

Start with what a fax actually does. It takes a document that almost certainly began life as a file, prints it, scans the print at low resolution in black and white, encodes the scan as audio, plays the audio down a telephone line, and prints the result at the other end, where somebody scans it again to get a file. Every step loses information. The result is a degraded picture of text that no computer can read without guessing, sent over a channel that costs money per minute and fails if the line is busy.

Everything a fax does, email with an attachment does better on every axis: fidelity, cost, speed, searchability, storage, delivery confirmation, and the ability to reach more than one recipient. Where the concern is security, a fax is worse still. It is unencrypted, it goes to whichever machine holds that number today, and it prints out onto a tray in a corridor where anyone may pick it up.

There is no technical argument for the fax. There has not been one for thirty years.

Why It Stays

The persistence is the interesting question, and the answer is not that people are stupid. It is that the fax sits at the intersection of three forces that individually look reasonable.

Legal habit. In many jurisdictions a faxed signature acquired legal standing decades ago, through case law and regulation written when the fax was the newest thing. Email signatures arrived later and their standing was established piecemeal, unevenly and with caveats. So a lawyer who wants certainty reaches for the channel with the longest settled record, which is the worst channel available. The law does not prefer the fax. It has simply had longer to stop worrying about it.

Institutional risk aversion. Nobody in a hospital or a ministry has ever been blamed for sending a fax. Plenty of people have been blamed for sending an email to the wrong address or losing a laptop. The fax is a channel with no recent scandals attached, which in a bureaucracy is worth more than any technical merit. The person who proposes replacing it takes on a risk; the person who leaves it alone takes on none. I have written about how organisations select for exactly this reluctance, and the fax is a small monument to it.

The network effect running backwards. A fax is only useful if the other side has one. As long as the courts, the insurers and the pharmacies keep theirs, the hospital must keep its own, and as long as the hospital keeps its own, the others see no reason to drop theirs. Each participant is behaving rationally given the others, and the collective result is that a technology nobody would adopt today cannot be abandoned by anyone.

The Regulatory Layer

There is a fourth force that deserves its own paragraph, because it is the one that actively keeps the machine alive rather than merely failing to kill it.

In several countries, health and legal regulations name the fax specifically as an approved channel for sensitive documents and are silent about, or hostile to, email. The regulations were written by people who understood the fax and did not understand encryption, and they have not been revisited because revisiting them would require someone to take responsibility for the new rule. So a machine with no security is mandated on security grounds, and a channel with real security is forbidden on the same grounds, and everyone involved knows this is absurd and nobody is positioned to change it.

This is the general pattern of which the fax is the comic instance. A rule written for one era outlives the era, the people who could change it have no incentive to, and the cost is paid diffusely by everyone who must comply. I have made the same argument about legislation as a form, and about the complaints every new technology attracts. The fax is what those two arguments look like when they meet in a hospital corridor.

What Would Kill It

Not a better technology. The better technology has existed for a generation and has not killed it.

What would kill it is a decision, taken at the level of the rules rather than the level of the machine: that a signed, encrypted document sent electronically has at least the legal standing of a smudged page arriving by telephone, and that regulations naming specific technologies expire unless renewed. Make those two changes and the fax machines will be in the skip within a year, because nobody actually wants them. They are kept by rules, not by people.

Until then the machine hums on, printing degraded copies of documents that were already files, in institutions that know better, because the cost of knowing better is borne by whoever moves first.

Philosophy and the Scientific Revolution: The Failure of Ambiguity

2026-09-14T00:00:00+00:00

The usual story about the Scientific Revolution is that people started looking at the world instead of reading Aristotle about it. That is part of it, but observation alone does not explain the break. Medieval scholars observed plenty. Astrologers kept meticulous tables, physicians watched patients for a lifetime, and alchemists ran more experiments than most modern chemistry undergraduates. What changed in the seventeenth century was not the amount of looking. It was a decision about what you are allowed to build a theory out of.

