What This Part Is Not
This is an advocacy document. It is the case for the prosecution against Congress governments in power. The selection of material is one-sided on purpose. The evidence in it is not. Every serious claim carries a grade showing how well it is proved, the other side is given its strongest argument, and where the other side wins, this book says so. A second series applying the same test to the government of 2014 onwards is planned.
This is the part where a book like this one usually stops being worth reading.
Seventeen parts of prosecution have to be followed by something, and there are only two things it can be. It can be a comparison in which the current government comes out well, in which case a reader concludes — correctly — that the whole exercise was a campaign document with footnotes. Or it can be a second prosecution, in which case the author has demonstrated only that he can write the same book twice with the names changed.
Neither of those is worth sixty pages, and this part is neither.
The purpose is to test the ruler, not to judge the government.
Parts One to Seventeen produced a method: graded evidence, the other side given its strongest case, and — at the end of eleven separate parts — a table of ordinary laws that would have prevented the failure described, none of which was ever made.
A method that has only ever been applied to one subject cannot be evaluated. It might be a good method. It might be a machine for producing charges against Congress. There is exactly one way to find out, which is to point it at something else and see what comes out.
What comes out is in Chapter Eight, and it was not what this author expected.
Which requires disposing of the question everybody arrives with.
Every reader who has reached Part Eighteen assumes that a comparison part is about which side is worse. That is what comparisons are for, and it is what both political camps will read this part looking for.
It is the wrong question and this book cannot answer it. Part One’s closing box conceded that a list of failures is not a judgment on a government, because there is no denominator. Part Sixteen conceded that sixteen parts of careful grading had produced, on the largest case, a conclusion too honest to be useful as advocacy. Part Seventeen conceded that words and acts do not correlate.
Those concessions do not become inoperative because the subject changed. A book that cannot deliver a verdict on Congress after seventeen parts cannot deliver one on anybody else in one.
What it can do is different and narrower: take the specific findings this series has made — these instruments, these unmade rules, these unreleased documents, this pattern of delay — and ask what has happened to each of them since 2014. Those are checkable. They do not require a verdict on anybody.
And the answer to almost all of them is the same, which is the finding of this part.
What the previous parts committed this one to
Four debts fall due here and all four are paid.
Part Two named Article 370’s “temporary” heading and said Part Eighteen owes the account of the 2019 removal. Chapter Four.
Part Seventeen promised that Part Eighteen would carry the equivalent list of statements, under the identical rule of date, venue and operative wording, and said in terms that if it did not, Part Seventeen’s fifth defence point was rhetorical. Chapter Six.
Parts Nine, Eleven, Thirteen, Fourteen, Fifteen, Sixteen and Seventeen each ended with a table of ordinary laws that were never made. Chapter Seven asks what has happened to every one of them.
And §3.3 of this book’s own working brief, written before Part One, said that if this part read as an advertisement it would retroactively discredit everything before it. That is the constraint under which the following ten chapters were written, and Chapter Ten reports honestly on whether it held.
1What “The Same Ruler” Means
A comparison is only worth anything if the thing being compared with is measured the same way. This chapter sets out how, in enough detail that a reader can check whether it was done.
1.1 — The four tests
Every part of this series has asked the same four questions of the material in front of it. They are asked here in the same order.
| Test | The question | Established in |
|---|---|---|
| Grade | Who found this, and to what standard? A court, a commission, an auditor, or nobody? | Part One |
| Instrument | What tool was used, who built it, and had it been used before? | Parts Nine and Ten |
| Rule | What ordinary law would have prevented this, and was it made? | Parts Nine, Eleven, Thirteen to Seventeen |
| Document | What would settle the question, and is it public? | Parts Five, Six, Thirteen to Seventeen |
Chapter Eight’s finding is that three of the four tests produce almost identical answers whichever government they are applied to, and that this is a fact about the Indian state rather than about either party.
1.2 — What this part will not do
Four exclusions, all of them deliberate, all of them costly to the appearance of this part.
No comparison of totals. This book will not say that one government has been worse than another. Part One conceded there is no denominator and nothing has supplied one since.
No adjudication of live political disputes. Where a matter is contested and no court or commission has ruled, this part sets out both cases at full strength and stops. That applies to several of the largest questions in these twelve years and Chapter Ten lists them.
