Why I Wrote This
This book argues one side. It chooses like a prosecutor and checks like an auditor. Every claim carries a grade, and Chapter Six is about the accusations that are false.
There are two ways to write an angry book. One of them works.
The Indian National Congress held power at the centre of India for about fifty-four of the seventy-nine years since independence. No other party comes close. For long stretches it also ran most of the states at the same time. That means something simple and quite hard to sit with: for most of the life of independent India, if something was decided, Congress decided it.
That is not an accusation. It is arithmetic. But it does mean that a serious account of what has gone wrong in India has to spend most of its time looking at what Congress did, because for most of the period there was nobody else in the room.
Books about this already exist. Most of them are useless, and they are useless in one of two ways.
The first kind is written by people who admire the party. These books explain. Every failure becomes a difficult situation. Every disaster becomes a lesson learned. The Emergency becomes a period. Nobody decided anything; things simply happened, and then the country moved on. You finish the book knowing more dates and understanding nothing.
The second kind is written by people who hate the party. These are worse, and I want to be precise about why. They are worse because they mix true things with invented things, and they do it in the same paragraph, in the same tone, with the same confidence. So a reader who checks one claim and finds it made up throws the whole book away. And the book deserved to be thrown away. But the true parts of it did not.
That is the problem this series is built to solve. The record of what Congress governments actually did is bad enough. It does not need help. Every made-up story added to it is not a bonus. It is a subtraction, because it hands the party an easy escape from all the real ones.
So this is a book of one-sided selection and honest evidence. Those two things are not in conflict, and confusing them is the mistake almost everyone makes.
Think about how a court works. The prosecutor is not neutral. The prosecutor’s whole job is to build the strongest case against the accused, and nobody thinks this is dishonest. What would be dishonest is inventing a witness. Choosing what to put in front of the court is the job. Making things up ends the career.
I am doing the first thing. I am not doing the second.
You should know where I stand, because it could pull the writing. I am an Indian. I am not neutral about this material and I am not going to pretend to be. I think the record I am about to lay out is a serious one and that most Indians have never seen it collected in one place in language they can actually read. That is why I am writing it.
But an author who wants a particular answer is an author who will cut corners to reach it. So I have built the corners into the structure, where I cannot quietly skip them. Every serious charge in this series gets a grade telling you how solid the evidence is. Every argument gets the other side stated at its strongest, in its own words, before it loses. Chapter Nine of this part is a list of what would prove me wrong. And one full chapter here — Chapter Six — is given over to accusations against Congress that are false, because I would rather kill those myself than have a reader find them and stop believing me.
One more promise, and I want it on the record early. This series is about Congress. Part Eighteen compares them with the government that has held power since 2014, and that comparison is measured on the same ruler in both directions. A second series will audit that government the way this one audits Congress, using the same method and the same grades. If I never write it, you are entitled to conclude something about me.
How to read this book
The Six Boxes
Coloured boxes run through every part of this series. Each colour does one job. You will meet all six in the next few pages, so here they are first, with a real example of each rather than a description.
A Word Box appears the first time a hard word turns up. It explains the word right there, on the spot. You never have to hold a term you do not understand and hope it gets explained later.
Exchequer: the government’s money. All of it, in one pot — what it collects in taxes and what it spends. When you read that something “cost the exchequer” a sum, it means the government either lost that money or never collected it.
Why it matters here: most of the big scandals in this series are arguments about whether money was lost from this pot, and if so, how much.
An In Real Terms box takes a number too big to picture and turns it into something with a body. A number you cannot picture is decoration. It sits on the page looking serious and tells you nothing.
Congress has been in power at the centre for roughly fifty-four of seventy-nine years. Put it another way. If independent India were a person of seventy-nine, they would have had the same employer for fifty-four years, three other employers for the rest, and would have spent about two-thirds of their working life under one management.
A How We Actually Know This box shows you the evidence. Not the conclusion — the actual thing. The register, the audit, the court file. And then it tells you what that evidence cannot show, because every source is blind to something.
Much of what we know about the Emergency of 1975 to 1977 comes from one source: a commission of inquiry led by Justice J.C. Shah, a former Chief Justice of India, appointed in 1977. It held around a hundred sittings and went through roughly forty-eight thousand papers. It reported in three volumes, the last on 6 August 1978.
What it cannot show: it questioned officials and ministers who knew they were being questioned. People being investigated shade their answers. The Commission recorded what it was told, under oath, by people with reasons to shape it.
An Argument box appears where serious people genuinely disagree. Each side gets its strongest case, in its own words, not a weak version I built so I could knock it over. Then a verdict tells you where things actually stand.
You are holding a book that says on its own cover that it argues one side. Should you trust anything in it?
An account that has decided its answer will find its answer. It will include the damning file and skip the ordinary one. Even if every individual fact checks out, the shape of the thing is a lie, because the shape is made of what was left out. You cannot audit a book for absences.
Every account selects. The pretence of neutrality just hides the selecting. A book that announces its side lets you correct for it, the way you correct for a defence lawyer. And the alternative on offer is not a neutral book. It is a book with the same bias, undeclared.
What would settle it: nothing settles it in general. It is settled case by case, by checking. This book is built to be checkable and that is the only defence it has.
Why people care so much: because in Indian politics “biased” has become the standard way of not answering a charge. Calling a source biased is cheaper than showing the source is wrong.
A Hidden Assumption box is the signature of this series and the hardest thing in it to write. It is not a warning or a note about limits. It digs out a belief sitting underneath an argument that both sides accept without discussing it. Those beliefs do most of the work and nobody examines them.
Everybody in this argument — the people defending Congress and the people attacking it — agrees on one thing without ever saying it out loud. They agree that “Congress” is one continuous thing.
It is not obviously one thing. The party of 1952 was a mass movement that had just won independence, run by people who had been to prison for it. The party of 1975 suspended the constitution and jailed its opponents. The party of 2014 was a family office with a national footprint. They share a name, a symbol, and for most of that stretch a family. Beyond that, the connection is a matter of argument, not fact.
Watch how convenient the assumption is for everyone. It lets me carry every sin across eight decades and pile them on today’s party. It lets today’s party disown any particular sin as a different era. Both moves use the same unexamined premise, in opposite directions.
I have made a choice here, and you should see it as a choice. I am treating Congress as one continuous actor, because the family, the culture and many of the same instincts do carry across. But it is a choice, and where it does real work in an argument, I will say so.
Finally, every chapter ends with a Remember This box. It is the chapter in the plainest words I can manage. If you read only these boxes across all nineteen parts, you would still have the whole argument.
Congress governed India for most of its independent life. A serious account of what went wrong has to look mostly at them, because for most of the time there was nobody else deciding.
This book selects like a prosecutor and checks like an auditor. One-sided choosing is legitimate. Invented evidence is not, and it is also stupid, because it rescues the accused.
The true record is bad enough. Every made-up story added to it is a subtraction, not a bonus.
1What This Book Is, and What It Is Not
There is a way to attack a government that works, and a way that feels better and achieves nothing. Almost everybody picks the second one.
1.1 — The prosecutor and the liar
Imagine two people who both want the same man convicted.
The first is a prosecutor. She reads every file. She picks the six documents that damage him most and leaves the forty that are boring. She questions the witnesses who saw the worst of it. She does not call the witnesses who saw him being pleasant. When the defence produces something awkward for her, she deals with it, because she knows the judge will read it anyway.
The second is a liar. He cannot be bothered with files. He tells a story he heard. He adds a detail because it makes the story land better. He is not worried about the defence, because he has never really imagined the defence as a thing that will happen.
Both of them are one-sided. Only one of them is doing something legitimate, and here is the thing that matters most: only one of them is effective.
The prosecutor wins cases. The liar loses them, and worse, he loses them in a way that helps the accused. Every invented detail that gets exposed makes the true details look invented too. By the end, the man walks out and the real evidence walks out with him.
Indian political argument runs almost entirely on the second model. You have seen it yourself. Somebody forwards you a list of twenty terrible things a party did. You check one out of curiosity. It turns out to be nonsense. You now believe none of the other nineteen, and four of them were true and provable.
Prosecution brief: the written case a prosecutor builds against an accused person. It gathers the evidence that supports the charge and organises it into an argument.
It is not supposed to be balanced. Balance is the court’s job, once both sides have made their strongest case. What a brief is supposed to be is true in every particular, because the other side will check.
Why it matters here: this book is a prosecution brief. That tells you what to expect from it and what you are entitled to demand of it.
So the rule for these nineteen parts is short. I choose what goes in. I do not get to invent what goes in.
1.2 — What “one-sided” is allowed to mean
Being one-sided is allowed to mean three things in this series.
It is allowed to mean that I am looking for failures. When Congress did something well — and they did, more than once, and those moments are in this book — I note it and move on. I do not build a chapter around it. A prosecutor does not spend the afternoon on the accused’s charity work.
It is allowed to mean that the questions are hostile. “Why did this happen?” is a neutral question. “Who decided this, what did they know when they decided it, and what happened to them afterwards?” is a hostile one. I ask the hostile one. It is a better question and it gets better answers.
It is allowed to mean that I give the reader my view at the end of an argument, when the evidence supports a view.
Being one-sided is not allowed to mean four other things, and these are the ones that destroy books.
It cannot mean stating something I have not checked. It cannot mean quoting somebody without the date, the place and the actual words. It cannot mean presenting an allegation as a finding. And it cannot mean building a weak version of the other side so that I can knock it down.
That last one is worth pausing on, because it is the instruction people find hardest to accept. If you cannot make the other side sound reasonable, you do not yet understand the other side. You are arguing with a puppet you made. And a reader who has actually heard the real argument will spot the puppet immediately and stop reading.
Think of it as building a wall you want to be unclimbable. Every true brick makes it higher. Every false brick makes a hole, and a hole is not a small problem in a wall — it is the only part anyone will use.
Nineteen parts of this series will contain roughly two hundred separate charges. If one hundred and ninety-nine are solid and one is invented, the conversation about this book will be about the one. It always is.