The decision was this: start from a small number of primitives that mean exactly one thing, and construct everything else from them. Philosophy never made that decision, and that is why it is where it is.

What Newton Refused to Do

Before Newton, the explanation of why things fall was that heavy bodies have a natural tendency toward the centre. The explanation of why opium puts you to sleep, in Molière's famous parody, was that it has a dormitive virtue. Fire rises because its nature is to rise. These explanations were satisfying, and they were worthless, and the two facts are related. A word like "tendency" or "virtue" or "nature" can be attached to any phenomenon after the fact, so it explains everything and predicts nothing.

Newton did not write a treatise on the nature of motion. He picked a handful of bare terms, point mass, distance, time interval, force as a vector, and he gave each one a definition tight enough that two people using it could not disagree about what it referred to. Then he invented the calculus needed to say how those terms relate, and he derived the orbits of the planets from three laws that fit on one page. The primitives were poor in meaning, deliberately so. All the richness came out of the construction.

Chemistry did the same thing a century later, and it could only do it by throwing words away. Phlogiston was a real theory held by serious people. It had a name, a literature and a set of qualitative principles, and none of that saved it once weight could be measured to the milligram. The field became a science at the moment it agreed that its primitives would be discrete elements with fixed masses and nothing else.

The twentieth century repeated the move in the most abstract territory imaginable. Shannon founded information theory by refusing to say anything about meaning at all. Information was the reduction of uncertainty, the unit was the bit, and the question of what a message was about was ruled out of scope. Turing did the same for computation: an abstract machine with a tape, a head and a finite table of state transitions, and every algorithm anyone would ever write turned out to be expressible in it. Both men were asked, in effect, to define something deeply human, and both answered by defining something almost insultingly simple and then showing it was enough.

The pattern is the same every time. Take the complex phenomenon, decompose it into primitives that are simple, unambiguous and checkable, and rebuild the complexity from them. The direction is always bottom up.

Philosophy Went the Other Way

Metaphysics, epistemology and most of what gets called continental philosophy did not adopt this discipline, and they did not adopt it for a specific reason: they took natural-language words as their starting point. Mind, free will, substance, knowledge, justice, goodness, being. These are the primitives of philosophy, and not one of them is primitive.

Natural language evolved to coordinate a band of primates. It is a noisy heuristic for getting other people to do things, and its words are compressed bundles of use cases that happened to travel together. "Knowledge" covers knowing a fact, knowing how to ride a bicycle, knowing a person and knowing that something feels wrong, and there is no reason to believe those share an underlying structure just because English gave them one verb. To treat the word as a unit of reality is to assume that the accidents of a language's history carve the world at its joints.

Three things go wrong when a discipline builds on such words.

The first is reification. Because there is a noun, there must be a thing. Because people say "consciousness", there must be a single coherent entity that either is or is not present in a system, and the debate becomes about where it is rather than about whether the noun names one thing at all.

The second is circularity. When your terms are fuzzy, the only move available is to propose a definition, and the only counter-move is to propose another one. Two and a half millennia of argument about what justice is have this shape. It looks like inquiry, but no mechanism is ever uncovered, because the argument is over which words to prefer, and preferences do not converge.

The third is mistaking confusion for depth. When ambiguous terms produce a paradox, which they reliably do, the philosopher announces a profound mystery of the universe. Zeno's arrows, the ship of Theseus, the problem of the heap. In every case the "mystery" evaporates the moment someone supplies a primitive precise enough to compute with, and in every case the philosophical literature had treated the vagueness of the original words as a feature of the world rather than of the words.

Wittgenstein saw much of this. Philosophical problems arise, he said, when language goes on holiday, when words are lifted out of the practices that gave them a use and treated as metaphysical absolutes. That is a correct diagnosis. But the linguistic turn stopped at the diagnosis. Having noticed that the words were the problem, philosophy did not do the scientific thing and replace them with better tools. It kept the words and declared the disease incurable. I have written about that specific surrender elsewhere.

