Why I Wrote This
Before we begin
A woman stood in front of a police vehicle at a protest. She had one hand flat against it. Somebody took a photograph. The photograph travelled.
Within a few days a second story was travelling alongside it. The woman in the picture, people said, sold intimate photographs of herself online for money. I do not know whether that is true. I have not tried to find out, and I will explain in Chapter One exactly why not. What interests me is not her. It is what happened in the gap between the two stories.
In that gap, a large number of people performed the same manoeuvre at the same time, without coordinating and without noticing they were doing it. They took a claim about a woman’s body and used it to settle a question about a woman’s politics. Almost nobody stopped to ask what one had to do with the other. The move felt natural. That is the interesting part. When something that does not follow logically feels natural to millions of people at once, there is a machine underneath, and the machine is old, and it is worth taking apart.
I want to be honest about how this series started, because it started with irritation rather than curiosity. I found the manoeuvre disgusting and I wanted to say so. Then somebody told me that rules like this exist for reasons, that they have survived a long time, and that people who tear down old rules without understanding them cause disasters. That is not a stupid thing to say. It is a serious argument with a serious history behind it, and it deserved better than my irritation.
So I decided to do the work properly. Sixteen parts. Where these rules came from. What they were for. Who paid for them. What the evidence actually shows about what happens to a woman who breaks them, and what happens to a society that stops enforcing them. Whether men and women are the same, in what ways, by how much, and what follows from the answer. What sixty years of the equality project has actually produced, measured rather than asserted. What happens to a population that stops having children. And what nobody knows, which turns out to be a great deal more than either side admits.
I am going to be blunt. Not cruel, but blunt. This subject is drowning in politeness in one direction and abuse in the other, and neither produces knowledge. Where the evidence supports something that will annoy people who think of themselves as progressive, it goes in at full strength. Where it supports something that will annoy people who think of themselves as traditional, it goes in at full strength too. If you finish a part of this series without having been made uncomfortable at least once, I have failed at it.
One promise, which is the only promise this series makes. I am not going to pick a side for you. I will show you what each position claims, what evidence it actually has as opposed to the evidence it says it has, and where it is guessing. Then you decide. Where my own position could pull the writing, I will say so in the text at the point where it applies, rather than hiding it in a note at the front that nobody reads.
How to Read the Boxes
The notation, taught before you need it
Six kinds of box appear throughout. Each does one job. Here is a live example of each, so you know what you are looking at when it turns up in the middle of a chapter.
Norm: an unwritten rule about how people should behave. Not a law. Nobody votes on it and nobody signs it. It is enforced by ordinary people through approval and disapproval — a look, a comment, an invitation not extended, a marriage proposal that does not arrive.
Why it matters here: almost everything this series examines is a norm rather than a law. That makes it harder to see, harder to change, and much harder to argue with, because there is no office you can appeal to.
A Word Box appears the first time a hard word does. Not in a glossary at the back, not later. There is a glossary at the back as well, but you should never need it.
India has roughly 1.4 billion people. That number is unusable. Here is a usable version: if you shrank the whole world to a single village of one hundred people, about eighteen of them would be Indian, and about nine of those eighteen would be women and girls.
So when this series says something is true of Indian women, it is describing the lives of nine people in a village of one hundred. Whatever the rules do to them, they do at that scale.
Any number too big to picture gets converted into something with a body — a walk, a room, a wage, a village of one hundred.
Much of what we know about sexual behaviour comes from surveys where people are asked to report their own history. This is a weak instrument and it is worth knowing why before you read fifteen parts built partly on it.
In survey after survey, in many countries, men report substantially more opposite-sex partners than women do. In a closed population this is arithmetically impossible. Every act has two participants. The averages must come out close to equal.
What it can show: that men and women answer the same question differently, reliably, everywhere. What it cannot show: which of them is wrong, or by how much. The most likely answer is both, in opposite directions, because the two sexes are punished for opposite answers.
This box names the actual evidence — the survey, the register, the court record, the bone — and then says what that evidence cannot show. It is the box that should make you trust the rest.
This is the question the protest photograph raises, and people answer it with total confidence in both directions.
An argument is either sound or it is not. Whether the person making it is a saint or a criminal changes nothing about whether the police were right or wrong that day. Attacking the person instead of the claim is the oldest way of avoiding a claim you cannot answer.
We cannot personally check most of what we are told, so we necessarily rely on the character of the teller. A person’s conduct is information about how much weight to put on their word. Pretending otherwise is not high-mindedness, it is a refusal to use evidence you actually have.
What would settle it: nothing general. It has to be settled claim by claim, by asking whether the claim is checkable without trusting the speaker.
Why people care so much: because the second position is a licence. If character is always relevant, then any opponent can be answered by investigating their life rather than their argument, and the investigation only ever has to succeed once.
An Argument box appears wherever people who have studied something genuinely disagree. Each side gets its strongest case, not a version built to lose. Then a verdict that says where things actually stand, what would settle it, and why people are fighting so hard.
Both sides of the argument above assume the same thing without saying it: that the woman’s conduct, if real, is discrediting in the first place.
Notice that the defence and the attack agree on this. The attack says she sells images, therefore she is not worth listening to. The defence says her private life is irrelevant, so do not bring it up. Neither says the obvious third thing — that selling images might simply not be a mark against a person at all.
That silence is the shared premise. And a shared premise is invisible precisely because the two loudest voices in the room are not arguing about it.
This is the signature box of the method, and the hardest to write. It is not a caveat or a limitation. It digs out something both sides of a fight are taking for granted while they fight. There are between four and six in every part.
Every chapter closes with one of these. It restates the chapter in the plainest words available, with the key terms in bold.
If you read only the Remember This boxes and nothing else, you should still finish this document holding the whole argument.
A Warning About This Series
Read this before Chapter One
Three warnings, and then we start.
The material is live and it is personal. This is not a series about a dead empire. It is about rules that are being enforced on somebody today, possibly on you, possibly by you. Later parts contain figures on violence, on suicide, and on the deaths of women and girls who were never born. I am not going to soften those numbers, and some readers will find them hard.
I am not neutral and I will not pretend to be. I am an Indian man. I have never been on the receiving end of a single rule this series examines. That is a serious limitation and it does not go away by my mentioning it, so I will flag it again at each point where it could bend the writing. There is also a well-worn habit among educated Indian men of writing about women’s freedom in a way that is mostly about sounding modern to other men. I know that habit is available to me. Watch for it, and write to me when you catch it.
Simple words, serious content. Every hard idea in this series is explained in ordinary English, on the spot, the first time it appears. That is a choice about sentences, not about depth. Nothing is left out to make it easier. If a passage looks simple, it is because the sentences were made simple, not because the argument was made small.
1A Photograph and a Rumour
A woman put her hand on a police vehicle and somebody photographed it. Days later, a different kind of story arrived. What happened in between is the subject of this entire series.
1.1 — What happened, in order
Start with the sequence, because the sequence is the thing people stop being able to see once they have picked a side.
First: a protest. A woman is standing in front of a police vehicle. One hand rests flat against it. She is not throwing anything. She is standing still, which is why the image works — a still body against a machine built to move.
Second: the photograph spreads. People share it. Some share it because they support what she is protesting. Some share it because a single figure against a vehicle is a good picture, and good pictures travel whether or not people care about the cause.
Third, and this is where it gets interesting: a few days pass. Then a second story appears next to the first one. The woman in the picture, it is said, sells intimate photographs of herself on the internet for money.
Fourth: a very large number of people treat the second story as an answer to the first. Not as extra information. As a reply. As though somebody had asked a question and this settled it.
Nobody organised this. There was no meeting. Thousands of people who have never spoken to each other performed the same move within a day or two of one another, and almost all of them experienced it as obvious. That is what I want you to hold on to. Not that it was wrong — we will get to that — but that it felt like nothing. It felt like a reply.
1.2 — The manoeuvre, named
Let us be precise about what the move actually is, because naming it correctly matters more than condemning it.
Ad hominem: Latin, meaning “to the person”. It describes any move that responds to an argument by talking about the person who made it rather than the argument itself.
Why it matters here: the phrase is often used as though naming it wins the fight. It does not. As Chapter One will show, going after the person is sometimes legitimate and sometimes not, and the difference is not a matter of manners. It depends entirely on whether the claim under discussion can be checked without trusting the speaker.
There is a second word worth having, because it describes something slightly different and more specific.
Genetic fallacy: judging a claim by where it came from rather than by whether it is true. “That idea came from a bad person, so it is a bad idea.” The word has nothing to do with genes. It comes from the Greek for origin.
Why it matters here: it is the exact shape of the whole conservative argument this series has to deal with, only pointed the other way. “This rule came from a wise tradition, therefore it is a good rule” is the same move as “this claim came from an immoral woman, therefore it is a false claim.” Both judge a thing by its parentage. We will meet this again in Chapter Three and it will be much less comfortable there.
So the manoeuvre in full: a claim was made about how the police behaved at a protest. That claim is a matter of public record. There is footage. There are witnesses. There are, in most such cases, official statements. Whether the police acted lawfully can be established without knowing a single thing about the woman in the photograph.
The rumour does not address the claim. It cannot address the claim. It changes the subject from the conduct of an institution to the conduct of a woman, and it does this so smoothly that most participants did not notice a subject had been changed at all.
1.3 — Why I am not going to look
I could try to find out whether the rumour is true. I have decided not to, and the reason is not delicacy. It is method, and it is worth setting out plainly because it applies to the whole series.
She is a private person. She did not stand for office. She did not publish a book. She stood in a street and somebody pointed a camera. Nothing about that act makes her life a public archive, and a series that begins by rummaging through it would be performing the exact behaviour it claims to be studying. You cannot dissect a specimen by becoming one.
There is a second reason, and it is the stronger one. The truth of the rumour does not matter to any question this series asks. Consider both outcomes.
Suppose it is false. Then a woman was defamed for standing in a street, and we learn that the accusation is available as a weapon whether or not it is earned. That is a fact about the weapon, not about her.
Suppose it is true. Then a woman who sells images of her body also went to a protest, and the question becomes why one of those facts is supposed to cancel the other. That is a fact about the rule, not about her.
Either way, the interesting object is the same. She is the occasion for this series, not its subject. From here on she does not appear again as a person. She appears as a shape — the shape of what happens when a claim about a woman’s body is offered as a reply to a claim about power.
It is worth understanding the scale of what a rumour like this does, because “it spread online” hides the size of the thing.
A moderately viral post in India reaches a few million people. Take three million as a working figure. If you tried to tell three million people something in person, at one person per minute, working eight hours a day without a single day off, it would take you a little over seventeen years.
Now consider the correction. Corrections travel to a small fraction of the audience of the original — studies of misinformation consistently find the follow-up reaching a much smaller group than the claim. So the practical effect is close to permanent. Seventeen years of speaking, delivered in an afternoon, and no realistic way to take it back.