No prediction. This government is in office. Every judgment in Parts One to Seventeen had the advantage of knowing how the story ended, and this one does not.
No second prosecution. That is what the promised second series is for, and Chapter Ten states what has now been committed to.
Comparing two things by holding everything else constant. In this part it means four specific disciplines.
Same standard of proof: a Supreme Court finding against one government counts exactly as much as a Supreme Court finding against the other. Same treatment of acquittals: Part Sixteen refused to treat a failed prosecution as a proved charge and so does this part. Same treatment of achievements: Parts Fourteen to Seventeen put the credit chapter second and so does this one. Same treatment of statements: date, venue and operative wording, per Part Seventeen.
Where this part departs from any of those, it says so in the text.
Each of those four exclusions has a price and it is worth naming before the chapter moves on.
An exclusion list is easy to write and easy to ignore, so it is worth showing what these four cost.
The second exclusion — no adjudication of live political disputes — removes from this part the five matters listed in Chapter Ten, which are the five things most readers will have arrived wanting an answer on. That is not a small omission and Chapter Ten’s argument box states the objection to it in the strongest terms this book can manage.
The fourth — no second prosecution — removes the entire genre this part would otherwise belong to. Seventeen parts of material exist about one government. The equivalent for this one would be another nineteen parts, and producing a compressed version in sixty pages would be worse than producing none.
Which is why section 10.5 exists and why the commitment there is written in terms specific enough to be held to. A part that declines to prosecute and does not commit to prosecuting later has simply declined.
1.3 — The fifth discipline
Four tests were listed above. There is a fifth thing this part had to do, which is not a test but a habit, and it is the one most likely to have failed.
Every part of this series has recorded, somewhere in it, a place where the evidence went against what the author wanted. Part Five found the Punjab rivers verdict to be weak on the water and strong only on the procedure, which is not the conclusion most Punjabi accounts reach. Part Six conceded that the correction after 1962 worked and the same army won in 1971. Part Ten recorded that the majority of the Sikhs who died in Punjab were killed by Sikhs. Part Fourteen conceded Bommai outright. Part Sixteen conceded that the presumptive figures were soft.
Those concessions are the only reason the rest of this series is worth reading, and they were all made about a party the author was prosecuting.
A concession made against your own case is evidence about your method. A concession made in favour of it is not evidence of anything, because there was never any resistance to overcome.
Which sets a specific and awkward test for this part. If Part Eighteen contains no place where the evidence went against what a critic of Congress would want to find, it has failed — not because balance is a virtue in itself, but because a method that produced resistance seventeen times and none on the eighteenth was never a method.
Chapter Two is where that was hardest, Chapter Nine’s fifth defence point is where it was sharpest, and Chapter Eight is where it turned out to matter most. A reader should check all three.
1.4 — The problem this part cannot solve
There is one asymmetry that no amount of discipline removes, and it should be stated before anything else.
Congress governed for roughly fifty-four of the seventy-nine years covered by this series. This government has been in office for twelve.
More than that: seventeen parts of this book rest on documents that took decades to surface. The Shah Commission reported in 1978 on 1975. The Liberhan Commission reported in 2009 on 1992. The urea convictions came in 2018 for a payment made in 1995. Part Sixteen’s central case is still on appeal.
Most of what will eventually be known about 2014 to 2026 is not knowable in 2026. A judgment made now is a judgment made without the evidence that decided every other part of this series.
That cuts one way only, and this part says so at the front rather than hiding behind it: everything in the charge column here is provisional, and everything in the credit column is provisional too.
Two objections to this whole part will be made from opposite directions, and both belong on the page before anything is measured.
The objection that will be made by supporters of this government: that a book written by a critic of Congress, applying a method designed to prosecute Congress, cannot fairly assess anybody else — and that the exercise is a way of borrowing credibility from seventeen parts of hostile work in order to spend it here.
The objection that will be made by critics of this government: that twelve years cannot be handled in sixty pages, that the most serious questions are precisely those no court has ruled on, and that a part which refuses to adjudicate them is not neutral but is producing a specific result by omission.
Both objections are substantially correct and this book cannot answer either. What it can do is make the method visible enough that a reader can see where it was applied and where it was not, which is what Chapter One is for and why it comes before any material.