Every Success Left the Building

There is a clean historical test of this argument. Look at what happened to the parts of philosophy that did adopt bottom-up construction.

Natural philosophy became physics as soon as it swapped qualitative discourse about natures and tendencies for measurable quantities and laws. Nobody now calls Newton a philosopher, although his book has the word in its title. The philosophy of the organism, the long argument about what distinguishes living matter, became molecular biology as soon as heredity was located in a discrete chemical structure rather than a vital principle. The vitalists were not defeated in debate. They were made irrelevant by a molecule. Logic, to the extent it became formal, moved into mathematics and computer science, and the parts that stayed behind are the parts that resisted formalisation.

What remains in the philosophy department, then, is not a random sample of hard questions. It is a selected sample: the residue of questions that refused to be rebuilt from primitives, or were protected from it. This is worth stating plainly, because it inverts the usual self-image. Philosophy did not keep the deepest problems. It kept the ones whose vocabulary was never fixed, and it kept them precisely by never fixing the vocabulary. The ambiguity is not an obstacle the discipline has been fighting. It is the mechanism by which the discipline's problems have been preserved.

The Case of Meaning Itself

The clearest demonstration is the one closest to home for philosophy, which is the theory of meaning.

Philosophers have argued about what a concept is for as long as there have been philosophers. Plato put meanings in a realm of forms. Frege and Russell dissected sense and reference. The ordinary-language school catalogued how words are actually used. The literature is enormous and it is very hard to name one thing it settled.

Then, over about a decade, computer scientists built a working theory of meaning, and they did it by ignoring the entire debate. They did not ask what a word's essence is. They asked which words occur near it, counted, and mapped each word to a point in a space of a few hundred or a few thousand dimensions such that words with similar contexts land near each other. The primitives were vectors, dot products and a probability distribution over a corpus. Nothing in the construction refers to meaning at all.

And meaning fell out. Similarity became distance. Categories became regions with graded membership, exactly the fuzzy family-resemblance structure philosophers had said was unformalisable. Analogy became subtraction: king minus man plus woman lands next to queen, and nobody put it there. I have written about why those vectors need so many dimensions and I will not repeat it, but the point for this argument is the method. Meaning became measurable the moment somebody agreed to build it from primitives that were not themselves words.

This is the pattern from the first section, replayed on the subject philosophy considered most its own. The result was not a contribution to the philosophical debate. It was the end of the debate's relevance, in the way that chemistry was the end of alchemy's relevance, and for the same reason.

What Would Have to Change

The lesson of the Scientific Revolution is not that observation beats speculation. It is that understanding is built from the bottom up, and that the bottom has to be made of things that mean one thing each. Inherited words do not qualify. They are what a theory has to explain, not what it is allowed to explain with.

I have argued before that philosophy as practised fails because it never fixes its terms. This post is the historical case for the same conclusion. For philosophy to become something other than a permanent scholastic backwater, it would have to do three things that the sciences did centuries ago. Stop treating vernacular nouns as windows onto ontology. Break every loaded concept down into primitives precise enough that a disagreement about them can be settled. Prefer a formal model that can be wrong to a dialectic that cannot.

I do not expect this to happen, because a philosophy that did it would stop being called philosophy. That has been the fate of every branch that tried. But it is worth being clear about why the questions that remain have remained. They are not too deep for the method. They have simply never been submitted to it.

A Cellular Network Nobody Owns Is Technically Easy and Politically Forbidden

2026-09-12T00:00:00+00:00

Every phone in your pocket is a radio. Every one of them can talk to every other one within a few hundred metres. There are billions of them, densely packed exactly where people are. And yet to send a message to the person standing next to you, the message travels to a tower owned by a corporation, through that corporation's core network, and back to the tower, and you pay the corporation for the privilege.

I want to sketch how a network without the corporation would work, and then say why it does not exist, because the reason is not the one people assume.

The Technical Sketch

A decentralised cellular network has three layers, and none of them requires an invention.