1.4 — Everybody uses this move
Here is where I have to be careful, because it would be easy to write this chapter as though only one political side does this. That would be false and you should not trust a series that pretends otherwise.
The move is universal. It has no politics. It has a shape, and any group with a grievance will reach for the shape.
A conservative commentator is found to have had an affair, and his arguments about family are treated as thereby refuted. A feminist writer turns out to employ domestic staff on poor wages, and her arguments about exploitation are treated as thereby refuted. A climate scientist takes a flight. A vegetarian is photographed eating fish. A politician who campaigned on austerity is found to have an expensive watch. In every case the same operation runs: locate a gap between the person and their position, then treat the gap as an answer to the position.
Sometimes it genuinely is an answer. If a man argues that his diet is easy to follow and he is secretly not following it, his failure is evidence about the diet. The claim was partly about him.
Usually it is not. Whether the police acted lawfully on a particular afternoon is not partly about the woman standing in front of the vehicle. There is no version of the question where her private conduct is an input.
Everyone in this fight — the people making the accusation and the people rejecting it — is assuming that the audience was ever weighing an argument.
The defence says: her private life is irrelevant to whether the police acted lawfully. True. But it assumes the crowd was in the business of evaluating police conduct and has now been distracted from that task.
Look at the behaviour instead of the words and a different picture appears. Most people did not examine the police claim before the rumour arrived, and did not examine it after. What they were doing, before and after, was deciding which side to be on. The rumour did not distract them from an evaluation. It gave them a reason to feel good about a decision they had already made.
The general form: treating persuasion as the thing that happened, when the thing that happened was permission. Most public argument is not people changing their minds. It is people acquiring licence for a position they already held. Once you see it, a great deal of what looks like debate in the coming fifteen parts stops looking like debate.
1.5 — What the move is actually for
If the manoeuvre does not work as reasoning, why does it work at all? It clearly works. That is not in question.
It works because it does a different job from the one it appears to do. It appears to be an argument. It functions as a cost.
Think of what the woman in the photograph now faces, irrespective of whether the rumour is true. Her name is attached to a claim she cannot disprove. Employers will search for her. So will future landlords, future in-laws, and every person she meets for the rest of her life who has a phone. She did not lose an argument. She acquired a permanent expense.
And every woman watching learned the price of standing in the street. That is the actual output. Not persuasion — pricing. The accusation is a bill sent to one person and a price list published to everybody else.
Hold that idea, because it returns in Chapter Five in a much more uncomfortable form. When somebody warns a woman that she will ruin her life, we are going to have to ask whether that is a forecast or an invoice.
A woman was photographed at a protest. Days later a rumour about her body was offered as a reply to a claim about police conduct. Thousands of people performed this move simultaneously and experienced it as obvious.
Going after the person instead of the claim is legitimate only when the claim cannot be checked without trusting the person. Police conduct at a protest can be checked. So the move was not reasoning.
The move belongs to no political side. Every group reaches for it. Anyone who tells you only their opponents do this is selling something.
I will not investigate the woman, because the truth of the rumour changes nothing. If false, we learn about the weapon. If true, we learn about the rule. Either way she is the occasion, not the subject.
The accusation was not an argument that failed. It was a price, charged to her and published to every woman watching.
2Three Claims Wearing One Coat
“These rules exist for a reason, and if you break them there will be consequences, so behave yourself.” That is one sentence. It is three separate claims, and they can be true and false in any combination.
2.1 — One sentence, three jobs
When somebody defends the rules about women, they almost never make one claim. They make three, joined so smoothly that the joins are invisible. Pulling them apart is the single most useful thing in this document, and everything in the remaining fourteen parts depends on it.
Here is the sentence in its most common form:
Now here are the three claims inside it.
Claim one is about the past. These rules came from somewhere. They had a function. They solved a problem that a society actually had.
Claim two is about the future. If the rules go, something bad follows. Damage. Disorder. Collapse. The “blunder”.
Claim three is about you. Therefore a particular woman, today, should behave in a particular way.
These are not three versions of one idea. They are three different kinds of statement, and each requires a completely different kind of evidence.
Claim one is a historical and anthropological question. It is answered by looking at what societies actually did and why, and by comparing societies that did it differently. That is Parts Two and Three.
Claim two is a prediction. It is answered only by looking at what has happened in places where the rules already changed. Not by reasoning about what ought to happen. By counting what did. That is Part Thirteen, and it is the part I expect to annoy the most people.
Claim three is a moral claim about what a specific person owes. Nothing in claims one or two produces it by itself, and Chapter Four explains exactly why not.
Underneath the welded sentence sits a premise that neither side ever states: that a rule and its justification are the same object.
Watch what happens in a normal argument. The traditionalist defends the rule by explaining where it came from. The reformer attacks the rule by attacking the explanation — “that is not even where it came from, that is a myth.” Both behave as though disproving the story disposes of the rule, or as though establishing the story secures it.
But rules and their stories are separate things, and they come apart constantly. A rule can be excellent and its official story completely invented. A rule can be terrible and its story entirely accurate. Most surviving rules have picked up justifications long after the fact, from whoever needed one — which means the story you are given is usually evidence about the present rather than the past.
The general form: attacking the deed by attacking the receipt. You will see this in nearly every argument in this series, on both sides, and once you notice it you cannot stop noticing it.
2.2 — The claim about the past
Claim one — “it exists for a reason” — is the one that sounds most modest and is most often waved through. It should not be.
It contains a hidden strength and a hidden weakness. The strength: it is often true. Rules about women are not random. They cluster around a small number of recurring problems — who inherits, who is responsible for a child, which families may marry which, how a group holds its boundary. Those are real problems and they were solved somehow. Part Two shows the machinery.
The weakness: “it exists for a reason” does not specify whose reason. And that omission is not a small gap. It is the whole argument.
Consider a rule that says the eldest son inherits all the land and his brothers get nothing. That rule existed for a reason, and the reason was excellent: it stopped farms being divided into strips too small to feed anybody. It genuinely solved a genuine problem.
Ask the younger brothers whether it worked and you get a different answer. It worked for the estate. It worked for the family as an accounting unit. It worked for the eldest son, spectacularly. It did not work for the second son, who left for the army or the church or the sea, and it is not obvious that his interests are less real than the estate’s.
So “it worked” and “it worked for her” are separate findings, and the first is regularly used as though it establishes the second. Hold that thought for Chapter Three, where it becomes the central problem.
2.3 — The claim about the future
Claim two — “if you change it, there will be a blunder” — is the only one of the three that can be settled by evidence, and it is the one people are least interested in checking.
Notice what kind of statement it is. It is a prediction. It says that if a society removes a rule, specific bad things follow. That is exactly the sort of claim that can be tested, because the experiment has already been run, repeatedly, by real countries, at full scale, within living memory.
Sweden ran it. So did Japan, South Korea, Iran, Saudi Arabia, the United States and Ireland, each in a different direction and at a different speed. We have decades of data on marriage, birth rates, violence, employment, disease and reported happiness from every one of them.
This is a gift and almost nobody accepts it. The people most convinced of the coming blunder are usually least able to name what the blunder consists of, or which country has already suffered it, or what number would show it had happened.
So I am going to insist on a discipline throughout this series, and I will apply it to both sides equally.
The rule I will hold myself to: a prediction that cannot say what would count as it being wrong is not a prediction. It is a mood.
“Society will decay” is a mood. It names no measurement. Any outcome can be fitted to it afterwards, which means no outcome can test it.
“Divorce rates will rise, births will fall below replacement, and reported loneliness will increase” is a prediction. Each part is countable. Each can fail.
Here is why this matters more than it looks: when you convert the mood into a prediction and then go and check, some of it comes back true. Birth rates really did collapse in the countries that changed fastest. That finding survives the test, and I am not going to hide it because it is inconvenient. Part Nine is built on it.
What the method cannot do: tell you whether a fall in births is a catastrophe or a preference. Counting settles what happened. It does not settle what it is worth.
2.4 — The claim about you
Claim three is the one that actually does the work in a real conversation, and it is smuggled in behind the other two.
Suppose claims one and two are both entirely true. The rules did solve a real problem, and removing them does produce measurable damage. Does it follow that a particular woman today should not wear what she likes, or sleep with whom she likes?
Not by itself, and the gap is not a technicality. There are at least four separate steps still missing, and each of them is contestable.
You would need to show that the damage is large enough to outweigh what the rule costs the person carrying it. A rule that prevents a small harm at the price of a large one is a bad rule even when the small harm is real.
You would need to show that this individual’s compliance actually affects the outcome. One woman’s clothing does not move a national birth rate. Rules that only work in aggregate impose a real cost on each person for a benefit no person produces.
You would need to show that no cheaper solution exists. If the problem was uncertain fatherhood, we now have paternity testing. If the problem was disease, we have antibiotics and contraception. If the problem was destitution after abandonment, we have wage labour and, in some places, a welfare system. A rule justified by a problem that has since been solved another way is a fence around an empty field.
And you would need to show why the cost falls on her rather than on him. Every one of these problems has two participants. The rules are overwhelmingly written for one of them. Part Eleven is about that asymmetry and it does not have a comfortable answer.
2.5 — Eight combinations, and where real arguments live
Because the three claims are independent, they can take any combination of true and false. Three claims, two values each, gives eight possible positions. Most real arguments are two people occupying different boxes and thinking they are in the same one.
| Past | Future | Obligation | What this position looks like in the wild |
|---|---|---|---|
| True | True | True | The full traditionalist case. Rare in this pure form, because it requires the fourth step in §2.4 to be argued rather than assumed. |
| True | True | False | The most defensible conservative position, and the most interesting. Yes it had a function, yes removing it has costs, and no, that does not give anyone authority over this woman. |
| True | False | False | The standard modernising view. It made sense once, the problem has been solved another way, the fence can come down. |
| False | True | False | Unusual and worth taking seriously. The origin story is a myth, but the rule is now load-bearing anyway because a society has been built on top of it. |
| False | False | False | The strong reformist case. Invented past, imagined blunder, no obligation. |
| False | True | True | Honest authoritarianism. It does not matter where it came from; the consequences of dropping it are severe enough to justify compulsion. |
| True | False | True | Incoherent as stated, but common in practice — usually a person defending a rule on the basis of continuity alone. |
| False | False | True | Naked assertion of authority. Rare to hear out loud, not rare to meet. |
Look at the second row. That is the position I think is hardest to beat, and almost nobody argues for it, because it is unsatisfying to both camps. It concedes everything the reformer wants about authority and everything the traditionalist wants about consequences.
2.6 — Running the tool on a real sentence
Take a sentence you have certainly heard, in one form or another.
Separate the three claims and the sentence stops being a wall.
Past: there is a long-standing norm against it, and there was a reason. Very likely true. Part Two gives the reason and it is not mysterious.