1.3 — The other side of this book
This series is about Congress. It stops there on purpose, and you are entitled to be suspicious of that.
Here is the suspicion, stated fairly: a book that audits one party and not the other is not an audit. It is campaign material with footnotes. The author gets the reputation of a serious researcher and delivers the product of a partisan.
That suspicion is reasonable and I am going to answer it with structure rather than with a promise, because promises are cheap.
Part Eighteen of this series compares the Congress record with the record of the government that has held power since 2014. That comparison uses the same measuring stick in both directions. Where the present government did something Congress could have done and did not, it says so plainly. Where the present government carried on a Congress policy under a new name, it says that too. Where it has not moved at all on something, that goes in as well. If Part Eighteen reads like an advertisement, it will have destroyed the seventeen parts in front of it, and I will have wasted a year.
A second series will then audit the government of 2014 onwards the way this one audits Congress, with the same grades and the same rules.
Everything in this book could be true and it could still be propaganda, if the choice of target is doing the political work. So: is it?
Publishing nineteen parts against one party while the other waits its turn is not a neutral scheduling decision. It lands during that other party’s time in office and it serves them. “I will get to them later” is what everybody says, and later is a place nothing arrives. The author gets full credit now for a balance he has only promised.
Fifty-four years of central power against about twelve produces vastly more material, and material that has finished being contested. Court cases have closed. Commissions have reported. Files have been declassified. Auditing a government still in office is a different job with different evidence, and mixing the two would produce a worse book about both.
What would settle it: the second series existing. Nothing else. Not this paragraph, not my intentions, not the fact that I have written this box.
Why people care so much: because in India the demand “what about the other one?” is now the standard reply to any criticism of any party, by everyone, in every direction. It works because it is often a fair point and always an easy one.
1.4 — What the nineteen parts cover
A short map, so you know what you have agreed to read.
Parts Two and Three take the first years, from 1947 to 1959 — partition, the princely states, Kashmir, and the early changes to the constitution. Part Four takes the economic system that was built between 1950 and 1991 and never really examined, which is the single most expensive item in this series and has no scandal attached to it at all. Part Five is about which Indian states were made rich and which were made poor by decisions taken in Delhi.
Parts Six and Seven cover China, Tibet, the war of 1962, and the wars and the governing of Kashmir. Part Eight covers Indira Gandhi’s rise. Part Nine is the Emergency. Part Ten is Punjab from 1978 to 1993, and Part Eleven is November 1984.
Part Twelve is Rajiv Gandhi’s five years. Part Thirteen covers the parts of India these accounts always skip — the North East, the tribal belt, the corridor where the Naxal insurgency runs. Part Fourteen is 1991 to 1996: economic reform and the demolition of the Babri Masjid, in the same five years, under the same Prime Minister.
Parts Fifteen and Sixteen are the ten years from 2004 to 2014. Part Seventeen collects what was said — the statements, with dates and full wording. Part Eighteen is the comparison. Part Nineteen is the honest final accounting, including everything I could not prove.
The raw material for this series is mostly boring and mostly official. Reports of the Comptroller and Auditor General. Reports of commissions of inquiry. Judgments of the Supreme Court and the High Courts. Debates recorded in Parliament. Census and survey tables. Replies given to questions in Parliament, which are useful precisely because they are grudging.
Then a second layer: memoirs by people who were in the room, and contemporary journalism.
What this mix is bad at: it is heavily weighted towards what the state wrote down. The experience of people who were not officials survives thinly, and much of it was collected decades later, from memory. That gap is not neutral. It runs in a consistent direction, and Chapter Four is about it.
That is the whole apparatus. It is not glamorous and it is not secret. Almost everything in the nineteen parts ahead was sitting in a published document that anybody could have read, which raises a question worth carrying with you: if the record is this available, why does almost nobody know what is in it?
Part of the answer is that it is written in a way designed to be unreadable. That is the problem this whole project exists to solve.
There are two ways to attack a government. The prosecutor selects hard and checks everything. The liar selects hard and checks nothing. Only the first one wins.
One-sided selection is legitimate and this book does it openly. One-sided evidence is not, and it is also self-defeating, because a single exposed invention discredits every true charge standing next to it.
The other side of the ledger is promised in Part Eighteen and in a second series. Judge that promise when it comes due, not now.
2What Counts as a Blunder
The word gets used for six different things. Keeping them separate is most of the work, and almost nobody does it.
2.1 — Six things called by one name
Somebody says “Congress made a blunder”. What have they actually said? Six quite different things hide under that one word.
One. A decision that was wrong on the information available at the time. Somebody in the room said so. The warning is on the file. It was ignored. This is the strongest kind of charge and it is rarer than you would think.
Two. A decision that was reasonable at the time and turned out badly. This happens to every government that has ever existed. It is not a blunder. Calling it one is the fastest way to look unserious.
Three. A cost that built up slowly with nobody deciding anything. No meeting, no file, no villain. A rule stays in place for forty years and quietly moves a large amount of money from one part of the country to another. These are usually the most expensive items and they get the least attention, because there is nobody to be angry at.
Four. A promise that was made and not kept. Not a mistake — a broken commitment. Different thing, different test. You need the promise in writing and the outcome measured.
Five. A crime. Somebody took money, or ordered something illegal, or covered something up. This is not an error of judgment and should never be filed as one.
Six. A thing that was done on purpose, worked exactly as intended, and was monstrous. This is the category people find hardest, because the language of “blunder” and “mistake” quietly does the accused a favour. Some of what is in this series belongs here.
Culpable: deserving blame. In everyday use it means you could have known better and should have.
The test has two halves and both must be met. Could a reasonable person in that chair, with the papers that were actually on that desk, have seen the problem? And did they have the power to act on it? If either answer is no, it is a misfortune, not a fault.
Why it matters here: nearly every argument about Indian history is really an argument about these two questions, conducted without either side naming them.
2.2 — Wrong at the time, wrong only later
This distinction decides more arguments than any other in this series, so here is a concrete way to hold it.
In 1950 India chose an economic model built around state planning, public ownership of heavy industry, and government permission for private businesses to expand. Part Four takes this apart in detail. For now, the question is only how to judge it.
The defence is straightforward. In 1950, this was not a strange choice. It was close to the international consensus. The Soviet Union appeared to be industrialising at extraordinary speed. Britain had just nationalised coal, rail and steel. Newly independent countries everywhere were doing versions of the same thing. And India’s own experience of an unregulated market was two centuries of colonial extraction ending in a famine that killed millions. The people making these choices were not fools and they were not thieves.
That defence is genuinely strong and I am not going to pretend otherwise. But watch what it covers and what it does not.
It covers 1950. Perhaps it covers 1955. It does not cover 1975, by which point the results were in, the comparisons were available, and the people who benefited from the system had grown powerful enough to defend it. Somewhere in that stretch, a defensible choice became an indefensible refusal to look at the evidence.
The interesting question is never “was it wrong”. It is “when did it become wrong, and who was in the room when the warnings arrived”.
From roughly 1951 to 1991, an Indian company generally needed government permission to start, to expand, to change what it made, or to import machinery. India grew slowly for four decades. Was this a mistake, or bad luck?
The information available in 1950 pointed this way. Every serious economist of the period, in India and outside it, thought a poor agricultural country needed a planned push into heavy industry, because private capital was too thin and too timid to build steel plants. The bet did produce real things: dams, steel, engineering, universities, a technical workforce. It was tried honestly and it underperformed. That is not a blunder, that is a wager.
The bet may have been fair in 1950. By 1965 the evidence was arriving and it was ignored. Other countries starting from similar poverty were pulling ahead on a different model, in plain sight. Indian officials wrote critical assessments internally. The response was not correction but extension — more controls, more nationalisation, tighter permissions. A wager you refuse to settle for forty years is no longer a wager.
What would settle it: in principle, the internal record — who wrote what warning, when, and what happened to them afterwards. A good deal of this is available. Part Four uses it.
Why people care so much: because this is the argument about whether India’s poverty in 1990 was inherited from the British or manufactured after 1947. Both answers assign a very large debt to somebody.
2.3 — The cost with no scandal attached
Here is the strangest fact about writing this series.
The most expensive things Congress governments did were not the scandals. They were rules. Ordinary, technical, dull rules, passed without controversy, defended by nobody in particular, left running for decades.
A scandal has a shape people can hold. There is a person, a sum of money, and a moment. A rule has none of that. It has a file number and a slow effect, and by the time the effect is visible the person who signed it has retired.
Part Five of this series is built entirely around one such rule. Between 1952 and 1993 the central government subsidised the transport of coal, iron ore, steel and cement, so that a factory anywhere in India could buy them at roughly the same price as a factory sitting next to the mine. It sounded fair. It was called equalisation. What it actually did was take the one advantage the mineral-rich states had and hand it to everybody else. Eastern India sat on the resources and watched the industry that should have grown around them get built two thousand kilometres away, for forty-one years.
No minister took a bribe. No commission reported. There is no photograph.
Picture a village with the only well. The village council announces that from now on, water will be delivered to every village in the district at the same price, and the cost of the lorries will be shared by all.
It sounds like fairness. It is fairness, in a way. But the village with the well had exactly one thing going for it, and that thing has just been converted into a public utility. Nobody stole anything. Nobody meant any harm. Forty years later that village is the poorest in the district and nobody can point to the day it happened.
2.4 — A crime is not a mistake, and cruelty is not either
Language does quiet work here and I want to name it, because I will be watching my own words for the rest of this series.
When a newspaper writes that “riots broke out”, it has said something specific without appearing to. Riots that break out have no organiser. They are weather. If it later turns out that lists of addresses were distributed and transport was arranged, the word “riot” has already done its job of hiding that.
“Excesses” is another. It suggests somebody doing a legitimate thing slightly too enthusiastically. When it is applied to men being sterilised against their will to meet a district target, it is not a description. It is a defence, smuggled in as a noun.