The radio layer. Phones talk directly to nearby phones, and to small fixed cells that people put on their windowsills and rooftops, over spectrum that anyone may use without a licence. That spectrum exists; it is what wireless networking and Bluetooth already run on, and more of it is available at higher frequencies. Range per hop is short, but hops are cheap and nodes are everywhere. This is a mesh, and meshes have been built many times, from community networks spanning whole cities to emergency systems that work when the towers are down.

The routing layer. A message finds its way across many hops without any central directory, the way the internet's routing was originally designed to work before it re-centralised. Nodes learn who their neighbours are, exchange what they know, and forward packets toward the destination. Where the mesh is thin, a node with a wired connection bridges to the ordinary internet. The protocols for this are decades old.

The identity layer. Nobody assigns you a number. Your identity is a cryptographic key you generate yourself, and people reach you by it. Messages are encrypted end to end by default, because there is no operator to trust and therefore no reason to leave anything in the clear. This, too, is built and running in several messaging systems today.

Put the three together and you have telephony, messaging and data among everyone in range of everyone, at no marginal cost, with no operator, no subscriber database, no central point at which to listen or to switch it off.

Who Pays

The obvious objection is that networks cost money. Towers, backhaul, engineers. Who pays for a network with no owner?

Mostly, nobody needs to. The radios are already bought; they are in the phones. The fixed cells are cheap consumer hardware that people buy for the same reason they buy a wireless router. The expensive part of a conventional network, the core and the backhaul, largely disappears when traffic stays local, and most traffic is local.

Where a cost remains, for the bridges to the wider internet and for the nodes that carry more than their share, the settlement problem is exactly the one I described in how cryptocurrency lets a decentralised organisation handle money. A node that forwards your packets can be paid a fraction of a cent for doing so, automatically, by the protocol, with no company in the middle. The mechanism for paying strangers small amounts without an intermediary is the piece that was missing for decades and is missing no longer.

Why It Does Not Exist

So the technology is available, the hardware is deployed, and the payment problem has a solution. Why do you pay a carrier?

Because the spectrum that works well is licensed, and the licences were sold to the carriers. The frequencies that travel far and penetrate walls, the ones that make a network good rather than merely possible, are allocated by the state to a handful of companies for enormous sums. Using them without a licence is a crime. The unlicensed bands are the leftovers: short range, crowded, and deliberately kept that way.

This is presented as physics. Spectrum is scarce, interference is real, somebody has to coordinate. All true, and all beside the point. Coordination does not require ownership, and it certainly does not require ownership by three companies. The mesh coordinates itself; that is what the protocols are for. The licensing regime was designed for a world of a few high-power transmitters and has been kept, long after that world ended, because it produces revenue for the state and monopoly rents for the licensees.

Add the second obstacle: the regulator's need to be able to listen. A network with no operator has no one to serve a warrant on, no database to subpoena, no switch to throw. Every state finds that intolerable, and every state has arranged that phones sold within its borders will not do what their radios could do.

The network nobody owns is forbidden not because it would fail but because it would work, and its working would remove a revenue stream, a rent, and a point of control simultaneously. That is three powerful constituencies against and a diffuse public for, which is the configuration in which the public loses every time.

What Would Change It

Not a better protocol. The protocols are fine.

A decision, taken by the public rather than by the licensees and the regulator, that the spectrum is a commons and that the phones people buy will be permitted to use it. That is a political act, not an engineering one, and it is the kind of act a representative system finds nearly impossible, because the interested parties are concentrated, well funded and in the room, and the beneficiaries are everyone, which in that system means nobody.

I write this not because I expect the network to be built soon, but because it is useful to know precisely where the obstacle sits. It is not in the radio. It is in who was allowed to own the air.

Word Embeddings and the Arbitrariness of the Sign

2026-09-11T00:00:00+00:00

A friend building on an AI agent platform asked me a question that I have since heard from several engineers. The platform let him choose between two OpenAI embedding models, one producing vectors of 3,072 numbers and one producing 1,536. Why, he wanted to know, does a word need thousands of numbers? The Hebrew word for bird, ציפור, has five letters. Would a five-dimensional space not do? And if you need room for the inflections, birds and my bird and so on, maybe a few more?