Future: such women end up unhappy. This is a measurable claim. There is real research on it, some of which supports the claim more than reformers expect. Part Ten reports it honestly, including the parts I would rather not report.
Obligation: therefore you should not. Does not follow, even if both the others hold, until the four missing steps are argued.
And now the question that Chapter Five is built on, which the elder’s sentence conceals completely. Unhappy because of what? If the unhappiness comes from the act, that is a fact about the act. If it comes from what the village does to her afterwards, that is a fact about the village — and the elder saying the sentence is one of the villagers.
The standard defence of the rules about women is one sentence containing three claims: about the past (it had a function), about the future (changing it causes damage), and about you (therefore obey).
Each needs a different kind of evidence. History for the first, measurement for the second, moral argument for the third. They can be true and false in any combination — eight positions in all.
“It exists for a reason” never says whose reason. A rule can work perfectly for a family, an estate or a caste while working badly for the person carrying it.
Getting from “changing it causes damage” to “so you must obey” needs four further steps: that the damage outweighs the cost of the rule, that this individual’s compliance matters, that no cheaper solution now exists, and that the burden belongs on her rather than on him.
Almost every argument you have ever had about this was two people arguing about different claims while using the same sentence.
3“It Exists For A Reason”
The strongest argument for leaving old rules alone is about a hundred years old and takes the form of a fence in a field. It is a good argument. It is also missing one word, and the missing word changes everything.
3.1 — The fence in the field
Imagine you are walking across a field and you come to a fence. It runs across open ground. There is no gate, no animal, no crop, nothing on either side that the fence appears to be separating. It seems pointless.
A companion says: this is useless, let us pull it down.
The English writer G. K. Chesterton put the reply in an essay in 1929, and the reply has outlived almost everything else he wrote. He said the correct answer is not “leave it alone” and not “pull it down”. It is: go away and find out why it was put there. Come back when you know. If you then still think it should go, take it down. But the person who cannot say why the fence exists has not earned the right to remove it, because their confidence is built on their own ignorance.
Chesterton’s Fence: the principle that you should not remove something whose purpose you do not understand, until you understand it. Named after the writer who described it. It is not a rule against change. It is a rule against a specific kind of confidence — the kind that mistakes “I cannot see why this is here” for “there is no reason this is here.”
Why it matters here: this is the strongest form of the argument your elders are making, whether or not they have heard of Chesterton. Any honest treatment of the rules about women has to answer it rather than sneer at it.
Notice how good this is. It does not claim old things are wise. It makes a claim about the person doing the removing — that they are likely to be missing information, that a rule’s function is often invisible to somebody who has only ever lived downstream of it, and that the cost of the missing information falls on people who had no vote.
Applied here, it goes: you think a rule about how women dress is arbitrary. But you were born after it was already working. You have never seen the problem it prevents, precisely because it prevents it. Your sense that nothing bad would happen is not evidence. It is the expected experience of somebody standing behind a wall.
That is a serious argument. I want it on the table at full strength before I take it apart, because a series that dismisses it has nothing to say to anyone who does not already agree.
3.2 — What the fence argument gets right
Three things, and all three are real.
Invisible success looks like pointlessness. This is genuinely how prevention works. A vaccine that works produces no disease to see, and then people conclude the vaccine is unnecessary. A dam that holds produces no flood, and the town below it forgets what the river does. Any successful protective institution eventually looks like an unexplained fence, and the more successful it is, the more pointless it looks.
Rules encode knowledge nobody has written down. A community can carry a solution for centuries without any individual member being able to explain it. The practice survives; the explanation is lost or was never explicit. If you demand the explanation before you allow the practice, you will destroy working things at a considerable rate.
The people who remove rules rarely pay for removing them. Reform is generally designed by people with resources and absorbed by people without. If a change goes wrong, the person who wrote the article is fine. That asymmetry is real and it is a good reason for caution.
3.3 — The missing word
Now the problem. Chesterton’s fence tells you to find out why the fence is there. It quietly assumes there is a why — one purpose, held by the field.
Fields do not have purposes. People do. And when you actually go and find out why a fence was built, you almost never discover a purpose. You discover a beneficiary.
The honest version of the question is therefore not “why is this fence here?” It is: who put this fence here, what did they get, and who was on the wrong side of it?
Once you ask it that way, the whole argument changes shape. Because “this rule has survived four thousand years” is now a much weaker statement than it sounded. Survival tells you a rule served somebody’s interest reliably enough that they kept enforcing it. It says nothing whatever about whether it served the interest of the person it constrained.
In fact, think about the selection pressure. A rule that is enforced by the powerful on the powerless has an obvious survival advantage over a rule that runs the other way, entirely independent of whether it is any good. Longevity is evidence of enforcement, not of benefit. Those are different findings and they are being confused constantly.
Both the traditionalist and the reformer accept, without ever discussing it, that a custom which survived must have served the person it constrained.
The traditionalist uses it directly: it lasted, so it works, so it works for her. The reformer accepts the same premise but denies the antecedent: it does not work for her, so it must never have worked at all, so it was pure oppression from the start.
Both are wrong in the same way. A rule can work superbly — solve a real coordination problem, hold a lineage together, prevent a genuine harm — and be paid for entirely by one group. Those are not alternatives. That is the normal case.
The general form: reading a system’s survival as evidence about everybody inside it. A machine that runs well tells you it is well engineered. It tells you nothing about what it is made of, or who is inside it.
This single confusion is responsible for more bad argument in this field than any other, and both camps depend on it. Neither can afford to look at it.
3.4 — Four fences that came down
The fence argument is best tested against fences that were actually removed. Here are four, chosen because in each case the defenders made exactly the argument we are examining, and in each case we now know what happened.
Foot-binding in China. For roughly a thousand years, girls’ feet were broken and bound to keep them small. It was defended as beauty, as decency, and as the mark of a family that could afford not to have its daughters work. Every defence available today was made then, including the argument that unbound daughters would be unmarriageable — which was true, and is the most important thing about it.
It ended fast. The mechanism is worth knowing because it is the most instructive thing in this chapter. Reformers did not primarily argue that binding was cruel; everyone knew it was cruel. They formed pledge societies in which families swore two things together: not to bind their daughters’ feet, and not to let their sons marry bound-foot women. Once enough families signed, the marriage penalty inverted. Then it collapsed, in decades, after a thousand years.
Read that again, because it tells you what kind of object we are dealing with. The rule was not held up by belief. It was held up by every family’s correct prediction about every other family. Nobody had to be persuaded that binding was good. They only had to believe that others would punish them for stopping. Change the expectation and the fence falls over on its own.
Sati in India. Widow-burning was defended as scripture, as tradition, as a woman’s own devotion. It was outlawed in 1829, after campaigning by Indians and foreigners together, and the strongest arguments against abolition were about the danger of outsiders touching a people’s customs. That argument was not silly. Colonial interference in Indian practice was frequently self-serving and often catastrophic. It was also, in this instance, on the wrong side.
Duelling. Men killed each other over insults for centuries. The practice was defended as the foundation of honour and the only guarantee of civil speech among gentlemen. That last defence contained something true: without duels, what stops men insulting each other? The answer turned out to be that nothing much stops them, and this is fine, and the cost of the insults is enormously lower than the cost of the corpses.
Primogeniture. All land to the eldest son. Defended as the only way to stop estates being cut into unviable slivers. Genuinely solved that problem. It was dismantled across much of the world and the predicted agricultural collapse did not arrive, because other instruments — mortgages, partnerships, land consolidation, wage labour — did the same work without disinheriting four children out of five.
3.5 — And one that came down badly
I would be cheating if I only listed fences whose removal went well. Here is one that did not, because the lesson from it is the most useful thing in the chapter.
Through the middle of the twentieth century, Britain and the United States held large numbers of mentally ill people in institutions. Conditions in many were genuinely appalling, the criticism was justified, and from the 1960s onward the institutions were closed on a large scale.
The replacement — community-based care, funded and staffed — was promised and, in large part, never built. The result was not liberation. In both countries a substantial share of severely mentally ill people ended up homeless or in prison. Prisons became, by default, the largest mental health institutions in both nations.
Now notice precisely what went wrong, because it is not what a traditionalist would predict. The old institutions were not vindicated. Nobody looking at the outcome concluded that the asylums had been good. What went wrong was that the fence was removed and the thing that was supposed to replace it was never constructed.
That is the real lesson, and it applies directly to everything in the remaining parts of this series. The dangerous move is not removing a rule. It is removing a rule while leaving in place the conditions that made the rule necessary.
Hold on to that sentence. In Part Thirteen we will find that the countries which suffered most from changing the rules about women were not the ones that changed them hardest. They were the ones that changed half.
This is the disagreement underneath every conversation in this series, and both sides have more than they are usually given credit for.
Customs are not designed, they accumulate. Practices that damaged the groups holding them tended to disappear along with those groups. What survives has therefore been filtered by a long and brutal process, and it carries information that no individual possesses. A person reasoning from first principles in one lifetime is competing against a filter that ran for millennia. Betting against it requires better evidence than a strong feeling.
The filter selects for what keeps a practice being enforced, which is not the same as what makes people’s lives go well. A rule imposed by a powerful group on a weak one survives because the powerful group keeps imposing it. That is a fact about power, not about benefit. Foot-binding survived a thousand years and broke children’s feet the entire time. Longevity measures durability, and durability is exactly what you would expect from a rule the beneficiaries control.
Longevity is weak evidence that a rule solved something, and no evidence at all about who it solved it for, or whether the problem still exists. It sets a duty to investigate, not a duty to obey. It also raises the standard of proof for removal without making removal illegitimate.
What would settle it: in any particular case, an answer to three questions — what problem did this solve, does the problem still exist, and is there now a cheaper instrument? Every one of those is answerable for most of the rules in this series, which is why we are spending fifteen more parts answering them rather than debating in the abstract.
Why people care so much: because the abstract version of this argument is unwinnable, and an unwinnable argument is extremely useful. It lets a person defend any specific rule without ever defending that rule.
Chesterton’s Fence says do not remove something whose purpose you do not understand. It is a serious argument and three parts of it are true: successful prevention looks like pointlessness, rules carry knowledge nobody wrote down, and reformers rarely pay for their own mistakes.
Its flaw is one missing word. It tells you to ask why the fence is there. The honest question is who — who built it, what did they get, and who was on the wrong side.
A rule surviving a long time proves it kept being enforced. That is evidence about power, not about benefit. Both camps confuse these constantly.
Foot-binding lasted a thousand years and fell in decades, once families stopped believing other families would punish them. The rule was held up by expectations about other people, not by belief in the rule. Remember that. It is the most important mechanism in this series.
The genuine danger is not pulling down a fence. It is pulling down a fence and never building what was supposed to replace it.
4The Jump From Is To Ought
Suppose we establish exactly where the rules came from and exactly what they do. There is still a gap between that and what anybody should do. Both sides of this fight jump the gap, in opposite directions, and both think only their opponents do it.