I will use plain words for these things. Where the plain word is contested, I will say that it is contested and tell you who contests it and why. What I will not do is pick the soft word and pretend I have made no choice.
Both sides of almost every argument in this book share a belief that neither side states: that the government was trying to do the thing it said it was trying to do.
Watch how it works. The critic says the poverty programme failed to reduce poverty. The defender says poverty is hard to reduce. They are arguing furiously, and they have both quietly agreed that the purpose of the programme was to reduce poverty.
Suppose it was not. Suppose the purpose was to be seen announcing a poverty programme, and the money moving through it to particular hands was not a leak in the system but a feature of it. Then “it failed” is the wrong verdict. It succeeded. You were measuring the wrong thing, and so was the person arguing with you.
I am not claiming this is always true. I am claiming that neither side ever tests it, because the test is uncomfortable for both. The critic loses the pleasure of calling them incompetent. The defender loses the excuse that the problem was hard.
Throughout this series, when a programme fails for thirty years and is renewed every year, I will ask which of the two readings fits better. Sometimes the answer is genuinely incompetence. Sometimes it is not.
Keep the six categories from the start of this chapter somewhere you can reach them. Almost every argument you will ever have about Indian politics is two people using the same word for two different things and getting angrier as they go.
Six different things get called a blunder: wrong at the time; reasonable but unlucky; a slow cost nobody decided; a broken promise; a crime; and a cruelty that worked as intended. Mixing them up is how arguments go nowhere.
The real question is almost never “was it wrong”. It is when did it become wrong, and who was in the room when the warnings came in.
The most expensive decisions in this series have no villain, no bribe and no photograph. They are rules that ran for forty years.
3The Problem With “They Should Have Done Otherwise“
Every charge in this book contains a hidden claim about a world that never happened. That claim needs evidence too.
3.1 — You cannot run history twice
Every accusation against a government has two halves. People argue about the first half and forget the second one exists.
The first half is what happened. That part can be checked. Files, figures, judgments, records.
The second half is the claim that something better was available. When you say a government blundered, you are saying there was another road, that it could have been seen from where they stood, and that taking it would have produced a better result. Three separate claims, none of them checkable in the ordinary way, all of them smuggled in with the word “blunder”.
This is not a reason to give up. It is a reason to be explicit. A charge that names its alternative is a serious charge. A charge that does not is just disappointment with extra confidence.
Counterfactual: a statement about what would have happened if things had gone differently. “If the train had been on time, I would have made the meeting.”
You use these constantly and you are usually right, because you know how trains and meetings work. They get shaky when the system is large and the gap in time is long. “If India had opened its economy in 1965, it would be as rich as South Korea” is a counterfactual about sixty years and a billion people, and nobody can run the experiment.
Why it matters here: it is hiding inside every sentence in this book that contains the word “should”.
3.2 — What a real alternative needs
For an alternative to count in this series, it has to pass three tests. If it fails any of them, I have to soften the charge, and I will say so.
Test one: somebody actually proposed it, at the time. Not afterwards. A road that only became visible in 1995 was not available in 1962, and blaming people for missing it is a cheap way to feel clever. If a named person wrote it down at the time and was overruled, the charge is real. Much of the time, somebody did.
Test two: it was politically possible. A Prime Minister with a small majority and a hostile coalition partner cannot do everything a Prime Minister with four hundred seats can. This cuts both ways, and it is worth noticing which way. Congress spent long stretches with enormous majorities. “We could not have done it” is a much weaker defence when you hold three-quarters of Parliament, and I will hold them to that.
Test three: somebody comparable did it and it worked. Not a rich country. Not a small one. A place with roughly similar problems at roughly the same time. This is the strongest kind of evidence available for a road not taken, and it is why comparison matters so much in the chapters ahead.
3.3 — The four honest comparisons
There are four ways to test an alternative in this series without pretending to run history twice. Each has a weakness. Used together they are strong.
Other countries at the same starting line. In 1950, a number of countries were about as poor as India and had similar problems. Where they are now is evidence about what was possible. The weakness is that no two countries are alike, and someone can always name a difference. Watch for whether the difference named actually explains the gap, or is just a difference.
Other Indian states. This is much stronger, and it is badly under-used. Indian states shared a constitution, a currency, a central bank, a foreign policy and a legal system. When two of them diverge sharply, the explanation has to be something local, and the list of candidates is short. Part Five and the state ledger that runs through this series live here.
Before and after inside one place. A rule changes on a date. What happened to the same population afterwards? The weakness is that other things also changed. The strength is that the population is the same people.
What the government itself predicted. This is the most underrated of the four and it is nearly unanswerable. When a government announces a programme it usually states a target, and those statements are in the record. You are then not comparing reality with your opinion. You are comparing reality with their own stated expectation. Nobody can complain that the standard is unfair, because they set it.
The fourth method leaves a particularly good paper trail in India, because of how the system works. A programme is announced with a target. Money is allocated in a budget document. The Comptroller and Auditor General later checks what was spent against what was allocated. Parliamentary committees take evidence and publish reports. Ministers answer written questions with figures.
None of these documents was created to embarrass anybody. That is exactly why they are useful. A record kept for the machine’s own purposes has no stake in the argument you are having with it forty years later.
What it cannot show: money spent is not the same as work done. A road can be paid for and never built. The audit tells you the cheque cleared, not that anyone drove on it. Where I can find a physical check on the outcome, I use it, and where I cannot, I say so.
3.4 — The trap that catches everyone
The trap is hindsight, and it is worth understanding properly because it is not what people think it is.
Hindsight is not knowing how things turned out. That is unavoidable. Hindsight is forgetting that you know, and then feeling that the outcome was obvious at the time. It converts a genuinely difficult decision into an idiotic one, and it makes the person judging feel considerably cleverer than they are.
The test I use throughout this series is simple. Before I criticise a decision, I try to write down what was on the desk. What was known, what was rumoured, what was unknown, and who was saying what. If the case still holds once the desk is reconstructed, the case is real. Several charges I expected to make did not survive that test, and Chapter Nine lists them.
But there is a mirror version of this trap and it is used far more often, so guard against it in the other direction too. It is the claim that nobody could have known. That claim is checkable, and it is frequently false. Somebody usually did know. The interesting question is what happened to them.
Underneath every “they should have done X” sits a belief that neither side examines: that there was a menu.
The critic and the defender both picture a room where options were laid out and the wrong one was chosen. They disagree about which option was right. They agree completely that choosing was happening.
Sometimes it was not. Governments inherit commitments, treaties, staff, budget lines and voters, and much of what looks like a decision is a thing that had already effectively been decided by a chain of earlier commitments nobody now remembers making. The meeting happens. Papers are circulated. But the room has one live option in it and everyone present can feel that.
This cuts against me more often than it cuts for me, so I want it stated in my own book rather than left for someone else to point out. Where I can show a live choice — a warning on the file, a minister overruled, a note recording an alternative — the charge is strong. Where I cannot, I am describing a current, not a decision, and I will use the weaker word.
The uncomfortable follow-up, which I will return to in Part Nineteen: if a great deal of government is current rather than choice, then a very large part of what every country argues about is theatre. That is not a comfortable conclusion for a nineteen-part book of blame.
So the standard for the rest of this series is set here, and it is deliberately hard to meet. Before I call something a blunder, I owe you the road not taken, the person who pointed at it, and some evidence that walking down it was possible. Where I have all three, the charge is heavy. Where I have two, it is a criticism. Where I have one, it is an observation, and I will use the smaller word.
Every accusation contains a hidden claim about a world that never happened. A charge that names its alternative is serious. A charge that does not is disappointment with confidence.
An alternative counts here only if somebody proposed it at the time, it was politically possible, and somebody comparable did it. Congress held huge majorities for long stretches, which makes “we could not have” a weak defence in their case.
The strongest test in this book is the government’s own stated target. Nobody can call the standard unfair when they set it themselves.
4Where the Evidence Comes From
Six kinds of source hold up this series. Each is good at one thing and blind to another. The blindness is the part worth learning.
4.1 — The six sources
Anybody can assert. What separates a book you can use from a book you cannot is whether it tells you where each thing came from. So here is the whole toolbox, in order of how much weight I put on it.
Audit reports. India has an office whose only job is to check what the government did with money. It reports to Parliament, not to the government. Its reports are dry, numbered and specific.
Comptroller and Auditor General, usually shortened to CAG: the constitutional auditor of India. Its job is to examine government spending and report to Parliament on whether the money went where it was supposed to go, and whether the rules were followed.
The important part is who it answers to. The CAG is not part of the government it audits and cannot easily be removed by it. That independence is exactly why its reports have started so many political fights.
Why it matters here: several of the largest charges in this series began as a paragraph in a CAG report, and Chapter Seven is about how easily those paragraphs get misread.
Commissions of inquiry. When something goes badly wrong, a government may appoint a judge to investigate it. The judge takes evidence, questions officials and publishes findings.
Court judgments. The strongest single kind of evidence in this book, because a court has heard both sides, tested the evidence, and taken responsibility for a conclusion in public.
Parliamentary records. Debates, questions and answers, committee reports. Grudging, precise, and produced under an obligation not to lie.
Official statistics and files. Censuses, surveys, budget papers, and material released years later under the right-to-information law or through declassification.
Memoirs and journalism. The people who were in the room, and the people who were outside the door at the time. The most vivid and the least reliable.
4.2 — What each one is blind to
Now the part that most books skip.
An audit is blind to whether the work was done. It follows money. If the money left correctly and arrived correctly, the audit is satisfied. Whether the road exists is a different question, and the CAG usually cannot answer it.
A commission of inquiry is blind to whatever it was not asked about. Its terms of reference are written by the government that appointed it. This is the quietest and most effective way to control an investigation: you do not have to interfere with a judge if you have already decided the question. Watch the terms of reference before you read the findings, always.