A reader who concludes at the end that this part is too soft, and a reader who concludes it is too hard, are both reading the same pages. Chapter Ten records which of the two the author expects to hear from more.
With that said, the measuring can begin.
This part exists to test the ruler, not to judge the government. A method applied to only one subject cannot be evaluated at all.
Four tests, applied identically: who found it, what instrument was used, what rule would have stopped it, and what document would settle it.
And one asymmetry nothing fixes: Congress governed for fifty-four years and this government for twelve, and most of what will be known about these twelve years is not knowable yet.
2What Was Done That Was Not Done Before
Parts Fourteen through Seventeen all placed the credit chapter second, before the charge, and said why: at the end it reads as a sweetener produced late. This part does the same, and here the reasoning runs in the opposite direction — placed late, in a comparison, it would read as the point.
2.1 — The tax
The Goods and Services Tax came into force on 1 July 2017, replacing a system in which goods crossing a state border were taxed by each state they passed through.
Part Sixteen recorded that this was on the UPA’s own reform agenda and did not move, and gave the reason its defenders gave: that it required a constitutional amendment and the consent of the states, which took another six years to obtain.
Before 2017 a lorry travelling from Chennai to Delhi passed through checkposts at every state line, paying and being assessed at each.
Studies of the period found trucks spending a substantial share of their journey time stationary at state borders. The checkposts were dismantled.
Part Five of this series is about a national policy that decided which states got rich by manipulating what it cost to move goods across the country. The GST is the first structural change to that system since freight equalisation was dismantled between 1991 and 1993 — and unlike freight equalisation, it was done by agreement with the states rather than to them.
The honest qualification, which the credit survives: the design that emerged has multiple rates rather than the single rate most of its economist advocates wanted, compliance was difficult for small businesses for several years, and the compensation promised to states became a serious dispute during the pandemic. It is a real reform, imperfectly executed, that nobody had managed to execute at all.
Whether that counts as an achievement can be tested without taking anybody’s word for it.
The test of whether a tax reform worked is not what either side says about it. It is what happened to the thing it was meant to remove.
The checkposts are gone. That is a physical fact, verifiable by anybody who drives, and it is not disputed by the reform’s critics — who argue instead about rate structure, compliance burden and the compensation owed to states.
When an opposition’s criticism of a reform is entirely about its design and not at all about whether it happened, the reform happened. Part Sixteen used the same test on the 2013 land law, from the other direction: a law the next government spent a year trying to weaken and could not is a law with real content in it.
2.2 — The bankruptcy law
The Insolvency and Bankruptcy Code was enacted in 2016. Before it, a company that could not pay its debts could remain in that condition more or less indefinitely, because there was no single process for resolving it and no deadline.
When a business fails, somebody has to decide quickly whether it can be saved and, if not, who gets what is left. A country without that process has banks carrying loans that will never be repaid and factories that nobody can buy, sell or restart.
Part Four of this series described a permission system in which nobody both decided and paid. An insolvency regime is the opposite arrangement: the person who lent the money decides, and bears the loss.
It is also the mechanism that makes a promoter’s control of a company conditional on repaying, which had not previously been the case in India.
Its record is mixed and contested — recovery rates and timelines have fallen well short of what the statute intended, and the resolution process routinely exceeds its own deadlines. It remains a structural change that had been discussed for decades and was made.
2.3 — Delivery at scale
Part Sixteen recorded that the previous government built the biometric identity system and began direct benefit transfers in January 2013, and that the identity authority operated for seven years without a statute.
What was done after 2014 with that architecture is the largest item in this chapter, and this book will not let the argument about who built it obscure what was done with it.
Bank accounts opened under the financial inclusion programme launched in August 2014 run into the hundreds of millions. Household cooking-gas connections, household toilets, rural road kilometres, rural electrification and piped water connections all moved by very large numbers over the same period.
The unified payments system built on that base processes a volume of transactions that has few parallels anywhere.
Part Fifteen found that a legal right which arrives two months late has already failed for the person holding it. The single most consequential thing about paying a subsidy into an account rather than through an intermediary is that it arrives, and that somebody can prove whether it did.
There is a serious objection to counting any of that as an achievement, and it has to be put properly.