4.1 — The gap
In 1739 a Scottish philosopher named David Hume noticed something about the way moral arguments are written. Authors would spend pages describing how the world is — what people do, what nature does, what God is said to have done — and then, without warning, the sentences would quietly change. Instead of saying what is, they would start saying what ought to be.
Hume’s point was not that the conclusion was wrong. It was that a new kind of claim had appeared without anybody explaining where it came from. A description had turned into an instruction and no step had been shown.
The is/ought gap: the observation that no collection of facts about how the world is, by itself, produces a conclusion about what anybody should do. Getting from one to the other always needs at least one extra ingredient — a value, a goal, or a standard — and that ingredient has to be argued for separately.
Why it matters here: nearly every argument in this series will present facts and then act as though an instruction has been proved. It has not. Watch for the moment the sentences switch, and ask what was added.
Here is the gap in ordinary form. “Most societies have restricted women’s clothing.” That is a fact, and as it happens a true one. It contains no instruction. To get to “therefore women should dress modestly”, you need something extra — perhaps “we should do what most societies do.” But that is not a fact. It is a value, and a very strange one, since most societies also kept slaves.
The extra ingredient is always there. The dishonesty is never in having one. It is in hiding it.
4.2 — Nature as an argument
The most common version in this field runs through the word natural.
It goes: this arrangement is natural, therefore it is right. Women are naturally more nurturing, so they should raise the children. Men are naturally more aggressive, so they should lead. Monogamy is natural, or polygamy is natural, depending on who is speaking.
Appeal to nature: arguing that because something is natural it is good, or because something is unnatural it is bad.
A related term you will meet is the naturalistic fallacy, coined by the philosopher G. E. Moore in 1903. Strictly it means something narrower — the error of defining goodness as some natural property. In everyday use the two have merged, and I will use the plainer phrase.
Why it matters here: this is the single most common move in arguments about men and women, and it is made with total confidence by people who would instantly reject it in any other setting.
The problem is not that the factual claims are always false. Some are true, and Part Four goes through them with actual numbers. The problem is that natural does no moral work whatsoever, and everyone knows this in every other context.
Cholera is natural. Childbirth killed a large share of women for most of human history and that was entirely natural. Tooth decay is natural. Dying at thirty-five is natural. Every one of us spends every day fighting nature with soap, spectacles, antibiotics and roofs, and nobody thinks this needs defending.
Meanwhile a great deal we consider obligatory is deeply unnatural. Universal literacy. Refusing to hit a child. Anaesthetic. Not killing your rival. The entire apparatus of law exists precisely because the natural outcome was judged unacceptable.
So when somebody says an arrangement between men and women is natural and therefore correct, they are using a standard they apply nowhere else in their life. The word is not carrying an argument. It is carrying a feeling of inevitability, which is a different thing wearing the same coat.
There is a stronger version of the argument, and it deserves a hearing because it is not silly.
The crude appeal to nature fails. But something in the neighbourhood might survive, and dismissing the whole area is its own kind of laziness.
Nothing about a difference being natural makes it good. But it does tell you what changing it will cost. A society trying to produce an outcome that runs against a strong natural tendency will need continuous force to hold it there, and will fail the moment the force is withdrawn. That is not a moral argument. It is an engineering one, and ignoring it wastes enormous quantities of effort and human happiness.
Every human trait develops in an environment. There is no version of a person raised outside a society, so there is no baseline to measure against. What gets called natural is almost always just what is common, and what is common is largely produced by the arrangements under examination. The word smuggles in the conclusion it is supposed to prove.
Naturalness is not binary and it is not unmeasurable either. Traits vary in how much effort it takes to shift them, and that is a real number that can be estimated — through twin studies, cross-cultural comparison, and natural experiments where societies changed policy sharply. Some differences shrink to almost nothing when conditions change. Others barely move. Both findings exist, and lumping them together is the actual error.
What would settle it: for any given difference — effect sizes across many societies, what happens to the gap when conditions change sharply, and whether it moves in the direction the cultural explanation predicts. This work has been done for a lot of traits. The results are not what either side expects.
Why people care so much: because both camps have accepted a bad bargain — the belief that if a difference is natural, inequality is justified. Both then fight over the facts to protect the moral conclusion. The bargain is what should be rejected. The facts are just facts.
4.3 — The same error, running the other way
Now the part I am obliged to write, having promised that this series would cut in both directions.
There is a mirror-image mistake and it is just as common. It goes: this fact would have unpleasant implications, therefore this fact is false.
Moralistic fallacy: concluding that something cannot be true because its being true would be morally inconvenient. The reverse of the appeal to nature. If the first says is proves ought, the second says ought proves is.
Why it matters here: it is the characteristic error of the progressive side of this argument, and it is not a minor one. It has led to real findings being ignored, real researchers being attacked for reporting them, and real policies being built on claims that were never true.
You can hear it in sentences like these. Men and women cannot differ in any average trait, because admitting a difference would justify discrimination. Traditional societies cannot have offered women anything, because saying so would excuse them. Research showing a cost to some behaviour must be biased, because reporting it would help the wrong people.
Every one of these reasons from the conclusion backwards. And they share a structure with the appeal to nature that is worth seeing clearly: both accept that facts about difference automatically produce conclusions about worth. The traditionalist accepts it and embraces the facts. The progressive accepts it and denies the facts. Neither one questions the bargain.
Reject the bargain and both problems evaporate. Men and women can differ substantially in some measurable trait, and it can still be true that no person’s rights depend on where they sit in a distribution. Those two sentences are entirely compatible. Holding both is not a compromise. It is just accurate.
4.4 — Why this matters for what comes next
I am setting this out now because of what is in Parts Four, Five, Ten and Nine, and I would rather warn you than surprise you.
Part Four reports physical and psychological differences between men and women with their actual sizes. Some are enormous. That will read, to some people, as though I am building a case against equality. I am not, and Chapter Four is my defence in advance: a fact about averages produces no conclusion about anybody’s rights without an extra ingredient that has to be argued separately.
Part Ten reports evidence about sexual behaviour and later outcomes, including findings that support the traditional warning more than reformers expect. That will read to others as though I am building a case for control. I am not, and the defence is the same sentence.
Part Nine reports what is happening to birth rates and it is genuinely alarming. That will read as though I am joining a panic. The defence is again the same. A number tells you what happened. It does not tell you what to do, and anybody who says otherwise has added an ingredient without showing you the label.
So here is the discipline I am adopting, stated plainly so you can hold me to it. Where I report a fact, I will report it at full strength regardless of who it helps. Where I draw a conclusion about what anybody ought to do, I will show the extra ingredient and name it as a value rather than a finding. If you catch me skipping that step, I have broken my own rule and I want to know.
No pile of facts produces an instruction by itself. Getting from is to ought always requires an extra ingredient — a value or a goal — and honest argument shows that ingredient rather than hiding it.
Natural does no moral work. Cholera is natural; literacy is not. Everybody already knows this and applies it everywhere except here.
Naturalness does tell you something useful but different: the price of change. Some differences shrink easily when conditions change. Others barely move. Which is which is a measurable question, not a matter of opinion.
The mirror error — the moralistic fallacy — says a fact must be false because it would be inconvenient. It is just as common and just as wrong, and it is the characteristic mistake of the other camp.
Both errors share one bad bargain: that facts about difference decide questions of worth. Refuse the bargain and both collapse. A difference in averages settles nothing about what anyone is owed.
5When A Warning Is Really A Threat
“She will ruin her own life.” It sounds like a forecast made by somebody who has seen it happen. Sometimes it is. Sometimes the speaker is part of the machinery that does the ruining, and the sentence is not a forecast at all.
5.1 — The sentence
You have heard this sentence, or one very like it. An older relative says it. Sometimes a philosopher says it in longer words. Sometimes a woman says it about another woman, which is the version people find hardest to explain.
The sentence arrives dressed as kindness and experience. It says: I have watched this happen. I am telling you what I saw. I am trying to spare you.
And here is the thing that makes it so hard to answer — it is frequently accurate. Women who break these rules in India often do end up worse off by almost any measure you choose. Reduced marriage prospects. Family estrangement. Damaged employment. Sometimes violence. Sometimes death. The elder is not lying about the outcome.
So the reformer’s usual reply — that the elder is a bigot inventing consequences — is simply false, and everybody in the room knows it is false, which is why the reply never works. The consequences are real. Anyone arguing here has to start by conceding that.
The interesting question is not whether the prediction comes true. It is what makes it come true.
5.2 — Two sentences that sound identical
Compare these.
Both have the form of a warning. Both may be entirely accurate. But they are completely different kinds of object, and the difference sits in one place: what the speaker’s own role is in the outcome.
The first speaker has no role. Gravity does the work. If she leaves the room, if she dies, if she changes her mind entirely, the legs still break. The prediction is about the world.
The second speaker is the mechanism. The destruction is not something that will happen to the listener. It is something the speaker and people like the speaker will do. Remove them and the outcome changes.
Both sentences describe true futures. One is a forecast. The other is a sentence in the legal meaning of that word — a punishment announced in advance by somebody who will help carry it out.
Self-fulfilling prophecy: a prediction that causes the thing it predicts. Named by the sociologist Robert Merton in 1948. His original example was a bank rumoured to be failing. The rumour is false, but people withdraw their money because of it, and the bank then fails. The prediction was wrong when made and true when finished.
Why it matters here: it means “it came true” is not proof that a warning was knowledge. A prophecy that builds its own outcome will always look, afterwards, exactly like wisdom.
This is the crucial point and it is worth stating as flatly as possible. A rule that punishes people for breaking it will always produce evidence that breaking it is harmful. The evidence is manufactured by the punishment. And that evidence is then offered as the justification for the punishment, which completes a circle that can run forever without ever touching the outside world.
5.3 — Why this is not an accusation
I want to be careful here, because it would be easy to read the last section as saying that elders are liars, or that they are consciously threatening people. That is not what I am saying and it would be both unfair and wrong.
Nearly everyone saying this sentence means it kindly. The aunt who warns her niece is not plotting. She has watched women’s lives go badly and she is trying to prevent a repetition. Her motive is protective and her observation is real.
The problem is structural, not moral. She has watched the outcome many times. She has never once watched the counterfactual — the same woman making the same choice in a society that did not punish it. Nobody has shown her that comparison, so she attributes the damage to the choice, which is the only part she can see.
She is doing what any of us would do with the data available. She is also, without intending to, one of the hands that produces the pattern she is describing. Both things are true simultaneously and neither cancels the other.
5.4 — A test that actually works
So how do you tell them apart? If a warning is accurate either way, and the speaker is sincere either way, what distinguishes a genuine hazard from a manufactured one?
There is a test, and it is simple enough to state in one line. It is my own formulation rather than a technical term you will find elsewhere, so treat it as a tool rather than an authority.