A court is blind to everything outside the specific charge. A criminal court asks one question: did the prosecution prove this offence against this person beyond reasonable doubt? It does not ask whether the policy was a disgrace. An acquittal answers the narrow question and is routinely reported as answering the wide one.
Parliamentary answers are blind by design. A minister answers what was asked, in the narrowest reading available. The skill of reading them is noticing the question that was not asked, or was asked and slid past.
Statistics are blind to what nobody counted. This is the largest blindness in the book and Chapter Seven returns to it. Whole categories of Indian suffering have no number, not because they are small but because no office was ever given the job of counting them.
Memoirs are blind to the author. Everybody is the reasonable person in their own account. Memoirs are best used for a specific and limited purpose: what was known in the room, and when. They are almost worthless for who was to blame.
Think of six people describing a house fire.
The accountant can tell you exactly what the insurance paid, and nothing about the flames. The judge can tell you whether one named person lit it, and nothing about the wiring. The auditor can tell you the fire extinguisher was purchased, but not whether it worked. The clerk can tell you what the fire brigade was asked, but not what it was not asked. The statistician can tell you how many houses burned in districts where somebody was counting. And the neighbour, who saw everything, was also in a dispute with the owner about a wall.
You do not pick one. You put all six together and you watch where they disagree, because the disagreement is usually where something is buried.
4.3 — The commission that was recalled
One episode teaches more about Indian official evidence than any general explanation, so here it is in full.
In 1977, after the Emergency ended and the government that imposed it lost power, the incoming government appointed a Commission of Inquiry under Justice J.C. Shah, a former Chief Justice of India, to examine what had been done during those twenty-one months.
The Commission worked. It held around a hundred sittings. It examined roughly forty-eight thousand papers. It produced three volumes running to about five hundred and twenty-five pages. The first interim report, in March 1978, covered how the Emergency was declared and how the press was silenced. The second covered police action, including the demolitions and firing at Turkman Gate in Delhi. The final report, on 6 August 1978, covered prisons, torture, and the forced sterilisation programme.
Its central finding was not vague. The laws allowing detention without trial had not been misapplied by accident. They had been used deliberately against political opponents. The Commission recorded that decisions to arrest and release people were taken on political grounds to favour the ruling party, and that using the police that way is a certain route to destroying the rule of law. It also found that most senior civil servants carried out orders they themselves believed were improper.
Then two things happened. Special courts were set up in 1979 to try Emergency cases, but the government that created them fell in July of that year. When the previous Prime Minister returned to power in January 1980, the Supreme Court found those special courts had not been properly constituted. No trials followed.
And the report itself began to disappear. Copies were withdrawn. It vanished from the shops where official publications were sold. For years afterwards, researchers looking for it were told that no complete copy survived in India, and that the surviving sets were in a library in Australia and another in London. A former Member of Parliament, Era Sezhiyan, eventually reassembled and republished the full text from his own collection, under a title that tells you the whole story: Shah Commission Report: Lost and Regained.
The journalist Kuldip Nayar, who had checked the archive while the earlier government was still in office and been assured the records were intact, later found the National Archives reporting that it held nothing — although the Commission had stated on its final day that it was depositing everything there.
The physical evidence for that last paragraph is unusual, and worth naming precisely. It is an absence. There is a body that said it deposited records, an archive that says it received none, and a set of published volumes that stopped being obtainable in the country that produced them.
Three independent things point the same way here: the recollection of a journalist who checked before and after, the experience of researchers who could not find copies, and the fact that surviving sets turned up in foreign libraries rather than Indian ones.
What it cannot show: an absence does not name the hand. No document orders the destruction, and I have not seen one. What can be said with confidence is that the report became unobtainable in India after a particular government returned to power, and that this was not an accident of shelving.
4.4 — When the missing record is the finding
That episode establishes a rule I will use for the rest of this series, and it is one of the most useful ideas in this whole book.
When a government had the power to count something and chose not to, the absence is a finding, not a gap.
States count what they intend to manage. A census, a register, a survey — each one exists because somebody decided the number mattered. So when a state that counts crop yields by district and cattle by village has no figure at all for how many people died in a particular week, that is not an oversight. Somebody decided.
You will meet this repeatedly. Deaths in a police operation with no tally. Sterilisations recorded as totals with no record of consent. A displacement with no register of the displaced. In each case, ask a question the official history does not: who would have been embarrassed by the number, and did they have the power to prevent it existing?
Everyone who argues about Indian history — every side, every time — assumes something about the documents: that a record is a record of what happened.
It is not. A record is a record of what an institution wrote down about itself, for its own purposes, at the time. Those are different things and the gap between them is not random.
Consider what leaves a trace. Money leaves a trace, because money has to be accounted for. Orders leave a trace, because subordinates want cover. Anything the machine had to do in order to keep functioning leaves a trace.
Now consider what does not. A conversation. A verbal instruction. A thing everybody understood without being told. A death in a place with no registrar. In one case in this series, we know how many people arrived at a refugee camp mainly because arrivals had to be fed, and feeding required a count. The record of their survival exists as a by-product of a ration ledger.
So the archive is not a neutral window. It is systematically better lit on the side of the state, and systematically dark on the side of everyone the state was doing things to. Both the defenders and the critics quote it as though it were a window, and both are wrong in the same direction — because the material that survives is the material the institution produced about itself.
This weighs against my own book too, and I would rather say it here than have it said to me. Where my evidence is entirely the state’s account of its own conduct, I will mark it, and you should discount it accordingly.
One question hangs over all of this, and it is worth settling before the series starts leaning on commission reports in part after part.
India has appointed a great many judge-led inquiries into disasters, killings and scandals. Very few have led to anybody being punished. So what are they worth?
Their real function is to absorb public anger and convert it into waiting. A commission is appointed while feeling is high, reports years later when nobody is watching, and its findings carry no force at all — a government may simply decline to act on them, and routinely does. Several have reported after the officials they criticised had already retired with full honours. Judged by what changed afterwards, they are a machine for the passage of time.
That confuses two different things. A commission can fail completely as a remedy and still succeed as a record. It has powers no journalist has: it summons officials, takes evidence on oath, and compels the production of files. That nobody was punished is a finding about the political system, not about the quality of the document. Half the solid material in this series exists only because somebody with the power to compel asked a question.
What would settle it: comparing what commissions concluded with what later court proceedings and released files established, across many cases. Where both exist, they agree more often than the cynical view predicts.
Why people care so much: because appointing an inquiry is the cheapest thing a government under pressure can do, and everybody involved knows it. The demand for one and the granting of one are frequently the same negotiation.
The working rule for this series, then. A commission finding is strong evidence of what happened and weak evidence of what mattered, because the second question was settled when the terms of reference were drafted.
Six sources: audits, commissions, courts, parliament, statistics, memoirs. Each is good at one thing and blind to another, and the blindness is the part that matters.
A commission is controlled not by pressuring the judge but by writing the questions. Read the terms of reference before the findings, every time.
When a state that counts everything has no count of something, somebody decided that. The missing number is a finding, not a gap.
5The Grades
Every serious charge in nineteen parts carries a label saying how solid it is. Here is the label, and why it is the most important thing in the book.
5.1 — Six levels
The single most common failure in political writing is treating all accusations as the same weight. A conviction and a rumour appear in the same paragraph, in the same font, joined by the word “and”.
So every load-bearing claim in this series carries one of six grades, printed in the text where you can see it.
| Grade | What it means | How much weight |
|---|---|---|
| Convicted | A court tried somebody for this and convicted them. | Highest available. |
| Court-found | A court made the finding, but no conviction followed, or the matter is still live. | Very high, with the caveat stated. |
| Commission-found | A judge-led inquiry found it. The commission is named. | High, subject to its terms of reference. |
| Audited | The CAG, a parliamentary committee or a comparable body flagged it. Never tested in court. | Moderate. Establishes a problem, not a crime. |
| Alleged | Publicly alleged, never tested, no finding either way. | Low. Reported as an allegation and nothing more. |
| Fabricated | Positively shown to be false. | None. Goes in Chapter Six, not the charge sheet. |
Two rules govern the grades and I will hold to both.
First, the grade appears next to the claim, not hidden in an appendix nobody reads. If a thing is only alleged, you will see the word “alleged” at the moment you meet it.
Second, the grade can move, and when it moves the whole journey goes in. Charges do not sit still. A thing can be audited in one year, prosecuted in the next, and thrown out in the year after that. Reporting only the stage that suits me would be exactly the dishonesty this book is built to avoid.
Acquittal: a court’s decision that the prosecution did not prove its case against a particular person, to the standard criminal law requires.
Notice what it does not say. It does not say nothing happened. It does not say the policy was fine. It says: this specific charge, against this specific person, was not proved to this specific standard. A great deal of confusion in Indian public argument comes from treating the narrow verdict as a wide one.
Why it matters here: the largest financial scandal in this series ends in an acquittal, and both sides describe that acquittal wrongly.
5.2 — The worked example: one allocation, five verdicts
In 2008 the government of the day allocated licences and radio spectrum for mobile telephone services. Spectrum is the invisible thing your phone uses to talk to a tower. There is a limited amount of it, the government owns it, and it is worth an enormous amount of money.
The licences were given out on a first-come, first-served basis at prices set years earlier, rather than being auctioned. What followed is the single best illustration in modern India of why grades are necessary, because the same set of facts has now produced five different institutional answers.
In November 2010, the CAG estimated the loss to the exchequer at about ₹1.76 lakh crore. This is the number everybody remembers and it is an Audited figure, not a judicial one.
The investigating agency itself disagreed with that number. When the CBI filed its charges, it put the loss at about ₹30,984 crore — roughly a sixth of the auditor’s figure. Same event, same files, two official arms of the same state, and a gap of more than one lakh crore rupees between them. That gap is not fraud on anybody’s part. It is a difference of method, and Chapter Seven explains what the two methods were.
In February 2012, the Supreme Court cancelled all 122 licences that had been issued. That is a Court-found outcome and it is a heavy one. A court does not cancel a hundred and twenty-two licences because the paperwork was untidy.