The objection to this whole section, and it is serious: counting outputs is not the same as measuring outcomes. A toilet built is not a toilet used; a gas connection issued is not a cylinder refilled; an account opened is not an account with money in it. Several independent assessments have found substantial gaps between the first number and the second in each case.
And the architecture was inherited. The identity system, the payment rails and the direct-transfer mechanism were all built between 2009 and 2014 and are described in Part Sixteen.
The reply, which concedes the first half: the gap between issue and use is real, it is documented, and this book does not treat the headline numbers as achievements in themselves.
Where the objection fails is on the second half. Part Fourteen credited the 1991 reforms to a government that inherited a diagnosis made in 1969. Part Sixteen credited the coal auctions to a government that legislated them in 2010 and the implementation to the one that ran them from 2015. A series which has repeatedly said that executing somebody else’s good idea is a real thing to have done cannot withdraw that finding when the executor changes.
2.4 — The pandemic, which belongs in both columns
A comparison covering twelve years cannot omit the largest single event in them, and this book has no way of scoring it either.
What is not disputed on either side is the scale of two things: the vaccination programme, which was among the largest ever run anywhere; and the free-grain expansion of the food entitlement described in Chapter Three, which reached most of the population for an extended period.
What is disputed, bitterly and by serious people, is the first three months of it.
For: a national lockdown was announced in March 2020 at a point when very little was known about the disease and health systems everywhere were being overwhelmed. India’s health infrastructure per head is among the thinnest in any large country, and a government facing that had to choose between acting early on poor information and acting late on better information. The vaccination programme that followed, and the domestic manufacturing behind it, are achievements by any measure.
Against: the notice given before the lockdown was very short, and the consequence was that millions of migrant workers in cities lost work, income and shelter simultaneously with no way to get home, and walked. That was foreseeable and was foreseen. The second wave of 2021 found the health system without oxygen capacity that had been recommended and not built. And the official death toll is substantially lower than every independent excess-mortality estimate, by a margin large enough that the difference is itself a subject of dispute.
What this book will say: the vaccination programme goes in the credit column and the notice period goes in the charge column, and neither cancels the other. On the mortality figures it has no basis to prefer any estimate — Part Two faced the same problem with the Partition toll and gave the range and the method behind each end rather than choosing, which is what Chapter Ten does here.
That last refusal is not a dodge, and the reason is a comparison this series is uniquely placed to make.
Part Two of this series recorded that the number of people killed in Partition is given by serious scholars anywhere between two hundred thousand and two million, and that this book would give the range rather than pick.
Seventy-nine years later, in an event that happened inside a functioning administrative state with digital death registration, the gap between the official figure and the independent estimates is proportionally of the same order.
Part Thirteen found an unmade addition where every component existed. Part Fifteen found no national register of terrorist incidents. This is the same finding again and it is the most consequential instance of it in nineteen parts, because the components here are birth and death registrations that the state already collects.
2.5 — The economy, which this book will not score
A comparison part that says nothing about twelve years of economic performance would be evading the largest question anybody has about them. A comparison part that scores it would be pretending to a certainty that does not exist.
What follows is what is not disputed, followed by what is.
Not disputed: India moved from around the tenth-largest economy in the world to around the fifth over this period. Inflation, which ran in double digits for long stretches in the years Part Sixteen describes, came down substantially and has largely stayed down. Physical infrastructure — roads, ports, airports, rail electrification, transmission — expanded at a rate not previously achieved.
Also not disputed: the size of an economy is a function of population and time as well as policy, and India would have moved up that ranking under almost any government.
Disputed, by serious economists on both sides: the rate of growth relative to what was achievable; whether employment kept pace with output; whether household consumption and real wages rose in step with the aggregate; and how much of the last decade’s performance is attributable to policy at all rather than to demography, global conditions and a low base.
This book cannot settle any of those and does not try. Part Four made the same refusal about the licence system and Part Sixteen about the 1991 reforms — declining to put a number on a counterfactual because no honest number exists.
What the two sides actually claim is worth setting out, because the disagreement is narrower than the noise around it suggests.
For: macroeconomic stability was restored after the volatility Part Sixteen describes, inflation was brought under a formal targeting framework, the banking system was cleaned up through the insolvency code, and India came through a global pandemic without a debt crisis. Those are policy outcomes and they are attributable.