The test: does the harm travel?
Take the same behaviour and look for it somewhere the punishing community has no power. A different country. A different class. A city where nobody knows the family. If the harm still appears, something in the act itself is producing it. If the harm disappears, the harm was the punishment.
What this can show: it separates hazards that follow a person across a border from hazards that stop at the edge of a community’s reach. That is a genuine and testable distinction, and the data to run it exists — the same behaviours have been studied across dozens of societies with wildly different enforcement.
What it cannot show: it is not clean. The women who leave are not a random sample of the women who stay, so differences between them are partly differences in who moves. And no society has zero enforcement, so we are comparing more against less, never presence against absence. This is the central difficulty of Parts Ten to Thirteen and I will flag it every time it applies.
Run the test on the roof and it passes instantly. Broken legs are the same in Delhi and Stockholm. Gravity does not check your passport.
Run it on the caste marriage and it fails just as fast. The destruction is administered by named people with a known address. Move beyond their reach and it does not follow.
Most real cases sit in between, and the middle is where the actual work is. Take the claim that a woman with many partners will be unhappy later. Some of the mechanisms proposed for this — regret, difficulty forming attachment, effects on later relationships — would travel. If they are real, they show up in Denmark. Others — being unmarriageable, family rejection, reputational collapse — clearly do not travel. They are local, and they stop at the boundary of the enforcing group.
So the sentence “she will be unhappy” is not one claim. It is a bundle, and the bundle can be taken apart by asking which pieces cross a border. That is precisely what Parts Ten, Eleven and Thirteen do, and it is why they are separate parts.
This is the fight underneath the whole series, and it is usually conducted with both sides talking past each other.
The elder has data the young person does not. She has watched forty years of outcomes and she is reporting a pattern she genuinely observed. Calling that a threat is an insult to somebody trying to help, and it also happens to be a convenient way of avoiding an unwelcome empirical claim. The pattern exists. Denying it does not make the young woman safer; it just leaves her uninformed.
The harm is overwhelmingly administered by the community issuing the warning. Take away the family that stops speaking to her, the marriage market that prices her out, the employer who reads the rumour, and most of the predicted damage vanishes. The speaker is describing what she and her neighbours will do, in the grammar of weather. That grammar is doing enormous work: it converts a choice into a natural fact.
Some of the harm travels and some does not. This is not a philosophical dispute but an arithmetic one, and the arithmetic can be done. The honest answer will be a split — this much is intrinsic, that much is imposed — and the split will differ by behaviour, by country and by decade.
What would settle it: comparison of the same behaviours across societies with sharply different enforcement, tracking the same people over time rather than asking them once, and looking at what happens when enforcement changes suddenly in one place. All three of these exist. None is clean.
Why people care so much: because the answer decides where responsibility sits. If the harm is intrinsic, the woman caused it. If the harm is imposed, the community did — and the community includes the person doing the warning. Almost nobody can examine this question neutrally, including me.
5.5 — The uncomfortable practical problem
Now the part that neither camp likes, and it is the reason this chapter cannot end tidily.
Suppose you fully accept that the harm is imposed rather than intrinsic. Suppose you are certain the punishment is unjust. What advice do you actually give a nineteen-year-old woman in a small town tomorrow morning?
Because the punishment does not care whether it is just. It will be administered anyway. Telling her that the harm is socially constructed is true and completely useless to her, in the same way that telling somebody the road is unfairly designed does not stop the traffic.
This is the genuine dilemma and both sides dodge it. The traditionalist dodges it by pretending the harm is natural, which converts a political fact into a permanent one. The reformer dodges it by pretending the harm is small, which is a lie that costs somebody else and never the person telling it.
The honest position is harder and less satisfying to say. The danger is real. The danger is also manufactured. Both facts stay true at the same time, and a person deciding what to do tomorrow has to weigh a real cost that should not exist.
That is not a resolution. It is the actual shape of the problem, and I would rather leave it in that shape than tidy it into something false.
“She will ruin her life” is usually accurate. The consequences are real. Any argument that starts by denying them fails immediately and deserves to.
But accuracy is not the question. The question is what makes it come true. A warning where the speaker is part of the mechanism is not a forecast; it is a punishment announced early, in the grammar of weather.
A rule that punishes rule-breaking always manufactures evidence that rule-breaking is harmful. That evidence is then used to justify the punishment. The circle can run forever without ever touching reality.
The test: does the harm travel? Look at the same behaviour where the punishing community has no reach. What follows her across the border is intrinsic. What stops at the boundary was imposed.
Almost nobody saying this sentence is lying or threatening. They have seen the outcome many times and have never once seen the comparison. They are reasoning correctly from data that has been built by their own participation.
The danger is real and the danger is manufactured. Both stay true, and the woman deciding tomorrow morning still has to pay.
6Who Is Counting, And What Do They Get
When a question that could be settled by evidence is instead fought like a war, something other than evidence is at stake. It is worth working out what, before we spend fifteen parts on the evidence.
6.1 — A question nobody wants answered
Here is something odd about this whole field, and you should find it odd too.
Most of the disputed claims in this series are countable. Whether women with more partners are less happy later — countable. Whether birth rates fall when rules relax — countable. Whether violence rises — countable. Whether societies that liberalised collapsed — extremely countable, and several of them are still standing there available for inspection.
Yet the arguments do not proceed like arguments about countable things. Nobody says “let us look.” People arrive holding conclusions and search for numbers afterwards. When a number goes the wrong way it is attacked rather than absorbed.
That pattern is itself information. When people fight an evidential question politically, it is because something other than the evidence is riding on the answer. Naming what is riding on it does not settle anything, but it tells you what you are watching.
6.2 — What each answer is worth, and to whom
Let me go through this without accusation. These are not conspiracies. Nobody meets. They are incentives, and incentives shape what people find persuasive without anybody having to be dishonest.
Families with unmarried daughters. In a society where marriage is the main route to security, a daughter’s reputation is a financial asset. Anything that raises the price of female reputation raises the value of that asset for families who have invested in it. It also raises the risk. Both effects push the same way: towards enforcement.
Families with unmarried sons. Their interest runs the other way and it is rarely mentioned. Strict rules on women reduce the supply of acceptable partners, which is bad for a family holding a son. This is one reason enforcement is never uniform: the same community contains groups with opposing stakes in the same rule, and which group is louder shifts with the sex ratio and the economy.
Religious and community authorities. Any institution whose authority rests on interpreting rules gains when the rules are important and loses when they are ignored. This is not cynicism about faith; it is a straightforward observation about institutions. An office that adjudicates behaviour needs behaviour to adjudicate.
Political movements. Threat to women is among the most reliable mobilising devices ever discovered, and it works for every faction. “Our women are being corrupted by outsiders” builds one kind of movement. “Our women are being destroyed by our own men” builds another. Both are used, both recruit effectively, and neither is primarily a claim about women.
The outrage economy. This one is new and it changes everything, because it is the first stakeholder that profits from the argument itself rather than from either answer. Platforms are paid for attention. Content about women’s bodies and behaviour reliably produces attention. The result is a permanent industrial incentive for the fight to continue and never resolve — which is a genuinely novel condition, and Part Fourteen returns to it.
Researchers. Including the honest ones. Certain findings are easier to publish, easier to fund and safer to hold in some institutions, and the direction of that pressure has flipped at least once in the last fifty years. Part Seven deals with what this does to the evidence base and it is not a small effect.
6.3 — Nobody believes it and everybody obeys it
Now the most useful single finding in this chapter, because it shows that a rule can be enforced by an entire population that privately rejects it.
In the 2010s a team of economists ran a study in Saudi Arabia, published in a leading economics journal in 2020. They asked young married Saudi men, privately, whether they thought it was acceptable for women to work outside the home.
Close to nine in ten said yes. Then they were asked what they thought other men like them believed. They substantially underestimated it. Almost everyone privately agreed, and almost everyone thought they were unusual.
The researchers then told a randomly chosen group the true figure — nothing else, just the correct number. Months later, the wives of the men who had been told were measurably more likely to have applied for jobs and to have interviewed.
What it shows: the constraint was not belief. It was a shared mistake about other people’s beliefs. Correcting the mistake changed behaviour, with no argument, no persuasion and no change to anybody’s actual values.
What it cannot show: that all norms work this way. Some are held up by genuine conviction, some by force, some by real material interest. This mechanism explains a subset, not the field. It also comes from one country and one class of men.
There is a name for the condition that study uncovered.
Pluralistic ignorance: a situation where most people privately reject a norm but each wrongly believes that most others accept it. Everyone then goes along with it, and by going along with it, supplies the evidence that convinces everyone else.
Why it matters here: it means a rule can be maintained at full strength by a population in which nobody supports it. There is no need for a villain. The enforcement is real and the belief behind it is a shared mistake.
Now put this beside the foot-binding mechanism from Chapter Three and they turn out to be the same object. In both cases the rule was held up not by conviction but by each person’s prediction about everybody else. In both cases the collapse, when it came, was sudden — because that kind of structure does not erode gradually. It stands until enough people see that others are ready to move, and then it goes over all at once.
6.4 — Whose society, exactly?
Everything above should make you suspicious of the word that appears in every version of this argument: society. Something bad will happen to society. Society will break down.
Everyone in this fight assumes society is a thing that can be harmed — a single body with a single set of interests, which can be well or unwell.
Both sides need this. The traditionalist needs it to say the rules protect it. The reformer needs it to say the rules damage it. Neither pauses to ask what would actually be measured.
Because societies are not bodies. They are large numbers of people with conflicting interests. Almost any change that harms some of them helps others. When a rule relaxes, some people gain freedom, some lose security, some lose status and some lose income, and there is no scale on which those cancel into a single verdict.
So “society will be harmed” nearly always turns out, on inspection, to mean one of three things. That some particular group will lose something. That a specific measure will move — births, marriages, crime — with the choice of measure smuggling in the conclusion. Or that the speaker will feel that things are not as they should be, which is a real experience and not a fact about a society.
The general form: a collective noun standing in for the interests of whoever is speaking. Whenever you meet the word in the next fifteen parts, replace it with “which people, and what happens to them,” and watch how many arguments simply stop working.
6.5 — My stake, and yours
It would be dishonest to list everybody else’s interests and leave out my own.
I am an Indian man writing about rules that have never once been applied to me. I have never had my clothing assessed, never had a rumour about my body used to settle an argument, never had my worth calculated by a marriage market. That is not a small gap. It means I am describing a weather system from indoors.
There is a specific temptation available to men who write about this, and I want to name it precisely so you can watch for it. It is the temptation to write about women’s freedom as a way of demonstrating to other men that one is modern and clever. That produces writing which sounds liberationist and is really about the author. It is a well-worn genre in India and I am fully capable of adding to it.