In December 2017, a special court acquitted all seventeen accused, including the former telecommunications minister. That is also Court-found, and it points the other way.
In March 2024, after six years and around thirty hearings, the Delhi High Court granted the CBI permission to appeal against those acquittals, saying that there were contradictions in the trial court’s judgment that required deeper examination and a fresh look at the whole body of evidence. The matter is live.
And separately, when one of the companies involved asked for its licence fee back on the grounds that its promoters had been acquitted, the Supreme Court refused. A bench presided over by Justice D.Y. Chandrachud held that the acquittal did not entitle the company to a refund, and used the words fraud and sham transactions about how the licence had been obtained.
Both political camps in India describe the 2017 acquittal, and both descriptions are wrong. Here are the two serious versions.
A criminal court heard the case in full and found the prosecution had not proved it. That is what courts are for. Everything else — the auditor’s estimate, the newspaper coverage, the political campaign built on it — was noise around a case that collapsed when it was finally tested by someone whose job is testing it.
An acquittal answers one narrow question: was this offence proved against these people beyond reasonable doubt. It is silent on whether the allocation was a disgrace. And the surrounding record does not point the same way: the Supreme Court cancelled every licence, a later bench used the word fraud about how one was obtained, and a High Court has said the acquittal judgment contains contradictions needing deeper examination.
What would settle it: the appeal. It was admitted in 2024 and has not been decided.
Why people care so much: because ₹1.76 lakh crore became the number that ended a government. If it was never a real loss, one side has an enormous grievance. If it was, the other side does. Neither camp can accept a verdict that leaves them with only half of what they want.
Five institutional answers, one set of facts. Now notice what a partisan writer does with this. He picks the one that suits him and stops. The party’s critics quote the auditor’s figure and the licence cancellation, and go quiet after 2017. The party’s defenders quote the acquittal, and go quiet before 2012 and after 2024.
You are going to get all five, every time, in this series. Not because I am being fair-minded, but because the full arc is more damaging than any single piece of it. It shows a system in which an allocation can be judicially cancelled in its entirety and nobody is responsible for it.
Six grades: convicted, court-found, commission-found, audited, alleged, fabricated. Every serious claim in this series carries one, printed where you can see it.
Charges move up and down over the years, and the whole journey goes in. Quoting only the stage that suits the writer is the standard trick and it is the one this book is built to refuse.
An acquittal answers a narrow question, not a wide one. In the spectrum case, the policy is condemned by the record and no individual is proved guilty. Both sentences are true. Almost nobody quotes both.
6What Was Made Up
A chapter of a book attacking Congress, spent demolishing attacks on Congress. It is the most useful chapter here, and by the end you will see why.
6.1 — Why an angry book needs this chapter
You have been sent these stories. Everyone in India has. They arrive as forwarded messages, as videos with dramatic music, as confident paragraphs under a photograph.
Most people sorting them do it by feel. If a story fits what you already believe, it goes in the true pile. That method has an obvious flaw, and there is a better one, which is the whole point of this chapter: check the ones you want to be true first. Those are the ones your judgment is worst at.
Here is the argument for spending a chapter on this in a book whose purpose is to prosecute.
The false stories about Congress are, right now, the most effective protection Congress has. Not their spokesmen. Not their history. The lies told about them.
Every time somebody repeats an invented story and it gets exposed, the party gets a free win. They get to stand up, correct the record, and say — accurately — that their opponents make things up. And the person watching learns a lesson that is much broader than the single story: that accusations against this party are the sort of thing that turns out to be false.
The Emergency was real. November 1984 was real. The forced sterilisations were real. Those things do not need a single invented story standing next to them, and every invented story standing next to them does them damage.
6.2 — The case that was invented
Start with the most serious, because it is the one that circulates hardest.
There is a widely forwarded claim that Rahul Gandhi faced a rape case. It refers to a real legal proceeding, which is why it survives. The proceeding was a fabrication and a court said so in unusually strong terms.
A petition was filed by Kishore Samrite, at the time a former Member of the Legislative Assembly from a rival party, alleging that a young woman had been confined and assaulted. The Allahabad High Court dismissed it. The petitioner then took the matter to the Supreme Court, complaining about the manner of the dismissal.
In 2012, a Supreme Court bench of Justices B.S. Chauhan and Swatanter Kumar went through the entire body of evidence rather than disposing of it on procedure. It concluded that the petition was based on falsehood and driven by malice and political vendetta, and that the allegation had no substance and not an iota of evidence behind it. The court imposed a cost of ₹10 lakh on the petitioner. Half was directed to Gandhi. The other half was directed to the family of the young woman named in the petition, whose reputation the case had also damaged. Investigators had been unable to locate that family at all.
Grade: Fabricated. Not unproven. Not doubtful. Judicially demolished, with costs, by a bench that read the whole file specifically in order to say so.
The evidence here is a reported Supreme Court judgment. That is close to the strongest kind of document available for a claim like this, and it is strong for a particular reason: the court did not have to do it.
The easy route was to dismiss the appeal on a technicality. Instead the bench went into the merits, examined the material, and recorded findings about the petition’s origin and motive. Courts do that when they want something on the record.
A second thing worth noticing. The costs were split, and half went to the family of the woman named. A court awarding money to somebody who was not even a party, on the grounds that a false case damaged her too, is telling you what it thought of the petition.
What it cannot show: a court can find a case unproved and malicious. It cannot prove a negative about the world. The finding here is as strong as a finding of falsity gets, and I am stating it as exactly that.
One further trap, and it catches careful people. That name is a very common one in India. Searching it returns unrelated criminal proceedings involving entirely different private individuals who happen to share it. Those results then get screenshotted and circulated as though they were about the politician. If you are checking a claim about a well-known name, check that the document in front of you is about the person you think it is about.
6.3 — The surname that was invented
The second story is older and has spread further. It holds that Indira Gandhi’s husband Feroze was really a Muslim named Feroze Khan, and that the family disguised itself behind a borrowed surname.
The record here is documentary and it is dull, which is what makes it useful. He was born Feroze Jehangir Ghandy on 12 September 1912, at the Tehmulji Nariman Hospital in Bombay. His father, Jehangir Faredoon Ghandy, was a marine engineer working for the firm Killick Nixon. His mother was Ratimai, born Commissariat. The family were Parsis who had moved to Bombay from Bharuch in Gujarat. He was not related to Mohandas Gandhi. He changed the spelling of his own surname from Ghandy to Gandhi during the freedom movement, in tribute. When he died in 1960 he was cremated, and his ashes were placed in the Parsi cemetery in Allahabad.
Grade: Fabricated.
It is worth understanding why this particular invention has such a long life. It does something no true charge can do. It converts a political argument into a question of secret identity, and it offers the listener the pleasure of having seen through a disguise. Real evidence cannot compete with that feeling, which is precisely why the feeling should make you suspicious.
Parsi: a member of a small Indian community descended from Zoroastrians who came from Persia, mostly over a thousand years ago, and settled largely in Gujarat and Bombay.
Zoroastrianism is its own religion, older than both Christianity and Islam, and it is not a branch of either. Parsi names often sound Persian, which is where the confusion in this story starts and, for some who repeat it, where it conveniently stays.
Why it matters here: the entire force of the Feroze Khan story depends on the listener not knowing this one fact about a community of a few tens of thousands of people.
6.4 — The contested one
Not everything in this category is a clean fake. Some claims are real events wrapped in a disputed description, and those need a different treatment.
In 2019 the Prime Minister said that Rajiv Gandhi had used the aircraft carrier INS Viraat as a personal transport for a family holiday, citing a magazine article from 1988.
The trip happened. What it was is contested, and the contest is unusually well documented, because the officers involved are alive and spoke.
Vice Admiral Vinod Pasricha, who was commanding the ship at the time, said the claim was wrong. Admiral L. Ramdas, then the Southern Naval Commander and later Chief of the Naval Staff, said the Prime Minister of the day was travelling on official business, to chair a meeting of the Islands Development Authority, and that no foreigners were aboard and no parties were held. Wajahat Habibullah, who was the administrator of the islands then, supported that account. One other retired officer, Commander V.K. Jaitly, publicly took the opposite view.
Grade: Alleged, with the official record against it. The voyage is not in dispute. The characterisation is, and the weight of first-hand testimony runs against the version that circulates.
This one goes in the book, in Part Twelve, with both sides exactly as stated. It will not carry weight, and it should not, because there is a great deal about that decade that is documented far better.
This chapter has just restated three false or contested claims in order to knock them down. There is a real case that doing so is a mistake.
People remember claims and forget corrections. Weeks later the reader retains “there was something about a case” and has lost the part where a court called it a fabrication. Repetition is what gives a false claim its power, and a debunking is still a repetition, delivered to a new audience with the author’s credibility attached.
The claims are already everywhere. The reader has already met them. Silence does not un-send a forwarded message; it just leaves the reader with the false version and no correction. And a book that quietly avoids the famous accusations looks like it is dodging them, which costs it more than the repetition does.
What would settle it: knowing whether readers of this chapter later repeat the claims or the corrections. That is measurable in principle and I have not measured it.
Why people care so much: because everyone arguing about political misinformation is really arguing about whether ordinary people can be trusted with a correction. Both sides here are polite about it. Neither says it out loud.
6.5 — Why the lies are the party’s best asset
Put the three cases side by side and the pattern is visible.
Each of the fabricated stories is about identity or private life. None is about a policy, a file, a budget line or a decision. That is not a coincidence. Invented stories go where evidence cannot follow — into secret parentage, private conduct, hidden motive — because those are the places where nobody can produce the document that ends the argument.
Which means the invented stories are competing with the real ones for the same limited attention. Every hour spent on a surname is an hour not spent on the Shah Commission. And the party knows this perfectly well. Answering a fabrication is the easiest day’s work a spokesman ever has.