The question is a good one, because the right answer to it is the first lesson of linguistics, and most of us never had that lesson.

Meaning Is Not in the Letters

The intuition behind the question is that a word's representation should somehow be derived from the word itself: its spelling, its length, its root and its morphology. That intuition is completely wrong, and the reason it is wrong is over a century old.

Consider the word תפוח, apple. Is there any connection between those letters, or the sounds they stand for, and the crisp red or green fruit? None. Whoever first used the word did not extract its consonants from the fruit's skin. When Eliezer Ben-Yehuda coined עגבנייה for tomato, he picked a root and a pattern, and the pick was an invention. The only place in the universe where the relationship between עגבנייה and the red vegetable exists is inside the heads of Hebrew speakers who share the convention. Had the convention gone the other way, with apples called tomatoes and tomatoes called apples, not a single property of either fruit would be different.

This is what Ferdinand de Saussure called the arbitrariness of the sign. A word is a signifier, an arbitrary label, attached by social agreement to a signified, a concept. The label carries no information about the concept. That is not a defect of language; it is the essence of it. Onomatopoeia and a handful of sound-symbolic words are the exceptions that make the rule visible.

Once you see this, the letter-counting question dissolves. An embedding is not a representation of the word ציפור. It is a representation of the concept that Hebrew speakers happen to label ציפור and English speakers happen to label bird. The spelling never enters into it, and could not, because the spelling contains nothing to enter.

Why So Many Dimensions

The second question is better and survives the first answer. A language has perhaps 50,000 to 150,000 words in active use, inflections included. Even a 3,072-dimensional space restricted to zeros and ones has 2^3072 distinct points, a number that dwarfs every word in every language that ever existed. If you only wanted to tell words apart, 17 bits would do, and 120 dimensions would be absurd overkill. So what are the other 2,950 for?

The answer is that telling words apart is not the goal. If it were, an integer ID would be enough, and that is exactly what a dictionary index is. The goal of an embedding is to place concepts in a space where distance means something: where bird sits close to eagle and far from queen, where king minus man plus woman lands near queen. The vector is not an identifier. It is a position, and the position is chosen so that geometric nearness matches semantic nearness.

That requirement is what eats the dimensions. Think of a large apartment into which dozens of suitcases arrive from different countries. You want the ones from France near each other, but also the ones with clothes near each other, and the ones belonging to the same family, and the ones going on to the same destination. Each criterion is an independent axis along which things must be close or far, and if you have only the floor to work with, the criteria fight. Two dimensions let you honour one or two of them. Satisfying thousands of independent similarity relations at once, without any two of them colliding, takes thousands of degrees of freedom. High dimensionality is not there to store the words. It is there to give the relations between the words room.

Nobody places the words by hand. The positions are learned from a very large corpus, by observing which words occur in the company of which others, and adjusting until words that keep similar company end up nearby.

The Numbers Mean Nothing on Their Own

My friend then opened a vector and looked at it: 1,536 floating point numbers, all between roughly minus 0.1 and 0.1. What does the fifteenth one mean? Is it "has wings"? Is it "is an animal"?

It is neither, and not because the answer is secret. No single coordinate has a meaning that a human can name. The meaning of the vector lies entirely in its relation to every other vector: the angle it makes with each, the distance to each. Rotate the whole space and every number changes while every meaning stays exactly the same, because the angles and distances are preserved. Meaning in an embedding is purely differential.

That, too, is Saussure. His claim was that in language there are only differences, without positive terms: a word means what it means by virtue of the words it is not, by its place in the network of contrasts. The linguists were describing, in prose and from the armchair, a structure that the machine learning people rediscovered by gradient descent eighty years later. It is one of the more satisfying convergences I know of between a humanistic theory and an engineering result.