I also have a stake in the conclusion. I began this project irritated by what was done to a woman in a photograph, and irritation prefers certain findings. If the evidence in Parts Ten and Nine cuts against my instincts, the honest thing is to report it at full strength, and I have built the structure of this series to make that harder to avoid — the chapters that will be most uncomfortable for me have already been announced in the contents.
And you have a stake too. Notice, as you read the coming parts, which findings you find yourself hoping are true. That reaction is not a flaw to suppress; it is a signal. It marks the exact places where you are least reliable, and it is far more useful to you than any conclusion I could hand over.
Most of the disputed claims here are countable. They are not argued as though they were. When a countable question is fought like a war, something other than evidence is riding on the answer.
Real stakes exist on every side: families with daughters, families with sons, institutions whose authority rests on interpreting rules, political movements that recruit through threat, and a new one — an attention industry that profits from the fight continuing rather than from either answer.
A norm can be enforced at full strength by a population that privately rejects it. In the Saudi study, nearly nine in ten men approved of women working and nearly all of them thought they were unusual. Correcting the mistake changed real behaviour — with no argument at all.
This is the same machine as foot-binding. Rules held up by expectations about other people do not erode gradually. They stand, and then they fall over at once.
“Society will be harmed” almost always means one group loses something, or one chosen measure moves, or the speaker is unsettled. Replace the word with “which people, and what happens to them,” and a great many arguments stop working.
7How To Read The Next Fifteen Parts
The rest of this series is built on studies, surveys and national statistics. Numbers can be made to say almost anything to a reader who does not know the six ways they go wrong. Here are the six.
7.1 — Two things happening together
This is the first and largest problem, and it is responsible for more confident nonsense in this field than everything else combined.
When two things happen together, there are four possible explanations, and only one of them is the one people assume.
Take a real example we will meet in Part Ten. Women who have had more sexual partners before marriage have, on average, somewhat higher rates of later divorce. That relationship shows up repeatedly in large national datasets. It is not made up and I am not going to pretend it is.
Now, the four explanations.
One: the first thing causes the second. Something about having had more partners makes marriages less stable afterwards.
Two: the second causes the first. Impossible here, since the order is fixed. But in many other cases it is not, and the direction is assumed rather than shown.
Three: a third thing causes both. Perhaps people who are less inclined to stay in relationships have more partners and more divorces, and the two outcomes are both symptoms of the same underlying disposition. Nothing is causing anything; both are effects.
Four: it is an artefact. The measurement is bad, the sample is odd, or the effect is small enough to be noise.
Every one of these is live, and the honest reader holds all four until something rules some out.
Confounder: a third factor that causes the two things you are looking at, creating the appearance of a link between them.
The standard illustration: ice cream sales and drownings rise together. Ice cream does not drown people. Hot weather causes both. Heat is the confounder.
Why it matters here: nearly every claim about women’s behaviour and women’s outcomes has an obvious candidate confounder sitting right next to it — usually family income, education, religiosity or region, all of which shape both the behaviour and the outcome.
7.2 — Who ended up in the study
The second problem is subtler and it wrecks more research than the first.
Selection effect: a distortion caused by how people ended up in the group you are examining, rather than by anything the study measured.
Why it matters here: people are never randomly assigned to break social rules. The women who defy a norm differ from those who comply in a hundred ways before they defy anything — in temperament, in family, in money, in what they had to lose. Any difference in their later lives could be caused by those prior differences rather than by the defiance.
Here is the version that matters most for us. Chapter Five proposed comparing women who break a rule in India against women who do the same thing in a country with weak enforcement. That is a good idea and it has a serious flaw: the women who leave India are not a random sample of Indian women. They are richer, more educated, more mobile and differently supported. So a difference in outcomes might be caused by the move rather than the rule.
I flagged this in the box in Chapter Five and I am flagging it again here because it does not go away. It is the central weakness of the comparisons this series depends on, and I would rather you knew that before Part Thirteen than discovered it afterwards.
7.3 — How big is it, actually
Now the most useful skill in this chapter, and the one that will change how you read almost every headline about men and women for the rest of your life.
A finding can be completely real and completely trivial. “Men and women differ in X” tells you nothing until you know by how much. Researchers have a standard way of expressing this.
Effect size: a number describing how big a difference is, expressed in terms of how much the two groups spread out.
The plainest way to think about it. Take two groups. Pick one person at random from each. The effect size tells you how often the person from the higher group will actually be higher.
If the effect size is zero, it is a coin toss — fifty-fifty. Small differences get you to around fifty-six in a hundred. What researchers call a large difference gets you to about seventy-one in a hundred. Even a “large” difference means that nearly three times in ten, the individual from the supposedly lower group comes out on top.
Why it matters here: this single number is the difference between “men are taller than women”, which is true and enormous, and “men are better at maths”, which is a claim about a difference so small it is almost invisible at the level of any individual person.
That definition is abstract, so put it in a room with people in it. This is the picture I want you carrying through the rest of the series, because it is the one that keeps a true sentence from doing dishonest work.
Picture a school hall with a hundred men on one side and a hundred women on the other.
Height. One of the largest sex differences there is. Line everybody up and the pattern is obvious across the room. But even here, if you pick one man and one woman at random, the woman is taller roughly once every ten or twelve pairs. Even the biggest differences have overlap.
A typical cognitive difference. Most measured mental differences between men and women are so small that if you picked one person from each side, you would be right about which was “better” barely more often than by flipping a coin. Somewhere around fifty-three or fifty-four times in a hundred.
Both of those are real differences. Reporting them the same way — “men and women differ in X” — is how one true sentence gets used to smuggle in a completely different one.
Hold on to that hall, because Part Four takes you through every measured difference between men and women and puts each one in it. Some of them will be visible from the door. Most will not be visible at all.
7.4 — Whose data is it
In 2010 three researchers published a paper pointing out something that should have been noticed decades earlier. The overwhelming majority of published psychology studies had been run on people from Western, educated, industrialised, rich, democratic countries — and a very large share of those on university students in one country.
Those populations are a small minority of humanity. And on many measures they are not a representative minority; they are unusual, sometimes at the far end of the human range.
WEIRD samples: shorthand for Western, Educated, Industrialised, Rich, Democratic. The point of the acronym is that the people most studied by psychology are among the least typical humans available.
Why it matters here: a great deal of what this series has to draw on was measured on American and European undergraduates and is then applied to Indian women in towns with entirely different economies, family structures and law. Every time I do that, I will say so.
There is a second problem sitting on top of the first, and it is worth knowing about before you read anything built on psychology research.
In 2015 a large collaboration attempted to repeat one hundred published psychology studies, using the original methods and larger samples.
Fewer than four in ten produced the same result. The rest were weaker, absent, or went the other way.
What it shows: that a single published finding, even in a respected journal, is a weak object. Not worthless — but not a fact. What matters is whether a result has been repeated by other people, in other places, with different methods.
What it cannot show: which findings are the bad ones. It does not license dismissing any particular study you dislike. It raises the bar for everything equally, including results that flatter your position.
The rule I will follow: where a claim rests on one study, I will say so. Where it has been repeated across many countries and methods, I will say that too, and I will treat the difference as significant.
7.5 — What would actually settle it
Given all of the above, what counts as strong evidence here? Four things, in ascending order.
A single study. Weak. Interesting, worth reporting, not a foundation.
The same result across many societies. Much stronger, especially where the societies differ in the thing being proposed as the cause. If a difference appears in India and Sweden and Japan and Nigeria, “Indian culture causes it” is in trouble. If it appears in India and vanishes in Sweden, culture is back in the running.
Tracking the same people over years. Stronger again, because it establishes what came first. Most claims in this field are made from single snapshots, which cannot do that.
A sudden change in one place and not another. The strongest thing available outside a laboratory. When one country changes a law or a policy and a neighbouring one does not, you have something close to an experiment that no ethics committee would ever approve.
Natural experiment: a situation where something outside the researchers’ control divided people into groups in a way that resembles a real experiment. A law changing on a particular date. A border. A policy applied in one state and not the next.
Why it matters here: we cannot randomly assign women to societies. Natural experiments are the closest thing available, and Part Thirteen is built almost entirely from them.
One last thing to carry forward, and it is the most valuable habit in this chapter. When somebody makes a claim in this field, ask what would have to be true for them to be wrong, and whether they can say it. Anybody who cannot name the evidence that would change their mind is not holding a position. They are holding a flag.
I will state mine at the end of Chapter Nine.
When two things occur together there are four explanations, not one: the first causes the second, the second causes the first, a third thing causes both, or it is noise. Almost every claim in this field has a strong candidate for the third.
People are never randomly assigned to break social rules. Those who break them differ beforehand in ways that could explain everything afterwards. This is the central weakness of the comparisons this series relies on.
Effect size is the most useful number you can learn. It tells you how often a randomly picked person from one group actually beats a randomly picked person from the other. Even large differences leave nearly three in ten going the other way.
Most psychology was measured on unusual populations, and fewer than four in ten of a hundred well-known studies repeated successfully when retested. A single study is not a fact.
Ask what would make somebody wrong. If they cannot say, they are not holding a position — they are holding a flag.
8The Words Are Doing Work
Some words arrive with a verdict already inside them. Use the word and you have agreed to the verdict without noticing you agreed to anything. This chapter empties out the vocabulary before we start.
8.1 — A word that has already decided
Consider a woman wearing a short dress. Now consider the available descriptions.
She is dressed immodestly. She is dressed provocatively. She is dressed freely. She is dressed indecently. She is dressed confidently. She is dressed in a short dress.
Only the last of those is a description. Every other one contains a judgement, and in most cases a whole theory.
Immodest assumes modesty is the correct default and she has fallen short of it. Provocative assumes an audience being provoked, and quietly assigns her responsibility for their reaction. Indecent assumes a public standard she has violated. Free and confident assume she has escaped something, which is also a claim, and one she may not have made.
This is not a small point about politeness. Once a word is chosen, the argument is largely finished. Somebody who accepts the word provocative has already conceded that the reaction of onlookers is a property of her clothing rather than of the onlookers. Everything after that is decoration.
8.2 — Counting the words
Here is an exercise that takes two minutes and permanently changes how you read this subject.
Think of every word you know, in English or in your own language, for a woman who has many sexual partners. Do not censor. Just count them. Most people who try this get to a dozen or more without effort, and nearly all of the words are insults.
Now do the same for a man. Most people struggle past three or four, and the striking thing is not the number but the temperature. The male words are milder. Several of them are compliments wearing a thin disguise.
This asymmetry is not an accident of one language, and it is not new. Scholars of language have documented a long-running pattern in English in which words for women drift downwards over centuries into sexual insult, while their male counterparts hold their meaning. Master and mistress began as equivalents. Sir and madam began as equivalents. Bachelor and spinster began as equivalents. Look where each pair ended up.
The evidence is dictionary history. Words are traceable through dated written usage, so a word’s meaning at different points in time is a matter of record rather than of opinion.