So the discipline of this chapter is not decency. It is strategy. If you want a charge to stick, you have to hand your opponent nothing to knock down. Every false brick is a hole, and the hole is the only part anyone will use.
Three widely forwarded stories, checked: the rape case was judicially demolished with costs in 2012; the Feroze Khan surname story is documentably false; the warship trip is real but its description is contested, with the first-hand testimony against the popular version.
Fabrications cluster around identity and private life, never around files and budgets, because those are the places no document can settle.
The lies told about Congress are the best protection Congress has. Each one gives them an easy correction to make and teaches the listener that charges against them turn out to be false. Check the stories you want to be true first.
Check the stories you want to be true first. That is where your judgment is worst.
7How Numbers Lie
Nobody in this argument invents figures. They do not have to. There are three easier ways, and all three are legal.
7.1 — Lakh, crore, and the size of a human head
Start with something that sounds too basic to matter and is in fact the root of the whole problem.
A human being cannot picture a large number. Not “finds it difficult” — cannot. Past a few thousand, the picture in your head stops changing. A crowd of ten thousand and a crowd of ten lakh produce the same mental image, which is roughly “a lot of people”.
This is not stupidity and it is not a lack of education. It is how the equipment works. And every large figure in political argument is aimed straight at this weakness.
Lakh: one hundred thousand. Written 1,00,000. Crore: ten million, or one hundred lakh. Written 1,00,00,000.
So a lakh crore is one lakh multiplied by one crore. That is a one followed by twelve zeros. In the international system it is called a trillion.
Why it matters here: the difference between “crore” and “lakh crore” is a factor of one hundred thousand, and in speech it is one extra word. Political argument in India runs on figures whose scale most listeners cannot hold, delivered at speed by people who know that.
The defence against this is mechanical, and you can apply it to any figure you meet, in this book or anywhere else. Divide it by the number of people it concerns. A national total means nothing to anybody. The same total per person means something instantly, because you already know what money is worth to a person.
Take the most famous number in this book: ₹1.76 lakh crore, the auditor’s estimate in the spectrum case.
India had roughly 121 crore people at the time. Divide one by the other and you get about ₹1,450 for every man, woman and child in the country. A family of five, ₹7,250.
Now the number has a body. You can picture what ₹7,250 was to a household in 2010 — a month’s earnings for a great many of them. That is what “1.76 lakh crore” means, and it is why the figure ended a government.
Hold on to that conversion, because the next section is about whether the ₹1,450 was ever actually in anybody’s pocket.
7.2 — Money that was lost, and money that was never made
Here is the first of the three legal ways to mislead with a figure.
In the spectrum case, two arms of the same state produced two numbers about the same event. The auditor said about ₹1.76 lakh crore. The investigating agency, filing actual criminal charges, said about ₹30,984 crore. That is a gap of more than one lakh crore rupees, and neither body was lying.
They were answering different questions.
The auditor asked: what could this have raised, if it had been sold the way such things are sold when the state is trying to get the best price? To answer that, you need a benchmark, and the benchmark available was what comparable spectrum fetched later, when it was auctioned. Apply that price backwards across everything that was given away cheaply, and you get an enormous figure.
The agency asked a narrower question: what specific sums can we trace, tie to specific transactions, and prove in a courtroom? That produces a much smaller number, because most of the auditor’s figure is not a sum that moved. It is the gap between what happened and what might have happened.
Presumptive loss, also called notional loss: not money that was taken, but money that could have been earned and was not.
If you own a shop and rent it to your cousin for ₹5,000 a month when the market rate is ₹20,000, nobody stole ₹15,000 from you. But you are ₹15,000 a month worse off than you could have been, every month, by your own choice.
Why it matters here: it is a completely legitimate way for an auditor to measure a decision, and a completely illegitimate way to describe a theft. The word “loss” is doing two different jobs and the speaker rarely tells you which.
Both camps then used this gap exactly as you would expect. The party’s opponents quoted the large number as though it described a robbery. The party’s defenders quoted the small number as though the large one had been made up.
The honest position is unglamorous. The presumptive figure is a real measure of a real decision — the state gave away a scarce public asset far below what it later proved to be worth, and the subsequent auctions demonstrated that beyond argument. It is not a measure of money in somebody’s pocket. Both of those sentences are true, and there is no version of this story that lets either side keep only the half it likes.
7.3 — Lines that move, and series that get rebased
The second technique is subtler and it works on the honest and dishonest alike.
When a government says poverty fell, it is comparing the number of people below a line. That line is set by a committee, using assumptions about what a person needs. Change the assumptions and the number of poor people changes, without a single person’s life changing at all.
India has redrawn that line more than once. Each redrawing had a defensible technical reason. Each also produced a figure that could be quoted for or against a government, depending on which pair of years you chose.
The same problem affects the national income figures. The method for calculating them has been revised several times, with new base years and new data sources. Each revision was defended by professionals for professional reasons. Each also made some earlier period look better or worse than it had looked the day before.
The rule I use for this series is straightforward. Where a comparison depends on which method was used, I say which method, and I say what happens under the other one. Where the answer flips depending on the method, that is the finding, and it goes in as one.
The most reliable Indian numbers in this series share a family resemblance. They are counts that some office had to make in order to do its own job, where the count itself had no political stake at the moment it was made.
Crop procurement records exist because grain had to be paid for. Ration registers exist because people had to be fed. Electoral rolls exist because votes had to be counted. Court records exist because cases had to be listed. None of these was created to settle an argument, which is exactly why they can.
The least reliable numbers are the opposite: figures produced specifically to describe how well a programme is going, by the body running the programme, in a document intended to be read by the people funding it.
What this cannot show: routine records are honest about what they measure and silent about everything else. A ration register tells you who was fed. It does not tell you who was turned away, and it never will.
7.4 — The third technique, and the biggest
The third way to mislead with numbers is not to have any.
Chapter Four set the rule: when a state that counts everything has no count of something, somebody decided that. Here is the arithmetic consequence.
An argument between two people who both have figures is a real argument. It can be settled. An argument where one side has a figure and the other side has testimony is not a fair fight, and the side with the figure wins by default, regardless of whether they are right.
This shapes the whole of Indian public memory. The things that got counted — spending, output, votes, procurement — are argued about endlessly and precisely. The things that did not get counted are argued about with adjectives.
Both sides of every numerical fight in Indian politics share one belief and never state it: that the number is measuring the thing they are arguing about.
Watch a spectrum argument. One camp says ₹1.76 lakh crore. The other says ₹30,984 crore. They fight about the size. Neither asks what either figure is a measurement of — and they are measurements of two entirely different things, one hypothetical and one transactional. The fight is conducted as though the disagreement were about arithmetic. It is about definition, and definition is never on the agenda.
The same thing happens with poverty. Two people argue about whether poverty fell from thirty-seven per cent to twenty-two per cent. Both accept, without discussion, that “poverty” is a thing with a percentage. It is a line drawn by a committee. Move the line four rupees and tens of millions of people cross it without eating differently.
Here is why this matters more than it looks. A number that everybody accepts as the measure becomes the target. Once “the poverty figure” is the thing governments are judged on, the cheapest way to succeed is to affect the figure rather than the condition. And the cheapest way to affect the figure is to argue about the line.
I am going to use numbers throughout this series because the alternative is adjectives. But when I put a figure in front of you, the question I want you to ask is not “is it big” — it is what is this counting, and who chose what it counts.
None of this is an argument for distrusting figures. It is an argument for asking two small questions before you let one move you. What is it a count of, and did anybody have an interest in the counting. Most political numbers survive both questions comfortably. The ones that do not tend to be the famous ones.
Past a few thousand, you cannot picture a number. Every large figure in politics is aimed at that limitation. Convert it into people, wages or weeks before you let it persuade you.
Presumptive loss is not theft, and it is not nothing either. It measures a decision, not a transfer. Both camps quote the half that suits them.
The three legal ways to mislead: use the wrong kind of loss, move the line, or have no number at all. The third is the biggest, because a side with testimony always loses to a side with a figure.
8The Shape of the Charge
Two hundred items across nineteen parts, and they fall into four families. Knowing the families tells you what to watch for.
8.1 — Four families of failure
A list is not an argument. Two hundred separate charges, delivered one after another, produce a reader who is exhausted rather than convinced.
So here is the structure underneath the list. Everything in this series belongs to one of four families, and each family has its own kind of evidence and its own kind of defence.
8.2 — The first family: the rules themselves
This is the family that gets the least attention and does the most lasting damage, because it changes not what a government did but what any future government is able to do.
It includes the amendments to the constitution, particularly those affecting speech and the courts. It includes the repeated dismissal of elected state governments by the centre. It includes the Emergency, which is the largest item in this series and the only occasion on which the constitution was effectively suspended. It includes the treatment of the judiciary, and the practice of appointing and superseding judges with an eye to their views.
Article 356: the part of the Indian constitution that lets the central government dismiss an elected state government and rule that state directly. It is usually called President’s Rule.
It was written for a genuine emergency — a state where government has actually broken down. Whether it was used that way is one of the central questions of this series. The count is not in dispute and it is large.
Why it matters here: this single provision is the instrument that connects Delhi to every state chapter in the book, and it explains why a national party’s failures show up as local disasters.
The defence for this family is always the same and it is always about necessity. The country was fragile. The situation was urgent. The alternative was worse. That defence has to be tested case by case, and the test is whether the emergency described was actually happening. Sometimes it was.
8.3 — The second family: the economy
The largest family by cost and the smallest by drama. It covers the planning system, the permissions regime, the neglect of agriculture in the first two decades, the food crises that followed, the nationalisations, and the four decades during which India grew slowly while comparable countries did not.
It also covers the geography of it, which is Part Five: how central rules decided which Indian states became rich and which were hollowed out.
The defence here is the strongest defence in the whole series, and Chapter Two has already stated it. It deserves to be taken seriously and it will be.
8.4 — The third family: the blood
The hardest chapters to write and the ones where language has to be watched most carefully.