What This Buys You in Practice

None of this is philosophy for its own sake. Embeddings are a technology in their own right, older than and independent of the large language models that now use them as an input layer. You can take a corpus, turn each passage into a vector, store the vectors in a database that supports nearest-neighbour search, and ask it for the passages closest to "royalty" or "Shakespeare". No generative model is involved. This is what retrieval-augmented generation does with a company's documents: index them as vectors, fetch the nearest ones to the question, and hand those to the model so that it answers from the documents rather than from its imagination.

It also explains why serious systems keep the old full-text search alongside the new vector search. The two fail in opposite ways. Lexical search over tokens finds exact part numbers, identifiers and code fragments, and fails on paraphrase and synonyms. Semantic search over vectors finds paraphrase and synonyms, and is hopeless at exact identifiers, precisely because it never looked at the letters. So production systems run both and merge the ranked results, typically with reciprocal rank fusion. The engineering decision to combine them is a direct consequence of the linguistic point: one index knows the signifier and the other knows the signified, and you need both.

The Lesson

Engineers who never took a linguistics course arrive at embeddings with the intuition that meaning lives in the string. It does not. It lives in the network of relations between concepts, and the string is an arbitrary handle that a community agreed to attach to one node of that network. Once that is understood, every design choice in an embedding model, the size of the vectors, the meaninglessness of individual coordinates, the need for a second lexical index, follows naturally. Lesson one of linguistics turns out to be lesson one of vector search.

Mark Veltzer's personal site

The Primacy of the Scientific Method: Why Empirical Rigor Is the Only Path to Knowledge

The Mirage of Intuition

The Limits of Pure Logic

The Engine: Testing and Measurement

Replication and Falsifiability

Independent replication

Falsifiability

The Only Self-Correcting Method

Fighting the Wrong Battle: How the Academic Left Chose the Wrong War

The Pass That Religion Received

From Truth to Moral Standing

The Real Adversary

Why the Echo Chamber Stagnates

Back to the Real War

Where This Started

Beauty, Truth and the Size of Working Memory

The Budget

Proofs

Causes

Music

Conclusion

An Ode to Individuality

The Record

Why It Is Individuals

The Amendment

Where This Leaves the Groups

Emotion: The Enemy of Reason

The Ancient Hijack

The Tyranny of the Immediate

The Fog of Affect

The Psychology of Faith

The Hard Discipline of Reason

IQ and Religious Belief: What the Correlation Actually Says

The Bottom Line

The Numbers

How to Read the Numbers

Where the Effect Is Strongest

Proposed Explanations

Caveats

Religions Did Not Evolve, They Merely Survived

What Evolution Actually Requires

Religions Suppress Variation on Purpose

Look at the Product

Survival Is Not Design

Theology Belongs with Astrology, Outside the University

Premises, Logic, Conclusions

The Parallel, Attribute by Attribute

Belief Is Not Knowledge

Religion Can Be Studied, Theology Cannot Be Done

Why People Believe the Bible: The Social, Psychological and Cognitive Machinery

The Social Layer

The Psychological Layer

The Cognitive Layer

What the Three Layers Add Up To

Why Believers Wear Symbols and Non-Believers Do Not

The Minority Explanation Fails

Symbols as Costly Signals

Fragile Beliefs Need External Support

Symbols as Mutual Reality Construction

What the Asymmetry Tells You

Is Religion an Ideology?

The Descriptive Sense

The Pejorative Sense

What Religion Adds

Why the Label Is Contested

Phones Moved Toward Freedom One Concession at a Time, and So Does Power

Stage One: The Carrier Owns the Phone

Stage Two: A Different Master

Stage Three: The Rules Become Visible and Contestable

The Pattern Is Not About Phones

What the Analogy Predicts

Human Intelligence Is Probably on the Lower Rungs of the Ladder

The Overrun

Logic Does Not Come Naturally

A Civilisation of Prosthetics

The Hardware

Evolution Stops at Good Enough

We Cannot See Our Own Reasoning

The Objections