What it shows: a consistent direction of drift, over centuries, in which paired terms for men and women diverge, with the female term acquiring sexual and then derogatory meanings while the male term stays neutral or gains status. It shows up repeatedly across the vocabulary.
What it cannot show: the cause. Drift is not proof of intention; nobody decided this. It also cannot easily distinguish a language reflecting a society’s judgements from a language helping to produce them. And it is best documented for English, which is a WEIRD language in a WEIRD society — though the same asymmetry is reported widely elsewhere, including in Hindi and Punjabi.
Here is why this matters for everything that follows. If your language supplies twelve ways to condemn a woman’s sexual history and three ways to mention a man’s, then any conversation you have will be shaped by that supply before you open your mouth. Precision costs effort in one direction and is free in the other. That is not neutral ground.
8.3 — “Westernisation”
This word deserves its own section, because it does a very specific job and does it invisibly.
When somebody says Indian women are becoming westernised, what is being claimed? Unpack it and there are at least four separate things bundled inside.
A description: some Indian women now behave in ways more common in Western countries. Sometimes true.
A causal claim: they do this because of Western influence, rather than for reasons of their own arising from Indian conditions — urbanisation, wage work, education, delayed marriage. This is an assertion, rarely defended, and it is often false. A woman taking a job in Bengaluru because her family needs the income is not imitating anybody.
A judgement: that this is a decline. The word carries it. Nobody says a village became westernised when it got clean drinking water.
A claim about ownership: that the behaviour is foreign, and therefore that the people doing it are, in some measure, no longer properly ours. This is the part that does the real work and it is never stated openly.
The fourth is where the damage sits, because it converts a disagreement about behaviour into a question of belonging. A woman who is behaving differently can be argued with. A woman who has become foreign can be expelled.
Part Three will show that the historical claim underneath the word is mostly backwards — that a great deal of what is now defended as Indian tradition arrived with the British in the nineteenth century, and that some of what is now attacked as Western would have been unremarkable in India a thousand years ago. I am not going to make that argument here. I am only pointing out that the word has already assumed the answer.
8.4 — And the words on the other side
I would be doing exactly what this chapter warns against if I only examined the vocabulary of one camp.
Empowerment. Enormously flexible and therefore nearly meaningless. It is applied to a woman becoming a legislator and to a woman buying a face cream. When a word can describe both, it is not describing anything; it is signalling approval. It is also the favourite word of advertising departments, which should tell you something about what it costs to use.
Choice. Often used to end a conversation rather than to have one. Saying a woman chose something is true and frequently uninformative. Everyone chooses within constraints, and the interesting question — what the constraints were, and who set them — is precisely what the word is used to close off.
Patriarchy. A real term describing a real pattern, and Part Fifteen takes it seriously. But it is often used to explain everything, and a theory that explains every outcome forbids none, which makes it unfalsifiable. It also has a specific weakness this series will have to confront: it does not obviously describe a system in which the people at the very bottom — by imprisonment, by workplace death, by suicide — are overwhelmingly male.
Toxic, problematic, regressive. Each converts a disagreement into a diagnosis. That is exactly what immodest does, running the other way. A judgement dressed as a description is the same move regardless of which camp is making it.
Look at the entire vocabulary — both camps, every word above — and one thing is common to all of it. Female behaviour is the variable. Male behaviour is the weather.
A dress is described as provocative. The provocation is treated as an automatic fact about men, like rain. Nobody says a man is reactive, because his reaction is not understood as a choice he is making.
Look at what gets counted. There are hundreds of studies on the effects of women’s sexual behaviour on their later lives, and a small fraction as many on men’s. Both camps do this. The traditionalist studies what women’s freedom costs. The reformer studies what constraint costs women. Both have made women the object under examination.
Even the standard progressive sentence, “we must teach men not to harass,” treats male behaviour as an input to be adjusted from outside rather than as conduct chosen by an adult.
The general form: one party’s conduct is treated as behaviour and the other’s as conditions. Whichever side of this argument you are on, you have almost certainly been doing it. I have been doing it in this document — this series is called The Rules About Women, and the title was chosen before I noticed what it assumed.
8.5 — What I will do in this series
Some rules for myself, so you can check them.
I will describe before I judge. “A short dress” rather than “immodest” or “liberated”. Where I use a loaded word because there is no alternative, I will mark it.
I will name the word choice as a choice. If I write honour killing, I will note that the phrase contains the killers’ own justification and that some researchers refuse it for that reason. If I write tradition, I will say how old the thing actually is, because the word implies an antiquity that is often absent.
I will use both camps’ words for their own positions, not the insulting versions their opponents prefer.
I will not use “we” to mean Indians, or Hindus, or men, or the modern. That word quietly assigns the reader to a team and I have no right to do that.
Words like immodest, provocative and indecent are not descriptions. Each contains a verdict, and accepting the word concedes the argument before it begins.
Every language examined has far more words for a promiscuous woman than for a promiscuous man, and the female ones are harsher. Paired terms drift apart over centuries — master and mistress, bachelor and spinster. The supply of vocabulary shapes the conversation before anyone speaks.
Westernisation bundles four claims: a description, a causal story that is usually asserted rather than shown, a judgement of decline, and a claim about belonging. The last is the one doing the work, because a woman who is merely behaving differently can be argued with, while a woman who has become foreign can be expelled.
The other camp’s words do the same job. Empowerment describes both a legislator and a face cream. Choice is often used to close a conversation about constraints.
Underneath every word on both sides sits one assumption: female behaviour is the variable and male behaviour is the weather. Even the title of this series assumes it.
9The Shape Of The Fight
Before we spend fifteen parts gathering evidence, it is worth knowing what each side would have to establish in order to win. Written down plainly, the requirements are startling — mostly because so few people arguing have ever met them.
9.1 — What the traditionalist has to prove
Take the position at its strongest, in the form set out in Chapter Two. Not the version that shouts. The version that would convince a careful reader.
To establish that a rule about women should be maintained, the following would need to be shown. Not asserted. Shown.
That the rule solves a real problem. Not that it once did. That the problem exists now. If the problem was uncertain fatherhood, paternity testing exists. If it was disease, contraception and antibiotics exist. A rule justified by a solved problem needs a new justification.
That the harm from removing it is larger than the harm from keeping it. This requires actually counting the second one. Almost nobody does. The costs of enforcement — the surveillance, the constrained lives, the violence, the women who do not work — are usually left out of the ledger entirely, and a comparison with one column empty is not a comparison.
That no cheaper instrument exists. Most problems have several solutions. If a cheaper one is available that does not require restricting half the population, the rule has to explain itself.
That the burden belongs where it is placed. Every problem here has two participants. The rules land on one. That asymmetry needs an argument and almost never receives one.
That is the honest requirement, and it is heavy. It is also not impossible. For some rules, in some places, I expect the case can be made, and where the evidence supports it I will say so.
9.2 — What the reformer has to prove
Now the other side, held to exactly the same standard.
That the harms are genuinely imposed rather than intrinsic. This is the test from Chapter Five and it is hard work, not a slogan. It requires showing that the damage stops at the boundary of the enforcing community.
That the replacement has been built. This is the lesson of the emptied asylums, and it is the one reformers most consistently skip. Removing a rule while leaving in place the conditions that made it necessary is not liberation. If the old system provided security — badly, unequally, at enormous cost, but provided it — then something has to provide it afterwards, and “she will be free” is not an answer to “who will support her.”
That the predicted costs did not materialise, or were worth paying. Not that they were imaginary. Some were not imaginary. Birth rates really did fall. Marriage really did become less common. The honest reformist position has to either dispute those numbers, or accept them and argue the trade was worth it, and the second is a much harder and more respectable argument than the first.
That women’s own stated preferences are being taken seriously, including when they are inconvenient. Large numbers of women in India endorse the rules under examination. Explaining that away as false consciousness is available, and it is the same move as saying they do not know their own minds — which is precisely what the other side is accused of.
9.3 — Where the burden actually sits
Who has to prove what first? This is not a technicality. In practice it decides most arguments, because whoever holds the burden loses ties.
The traditionalist answer: the rule is in place, it has lasted, and the person proposing change carries the burden. This follows from Chesterton’s fence and it is reasonable.
The reformist answer: the rule restricts a person’s life without her consent, and restriction is what requires justification. Long-standing restriction is still restriction. This is also reasonable.
Notice that these are not competing claims about the world. They are competing claims about the default, and there is no neutral place to stand between them. Your choice of default is already a moral position, and it usually determines your conclusion before the evidence arrives.
What I will do is state mine, so that you can discount it. I hold that a rule which constrains a person’s life requires justification, and that its age reduces the strength of that requirement but does not remove it. That is a value, not a finding, and I am not going to pretend I derived it from evidence.
9.4 — The four questions
Everything in the remaining fifteen parts serves four questions. Here they are, with where each is answered.
One. Where did these rules come from, and what problem did they solve? Parts Two and Three. Paternity, property, farming, and the specifically Indian machinery of caste. Including the finding that a great deal of what is now called ancient Indian modesty is roughly as old as the railway.
Two. How different are men and women, really? Parts Four and Five. Every measured difference with its actual size, and the result that most damages the simple cultural explanation — that in the freest and most equal countries on earth, men’s and women’s choices diverge more, not less.
Three. What actually happens when the rules change? Parts Nine, Ten, Eleven, Twelve, Thirteen and Fourteen. What happens to a woman, what happens to a country, what the costs of the old system are when somebody finally counts them, and what happened in every society that already ran the experiment.
Four. What has the equality project actually produced? Parts Six, Seven, Eight and Fifteen. Sixty years of results, measured rather than asserted, including the reversal in education that nobody discusses and the finding that measured female happiness in the West declined across exactly the decades women gained most.
Time to turn this on ourselves, as promised.
Every question above, and the question that started this series, takes the same form: what happens to a woman who does X? The traditionalist asks it expecting damage. The reformer asks it expecting none. Both agree, without discussion, that this is the question.
But look at what the form does. It makes a woman’s life an outcome to be predicted. It makes her behaviour a variable to be assessed for consequences. It positions everybody else — including me, including you — as the analyst standing outside, examining.
There is a different question available that almost nobody asks: who is asking, and what do they get from each answer? That question was Chapter Six, and notice that it is the only chapter in this part that turns the instrument around.
The general form: a question so universally accepted that its shape becomes invisible. Fifteen further parts of careful evidence will not fix this. The most I can do is say it out loud: this series is fifteen parts of examining women, produced by a man, and that fact is not neutralised by the quality of the examination.
9.5 — What would change my mind
Chapter Seven said that anybody who cannot name the evidence that would change their mind is holding a flag rather than a position. So here is mine, in advance, before I have gathered the evidence.
I currently expect that most of the harm attributed to women breaking these rules is imposed by enforcement rather than intrinsic to the acts — that it mostly does not travel across a border.