It covers the handling of partition and its aftermath. The war of 1962 and what was known before it. Punjab from 1978 to 1993. November 1984. The North East, including an episode in 1966 that almost no Indian outside that region has heard of. The Naxal corridor. The security failures that preceded major terrorist attacks. And the communal violence that recurs across the whole period, under governments of every description.
The evidence here is a mixture of commission reports, court judgments, and the testimony of people who were there. It is the family where the missing-count rule from Chapter Four applies most often and most painfully.
8.5 — The fourth family: the money
The family everybody thinks the book is about, and the one I rank last in importance.
It covers the procurement scandals, the allocations, the commissions on defence contracts, the games, the housing, the spectrum, the coal. These are real and several are serious. But a bribe on a contract, however large, is a smaller event than a rule that keeps four states poor for forty years, and this series will keep saying so.
It matters for a different reason than the money. A scandal tells you what the internal culture permitted, who was protected afterwards, and whether the system that was supposed to catch it worked. Those are the questions I will ask of this family, rather than adding up the totals.
Congress has been led by members of one family for most of its independent-era history. Everyone agrees on the fact. What it explains is genuinely disputed.
A party organised around a household selects for loyalty rather than for judgment. People who are right at the wrong moment do not rise. Internal correction becomes impossible, because the one person who cannot be removed is the one whose errors matter most. On this reading, the Emergency, the succession, and the long refusal to change economic course all trace to the same structural fact: there was no mechanism by which the leadership could be told it was wrong.
Look across India. Dynastic succession runs through nearly every major party and almost every state, including parties founded specifically in opposition to Congress. If the pattern is that general, it is not a peculiarity of one family — it reflects how Indian political parties actually work: weak internal institutions, personal networks doing the job of party machinery, and enormous costs of entry for anyone without a name. Blaming one family for a nationwide structure is comfortable and mistaken.
What would settle it: tracking what happened to senior Congress figures who publicly disagreed with the leadership, decade by decade, and comparing that with other parties. The material exists. Nobody has assembled it.
Why people care so much: because “dynasty” is now the shortest available argument in Indian politics, usable without any knowledge of any policy. That makes it valuable to one side and infuriating to the other, and it means the actual question underneath rarely gets asked.
8.6 — The ledger that runs underneath
One thread runs through every part from Part Five onwards, and I want to flag it here because it is the part of this series I expect to be most useful to a reader.
Most accounts of national politics stay national. They discuss a policy as though it happened to India in general. It did not. It happened to particular places, and the places did not receive it equally.
So from Part Five, every part carries a running ledger: which states were affected by the material in that part, and how. Bihar and Odisha appear early and often. Punjab appears as both a beneficiary and a casualty, in that order. Uttar Pradesh appears as the largest state in the country slipping quietly from the middle of the income table towards the bottom of it. The North East appears in places where the national account has a blank.
By Part Nineteen that ledger is a single table, and it answers a question the national story cannot: if you live where you live, what did this cost you?
Four families: the rules, the economy, the blood, the money. Each has its own evidence and its own defence.
The money family is the one people expect and the least important. A bribe is a smaller event than a rule that keeps four states poor for forty years. Scandals matter for what they reveal about the culture, not for the totals.
A running state ledger travels through the whole series, because policies do not happen to a country. They happen to places, and unequally.
9What Would Prove Me Wrong
The defence, at full strength, in its own words. If you are going to stop reading this book, this is the chapter that should make you.
9.1 — Four arguments for the defence
A book that cannot state the case against itself is not a book, it is an advertisement. So here are the four strongest arguments a serious defender of the Congress record would make. Not the weak versions. The ones that give me trouble.
One. It held.
In 1947 India was not a country in any ordinary sense. It was several hundred princely states in various stages of joining or refusing, a partition in progress with millions moving in both directions, more than a dozen major languages, and a population that was overwhelmingly poor and largely unable to read. No democracy that poor had ever survived anywhere. Nothing in the record of the world suggested it would work.
It worked. The country stayed one country. Elections kept happening. Power changed hands when votes said it should — including, crucially, in 1977, when the government that had suspended the constitution held an election and lost it and left. Most of the countries that became independent in that same decade did not manage that once. India’s neighbour had its first military coup in 1958 and several more afterwards.
The defender’s point is not that this excuses anything. It is that a book listing failures over seventy-nine years has quietly taken the survival of Indian democracy as the background against which failures are measured, when that survival is the single most improbable political achievement of the century, and it happened on Congress’s watch.
Two. The comparisons are rigged.
Every argument that India grew too slowly rests on a comparison, and the comparison is usually South Korea, Taiwan or Singapore. A defender would say: look at what you have chosen. Populations a fraction of India’s. Far greater ethnic and linguistic uniformity. Long periods of authoritarian rule in which unpopular decisions could simply be imposed. Enormous strategic aid and security guarantees from the United States for reasons that had nothing to do with economics.
Choose the comparison honestly — large, poor, plural, post-colonial and democratic — and the list of countries India can be compared with is very short, and India does not obviously come last on it.
Three. You are prosecuting three different parties as one.
The front matter of this book already conceded that treating Congress as one continuous actor is a choice. A defender would press much harder on it. The party that won independence, the party that imposed the Emergency, the party that liberalised the economy in 1991 and the party of the last decade have different leaderships, different memberships, different ideas and, in some cases, opposed ones. Convicting a name is not the same as convicting an organisation.
Four. The instruments used to convict them are instruments they built.
This is the argument I find hardest and I want to give it its full weight.
Almost every source in Chapter Four exists because Congress governments created it or left it alone. The auditor’s independence is constitutional and was not stripped when its reports became inconvenient. The commissions of inquiry that produced the most damaging findings were appointed under a law those governments passed. The Supreme Court struck down their legislation, repeatedly, and the legislation stayed struck down. The press that exposed the defence-contract scandal of the 1980s was not shut, and the newspaper that led the exposure was not closed.
A defender would say: you are able to write this book because they did not build the kind of state in which this book is impossible. That is not nothing. Look around the region and count how many neighbours produced a comparable set of institutions and then submitted to them.
And the honest counter, which I will make in the relevant parts, is that this argument has one enormous hole in it, and the hole is twenty-one months long. During the Emergency, exactly this list of institutions was suspended, curtailed or bypassed — the press was censored, the courts were packed and overruled by amendment, and the election was postponed. The defence rests on a record of restraint that was itself broken, deliberately, by the same party, when it became inconvenient.
9.2 — The tests I have set myself
Those four arguments are why this series carries obligations rather than intentions. Four of them, stated here so you can hold me to them across nineteen parts.
Every serious charge carries a grade. If you find a claim in this series with no grade and no source, that is a defect and you should treat it as one.
Every Argument box gives the other side its genuinely strongest case. Not the version I find easy to beat. If a defender of Congress reads one of those boxes and does not recognise their own argument in it, I have failed at that box.
Where the defence wins, I say the defence wins. There are items I expected to include which did not survive contact with the record, and there will be more.
The second series happens. Part Eighteen compares, and a separate series audits the government of 2014 onwards on the same terms.
9.3 — What I would have to retract
Here is what would actually falsify the argument of this series, as distinct from irritating me.
If it turned out that the pattern of failures had no pattern — that the items are simply what any government accumulates over fifty-four years, at roughly the rate any comparable government accumulates them — then the series would be a long list rather than an argument, and I would have to say so.
If the state-by-state ledger showed no relationship between central decisions and state outcomes — if the poor states were poor for reasons that have nothing to do with anything decided in Delhi — then Part Five collapses and a good deal of the rest weakens with it.
If the internal record showed that the warnings I expect to find were not there — that nobody in the room said “this will not work”, and the alternatives really were invisible at the time — then Chapter Three’s tests turn most of these charges from blunders into misfortunes.
I do not expect any of those. But I have written them down before starting, which is the only way that sentence means anything.
Now the one turned on this book, and it is the most serious problem with the entire project.
Everyone who reads a book like this — and everyone who writes one — assumes that a list of failures is a judgment of a government.
It is not. A list of failures is a list of failures. To be a judgment it would need a denominator: how many decisions were taken in total, of which these were the bad ones.
Fifty-four years of governing a country of this size means a very large number of decisions. Budgets passed, appointments made, treaties signed, famines that did not happen, epidemics that were contained, elections conducted, railways run. Almost all of it was ordinary and none of it is in this book, because ordinary is not what a prosecution brief collects.
So when you finish nineteen parts, you will have an accurate picture of the worst of a record and no picture at all of its size. Those feel identical from the inside. That is the thing to guard against, and I cannot guard against it for you, because the structure of what I am writing produces the effect automatically.
Notice what this does not rescue. It does not make any individual charge weaker. The Emergency is not smaller because other things also happened. What it does is prevent a specific move — the move from “here are two hundred failures” to “therefore this was the worst possible government”. That second sentence needs a comparison this book does not contain and is not built to contain.
The most honest description of what you are holding: this is a complete account of one side of a ledger, and there is no version of it that is also the other side. Part Eighteen is the closest I get, and it is not close enough. I would rather tell you that now than have it be the thing you work out on your own in Part Fifteen.
That is the case against this book, made as well as I can make it. I do not think it wins. I think the four defences are real, that the fourth is genuinely strong, and that none of them touches the specific charges in Parts Nine, Ten and Eleven, which are not arguments about growth rates or comparisons but about what was done to people by a state that had decided to do it.
But you now have the material to disagree with me, which is the point of the chapter. A reader who finishes this series still holding the four arguments above has not been badly served. A reader who never encountered them has.
Four serious arguments for the defence: the country held together and stayed a democracy; the usual comparisons are rigged; three different parties are being convicted as one; and the instruments used to convict them are instruments they built and submitted to.
The fourth is the strongest, and it has a hole twenty-one months long. During the Emergency, exactly those institutions were suspended by exactly that party.
A list of failures has no denominator. This book gives you an accurate picture of the worst of a record and no picture of its size. Those feel the same from the inside, and that is my problem to declare and yours to remember.
10An Honest List of What We Do Not Know
Every part of this series ends here. Two lists: what is genuinely unknown, and what is solid enough to build on. The first list is usually more interesting.
10.1 — Why this chapter exists in every part
Most writing about politics has one setting: confident. Everything is asserted at the same volume, whether it rests on a Supreme Court judgment or on something the author feels.
So the last chapter of every part in this series is a separation. What do we actually know, and what are we guessing at? For this first part, the lists are mostly about the series itself — what nineteen parts will and will not be able to establish.
10.2 — Genuinely unknown
Whether the alternatives were visible at the time, in most cases. Chapter Three set this as the test for turning a bad outcome into a blunder. Meeting it requires the internal record — who warned whom, and when. Some of that record is public. A great deal of it is not. India has never had a systematic declassification programme for cabinet papers of the kind that exists in Britain or the United States, and the thirty-year convention is honoured unevenly.
Why it is unknown: because releasing it is discretionary, and every government of every party has found the discretion useful.
How many people died in several of the events in this series. Not roughly. At all. In more than one case the range between the lowest credible figure and the highest is a factor of three or more, and the gap is not caused by the difficulty of counting.
Why it is unknown: because a count was not taken by bodies that count everything else. Chapter Four’s rule applies. This is a finding, and the relevant parts treat it as one.
What was in the evidence the Shah Commission gathered. The reports have been reconstructed and republished. The roughly forty-eight thousand papers behind them, and the verbatim record of around a hundred sittings, are a different matter, and the archive that was supposed to hold them has said it does not.
Why it is unknown: the honest answer is that nobody has produced a document explaining it, and I am not going to invent one.
What the freight equalisation policy actually cost the eastern states. That it hurt them is not seriously disputed. Putting a number on it requires imagining the industrial map that would otherwise have formed, and that map does not exist anywhere except in models whose assumptions decide their own answers.
Why it is unknown: because it is a counterfactual, and Chapter Three explained why those cannot be measured directly.
Whether the family at the top of the party is a cause or a symptom. Chapter Eight’s argument box laid out both cases. The evidence that would settle it — what happened to senior figures who disagreed, across decades, compared with other parties — has never been assembled by anyone.
Why it is unknown: because nobody has done the work, which is a much better reason than most.
Whether named individuals committed crimes in the spectrum case. Acquitted in 2017, under appeal since 2024, undecided.
Why it is unknown: because the courts have not finished, and until they do, the honest word is the one on the grade.
10.3 — Solid
The electoral record. Who won, where, with how many votes, in every general election since 1951. This is as solid as anything in Indian public life gets, because it was produced by an independent body under adversarial conditions with every party watching.
The legal machinery of the Emergency. What was proclaimed, when, under which article, which amendments followed, which detentions were authorised under which law. All of it is in the gazette and the statute book. There is no dispute about the mechanics, only about the justification.
The number of times elected state governments were dismissed by the centre. Each instance is a formal act with a date and a document.
That the freight equalisation policy existed from 1952 to 1993 and did what it did. It was a stated policy, not a secret one. Its intended effect was published; its unintended effect is a matter of argument and the policy itself is not.
That the Supreme Court cancelled all 122 spectrum licences in February 2012. A judgment.
That the petition alleging a rape by Rahul Gandhi was fabricated. A reported judgment of the Supreme Court in 2012, with costs awarded, after the bench examined the merits specifically in order to record the finding.
That Feroze Gandhi was born into a Parsi family as Feroze Jehangir Ghandy. Birth records, family details, burial arrangements, and a large published biography.
What Congress promised. Manifestos, budget speeches, programme targets and answers given in Parliament. This is the most under-used solid material in Indian political argument, and Chapter Three explained why: nobody can call the standard unfair when the government set it.
10.4 — The difference that matters most
Look again at the two lists and you will see that “unknown” is doing two completely different jobs.
Some things are unknown because they are hard. A counterfactual industrial map is hard. Nobody is hiding it; it does not exist and cannot.
Other things are unknown because somebody decided. A missing death toll in a state with a functioning registry. An archive that was supposed to receive records and did not. A cabinet paper that could be released and is not.
Those two kinds of not-knowing look identical on the page. They are opposites. The first is a limit of the world. The second is a finding about the people who arranged it, and in this series it will be treated as one every single time.
Genuinely unknown: whether the alternatives were visible in most cases; several death tolls; what the Shah Commission’s underlying evidence contained; what freight equalisation cost; whether dynasty is cause or symptom; whether named individuals committed crimes in the spectrum case.
Solid: the electoral record; the legal mechanics of the Emergency; the count of dismissed state governments; the existence and design of freight equalisation; the licence cancellation; the two fabrications in Chapter Six; and everything Congress promised in its own words.
Some things are unknown because they are hard. Others are unknown because somebody decided. Those look the same on the page and they are opposites. Telling them apart is most of what this series is for.
Sources & further reading — Part 1
The Ground This Series Covers
A skeleton of the whole nineteen parts, so you can place anything you read against it. Dates only; the arguments are in the parts.
| Years | What happened | Part |
|---|---|---|
| 1947–1950 | Independence and partition. The princely states are absorbed. Kashmir accedes and the fighting is referred to the United Nations. The constitution is written and adopted. | 2 |
| 1950–1959 | The First Amendment restricts speech. Land reform begins. The first general elections. The elected Kerala government is dismissed by the centre. | 3 |
| 1951–1991 | The planning and licensing system operates. Freight equalisation runs from 1952 to 1993. | 4, 5 |
| 1954–1962 | The agreement on Tibet. The border dispute develops. War with China in 1962. | 6 |
| 1947–1989 | Wars with Pakistan in 1947–48, 1965 and 1971. The Simla agreement. The governing of Kashmir up to the insurgency. | 7 |
| 1966–1975 | Indira Gandhi’s rise. The party splits. Banks are nationalised. The judiciary is confronted. | 8 |
| 1975–1977 | The Emergency. Censorship, detention without trial, the forced sterilisation programme, the postponed election. | 9 |
| 1978–1993 | Punjab: the rise of militancy, Operation Blue Star in 1984, and the decade that followed. | 10 |
| Nov 1984 | The killing of Sikhs in Delhi and elsewhere after the assassination, and the four decades of impunity that followed. | 11 |
| 1984–1989 | Rajiv Gandhi’s government. Shah Bano. The Babri locks. Bofors. Sri Lanka. The 1987 Kashmir election. | 12 |
| 1958–present | The North East, the tribal belt and the Naxal corridor — the geography these accounts usually skip. | 13 |
| 1991–1996 | Economic liberalisation and the demolition of the Babri Masjid, under one government. | 14 |
| 2004–2009 | The first UPA government. The nuclear deal. The terror wave and Mumbai in 2008. | 15 |
| 2009–2014 | The second UPA government. Spectrum, coal, the games, and what the courts later found. | 16 |
| 1947–present | What was said: the statements, with dates and full wording. | 17 |
| 2014–2026 | The comparison, measured on the same ruler in both directions. | 18 |
| — | The final accounting: what held, what did not, and everything I could not prove. | 19 |
Part Two starts here. It opens in August 1947, with a government three days old, a border announced after the transfer of power, and several hundred princely states that had not yet decided whether they were joining anything.
Glossary
Every hard word used in this part, in plain English.
| Word | What it means |
|---|---|
| Acquittal | A court’s decision that a charge was not proved against a particular person to the standard the law requires. It does not mean nothing happened. |
| Article 356 | The part of the constitution allowing the central government to dismiss an elected state government and rule the state directly. Usually called President’s Rule. |
| CAG | Comptroller and Auditor General. The constitutional auditor of India, who checks government spending and reports to Parliament rather than to the government. |
| Chargesheet | The formal document in which an investigating agency sets out the charges it intends to prove against named people. |
| Commission of inquiry | An investigation, usually led by a judge, appointed by a government to examine a particular event. Its powers and its questions are set by the government that appoints it. |
| Counterfactual | A claim about what would have happened if things had gone differently. Hidden inside every sentence containing the word “should”. |
| Crore | Ten million. One hundred lakh. |
| Culpable | Deserving blame. Requires both that a reasonable person could have known better, and that they had the power to act. |
| Exchequer | The government’s money, taken as one pot: what it collects and what it spends. |
| Lakh | One hundred thousand. |
| Lakh crore | One lakh multiplied by one crore. A one followed by twelve zeros. Called a trillion elsewhere. |
| Leave to appeal | Formal permission from a higher court to challenge a decision. Granting it is not a verdict; it means the challenge is arguable. |
| Parsi | A member of a small Indian community descended from Zoroastrians who came from Persia. Zoroastrianism is its own religion, not a branch of Islam. |
| Presumptive loss | Money that could have been earned and was not, rather than money that was taken. Also called notional loss. A legitimate way to measure a decision and an illegitimate way to describe a theft. |
| Prosecution brief | The written case against an accused person. Not required to be balanced. Required to be true in every particular. |
| Spectrum | The invisible radio frequencies mobile phones use to reach towers. Limited in quantity, owned by the state, and extremely valuable. |
| Terms of reference | The written list of questions a commission of inquiry is appointed to answer. The quietest way to control an investigation is to write these carefully. |
What Comes Next
Part Two — The Inheritance, 1947 to 1950
Part Two takes the first thousand days. It asks:
- What was actually handed over in August 1947, and how much of the state that took delivery was already in place?
- Who decided how the princely states would be brought in, and what was promised to each of them that was later not honoured?
- Why was the fighting in Kashmir referred to the United Nations, who advised it, who opposed it, and what did the people making that decision expect to happen?
- What did the ceasefire line freeze into place, and was there a point at which it could have been settled differently?
- Was the government of a country three days old capable of protecting the people moving across the new border, and did anyone in that government believe at the time that it was?
- Which of the early constitutional choices were genuinely open, and which had already been made by the colonial administration and simply inherited?
Questions, not topics. If any of them is answered badly, you will be able to tell.