I would abandon that if the harms turn out to appear at similar strength in societies with weak enforcement, controlling for who moves and who stays. If women in low-enforcement countries show the same later damage, my position is wrong and the elders are describing something real about the world rather than something real about villages.
I currently expect that the fertility collapse is genuine, serious, and not something reformers have a good answer to.
I would abandon that if some country demonstrated a sustained recovery to replacement without reversing women’s freedoms. As far as I know, none has. If one does, that changes the picture substantially.
I currently expect that some sex differences are large, stable across cultures, and not explicable by socialisation alone.
I would abandon that if those differences shrank as societies became more equal. They do not. In several cases they widen, and that finding is not going to be pleasant for anybody. It gets a full part.
To win, the traditionalist must show the problem still exists, that removing the rule costs more than keeping it — counting both columns — that no cheaper instrument exists, and that the burden belongs on her rather than on him.
To win, the reformer must show the harms are imposed rather than intrinsic, that the replacement has been built, and either that the predicted costs did not appear or that they were worth paying. Some of them appeared.
Both sides are really arguing about the default, and there is no neutral ground between them. Mine is stated: constraint requires justification, and age weakens that requirement without removing it.
I have written down in advance what would change my mind on three claims. Hold me to it.
And note what every question in this series has in common: it asks what happens to a woman who acts. The shape of that question is not neutral, and fifteen parts of careful evidence will not make it so.
10An Honest List Of What We Do Not Know
Every part of this series ends this way. Two lists, and no hedging in either direction. What is genuinely unknown, with the reason it is unknown. And what is solid enough to build on.
10.1 — Genuinely unknown
This part has been about how to think rather than what is true, so the unknowns here are mostly about the tools themselves. Later parts will have longer lists and harder ones.
How much of the harm is imposed and how much is intrinsic. This is the central question of the series and I do not know the answer. I have stated my expectation and my grounds for abandoning it. The reason it is unknown is not laziness — it is that the clean experiment is impossible. You cannot randomly assign women to societies, and the ones who move are not like the ones who stay. Every estimate will carry that flaw.
Whether the vocabulary shapes the belief or merely records it. Chapter Eight showed that words for women drift downward across centuries. Whether language is producing that judgement or reporting it is not settled, and the honest position is that the evidence cannot currently separate the two.
What norms hold up any specific rule. Chapter Six showed a norm sustained entirely by mistaken beliefs about other people. But it does not follow that all norms are like this. Some rest on real conviction, some on force, some on material interest. For any particular rule in India, which mechanism dominates is an open empirical question, and one that could be answered by exactly the kind of study run in Saudi Arabia. As far as I know, it has not been run here at scale.
How much of the research base survives scrutiny. Fewer than four in ten of a hundred well-known psychology findings replicated. That figure is itself from one project. Which specific results in Parts Four to Twelve are the fragile ones cannot be known in advance, and anyone who tells you they can identify them — always conveniently the ones cutting against their own position — is guessing.
Whether the mechanism that ended foot-binding can be generalised. It is the most hopeful finding in this part, and it may be a special case. Foot-binding had unusual features: it was visible, binary, and imposed on children by parents who could coordinate. Rules about adult behaviour are none of those things. Whether the pledge-society mechanism transfers is unknown.
10.2 — Solid
That the three claims are separable. This is not an empirical finding but a logical one, and it holds regardless of what the evidence turns out to show. Origin, consequence and obligation are different kinds of statement requiring different kinds of support. Nothing in the next fifteen parts can overturn this, and it is the most useful thing in this document.
That no set of facts produces an instruction by itself. Also logical, also settled, also universally violated. The extra ingredient is always a value. It can be a good value. It cannot be hidden.
That longevity is evidence of enforcement rather than of benefit. This is close to definitional once stated. A rule imposed by a strong group on a weak one has a survival advantage independent of its merit. Foot-binding is the proof by example: a thousand years of survival, and it broke the feet of every girl it touched for the entire duration.
That men and women report sexual histories that cannot both be true. This is arithmetic, and it holds across many countries and many surveys. In a closed population the averages must be nearly equal. They are not, consistently, everywhere. Something is wrong with the reporting, reliably, in a patterned way — which is itself a finding about what each sex is punished for admitting.
That norms can be held up by nobody’s actual preference. The Saudi study is a genuine experiment with a real behavioural outcome, not a survey correlation. Its scope is limited — one country, one class of men — but within that scope the finding is solid, and it establishes that this mechanism exists.
That the harm to a woman accused in public is real and close to permanent, whatever the truth of the accusation. This needs no study. It follows from how information moves and from the documented asymmetry between the reach of a claim and the reach of its correction.
10.3 — Why the gaps are shaped this way
One last observation, and it is the kind that matters more than the lists.
Look at what is missing. We have a great deal of research on what happens to women who break rules, and very little on what happens to women who keep them. We have extensive study of the costs of change and thin study of the costs of continuity. We can tell you the divorce rate associated with a woman’s sexual history and we struggle to tell you the life satisfaction of a woman married at nineteen to a man chosen for her.
That imbalance is not an accident of funding. It reflects which questions somebody wanted answered. The literature was built by people who found one direction interesting, and the shape of the gap is a record of who was doing the asking.
Chapter Six said to notice who is counting and what they get. This is what that looks like when it shows up in a library. The missing research is itself a finding, and I will point at it every time it appears.
Genuinely unknown: how much of the harm is imposed against intrinsic — the central question, and the clean experiment is impossible. Whether language produces judgements or records them. Which mechanism holds up any particular Indian rule. Which specific findings in the coming parts will survive scrutiny. Whether the foot-binding mechanism generalises.
Solid: that origin, consequence and obligation are separable claims. That no facts produce an instruction without an added value. That longevity measures enforcement, not benefit. That reported sexual histories cannot arithmetically be true. That norms can stand with nobody supporting them. That a public accusation is a near-permanent cost regardless of its truth.
And the shape of the gaps is itself evidence. We know a great deal about what happens to women who break the rules and very little about what happens to women who keep them. That asymmetry records who was asking.
Sources & further reading — Part 1
When These Tools Were Named
This part is about thinking rather than about events, so its timeline is a timeline of instruments — the point at which each idea used in these ten chapters was given a name and became usable.
| Year | What was named, and by whom |
|---|---|
| c. 1000 CE | Foot-binding becomes established in China. It will last roughly a thousand years, defended throughout by every argument still in use today, and then collapse within a few decades. |
| 1739 | David Hume points out that writers slide from describing how the world is to instructing what people ought to do, without ever showing the step. The gap is still unclosed. |
| 1829 | Sati is outlawed in British India after campaigning by Indian and foreign reformers together. The strongest argument against abolition was the danger of outsiders touching a people’s customs. |
| 1903 | G. E. Moore names the naturalistic fallacy. In everyday use it merges with the older appeal to nature: the claim that natural means good. |
| 1929 | G. K. Chesterton describes the fence in the field. Do not remove what you cannot explain. It becomes the most durable conservative argument of the century. |
| 1948 | Robert Merton names the self-fulfilling prophecy: a prediction that manufactures its own truth. Chapter Five is built on it. |
| 1960s–1980s | Britain and the United States close their mental institutions. The replacement is promised and largely never built. Prisons become, by default, the largest mental health facilities in both countries. |
| 1975 | Linguists document the long downward drift of English words for women — master and mistress, bachelor and spinster — into sexual and then derogatory meaning, while the male terms hold. |
| 2010 | Three researchers point out that most of psychology was measured on Western, educated, industrialised, rich, democratic populations, and that these are among the least typical humans available. |
| 2015 | A large collaboration retests one hundred published psychology studies. Fewer than four in ten hold up. Every finding in the coming parts must be read against this. |
| 2020 | Economists publish the Saudi study: nearly nine in ten men privately approved of women working, and nearly all of them believed they were unusual. Correcting the mistake changed real behaviour. |
| Today | A woman stands in front of a police vehicle. Days later, a rumour about her body is offered as a reply. |
| ↓ | Part Two starts here — and runs the clock backwards, ten thousand years, to the moment somebody first owned a field and needed to know who would inherit it. |
Glossary
Every hard word used in this part, in plain English. You should not have needed this page — each term was explained where it first appeared — but it is here.
| Term | Plain meaning |
|---|---|
| Ad hominem | Latin for “to the person”. Answering an argument by talking about the person who made it. Legitimate only when the claim cannot be checked without trusting them. |
| Appeal to nature | Arguing that natural means good, or unnatural means bad. Cholera is natural; literacy is not. |
| Chesterton’s Fence | The principle that you should not remove something whose purpose you do not understand until you understand it. A rule against a certain kind of confidence, not against change. |
| Confounder | A third factor causing two things at once, making them look connected. Hot weather causes both ice cream sales and drownings. |
| Counterfactual | What would have happened otherwise. The comparison nobody ever gets to see, which is why patterns caused by punishment look exactly like patterns caused by nature. |
| Deinstitutionalisation | The closing of large mental hospitals from the 1960s onwards. Used here as the clearest case of a fence removed without building its replacement. |
| Effect size | A number saying how big a difference between two groups actually is. Read as: pick one person from each group at random — how often does the higher group win? |
| Foot-binding | The Chinese practice of breaking and binding girls’ feet, lasting about a thousand years and collapsing within decades once families coordinated to stop. |
| Genetic fallacy | Judging a claim by where it came from rather than whether it is true. Nothing to do with genes; from the Greek for origin. |
| Is/ought gap | The point that no pile of facts produces an instruction by itself. Getting from one to the other always requires an added value. |
| Moralistic fallacy | Concluding a fact must be false because it would be inconvenient. The mirror image of the appeal to nature, and just as common. |
| Natural experiment | A situation where something outside the researchers’ control split people into groups — a law changing, a border, a policy in one state and not the next. |
| Naturalistic fallacy | Strictly, defining goodness as a natural property. In ordinary use, merged with the appeal to nature. |
| Norm | An unwritten rule enforced by ordinary people through approval and disapproval, rather than by law. Harder to argue with because there is no office to appeal to. |
| Pluralistic ignorance | When most people privately reject a rule but each believes the others accept it — so everyone complies, and the compliance becomes the evidence. |
| Primogeniture | The rule that the eldest son inherits everything. Solved a real problem — land being divided into unviable slivers — at the cost of four children in five. |
| Replication | Repeating a study to see whether the result appears again. Fewer than four in ten of a hundred famous psychology findings did. |
| Sati | Widow-burning, outlawed in India in 1829. Defended at the time as scripture, tradition and devotion. |
| Selection effect | A distortion caused by how people ended up in the group being studied, rather than by anything the study measured. |
| Self-fulfilling prophecy | A prediction that causes what it predicts. False when spoken and true when finished. It always looks like wisdom afterwards. |
| WEIRD samples | Western, Educated, Industrialised, Rich, Democratic. The populations psychology mostly studied, and among the least typical people available. |