Against: unemployment, particularly among the young and the educated, has been the most persistent political complaint of the period, and the 2024 result is generally read as partly about it. Two large shocks — the currency measure of November 2016 and the initial GST rollout — imposed costs on small and informal businesses that are not captured in headline figures. Real wage growth has been weak.
What this book will say, and it is nearly nothing: both of those accounts describe real things, both are held by economists with no partisan attachment, and the evidence that would settle between them — long-run household and employment series that everybody accepts — is itself contested. An honest audit of an economy is a book, not a section, and this is a section.
2.6 — Two things that were legislated and one that was constituted
The Muslim Women (Protection of Rights on Marriage) Act, 2019 criminalised the pronouncement of instant triple talaq, following the Supreme Court’s judgment of August 2017 setting the practice aside. Part Twelve of this series is about what happened the last time an Indian government legislated on Muslim personal law after a Supreme Court judgment, and the contrast is the point: in 1986 a government legislated to reverse a court; in 2019 it legislated to follow one.
The Constitution (One Hundred and Sixth Amendment) Act, 2023 reserved a third of seats in the Lok Sabha and state assemblies for women. It has not yet come into effect, being tied to a delimitation exercise following a census, and Chapter Seven returns to that.
And the Lokpal was constituted in March 2019 — the office Part Sixteen recorded as legislated in December 2013 and left empty for five years and three months. It was filled by this government, after the Supreme Court had been asked twice to compel it.
Substantial, and larger than a hostile reader will expect. A tax reform that had defeated every previous government; an insolvency regime; delivery of basic services at a scale that is not in dispute even where its quality is; and an anti-corruption office finally filled.
The qualifications, all of which stand: the delivery architecture was inherited and Part Sixteen says by whom; the output numbers are not outcome numbers; the GST design is not the one its advocates wanted; and the women’s reservation is enacted and not in force.
What this book will not do is what a comparison part is usually for, which is to net this off against something. Chapter One said there is no denominator. There still is not.
Which is the honest shape of the credit column, and it is larger than the reputation of this book would suggest.
A national goods and services tax in 2017, which every previous government had wanted and none had passed. An insolvency law in 2016. Basic services delivered at a scale that is not seriously disputed.
The Lokpal, legislated in December 2013, was finally constituted in March 2019 — by this government, after the Supreme Court had been asked twice.
The architecture underneath much of the delivery was built between 2009 and 2014. This series has twice said that executing somebody else’s good idea is a real thing to have done, and it does not withdraw that here.
3The Same Policy, A Different Name
This chapter is the one both sides find least convenient, which is usually a sign that it is the useful one.
3.1 — What continued
| The thing | Before 2014 | After 2014 |
|---|---|---|
| Rural employment guarantee | Enacted 2005; Part Fifteen | Retained, renamed in usage, and its budget in several years the largest it had been |
| Biometric identity and direct transfers | Authority created 2009; transfers from January 2013; Part Sixteen | Given a statute in 2016 and extended to most subsidy delivery |
| Food security | Enacted 2013; Part Sixteen | Retained and, during the pandemic, substantially expanded |
| Auctions for natural resources | Coal auctions legislated 2010; spectrum from 2012; Part Sixteen | Implemented at scale from 2015 and made the default |
| Sanitation and rural roads | Rural roads programme from 2000; earlier sanitation campaigns | Continued under new names with much larger budgets and better completion |
| The land acquisition law of 2013 | Enacted 2013; Part Sixteen | Ordinance to weaken it in December 2014, re-promulgated twice, abandoned in 2015. The 2013 Act stands |
The last row is the interesting one and it runs against the pattern of the others: an attempt to reverse a predecessor’s law that failed, and the predecessor’s law survives. Part Sixteen used that failure as evidence that the 2013 Act had real content in it.
Both sides of Indian politics treat a renamed programme as a different programme — governments rename things to claim them, oppositions treat the renaming as proof that nothing else changed — and both readings are lazy. A name tells you nothing on its own. The statute, the budget line and the delivery mechanism tell you everything, and they move independently of the name and of each other.
The employment guarantee’s name in official usage changed; its statute did not; its budget rose. The identity system’s name did not change; it acquired a statute in 2016 that it had lacked for seven years; its use expanded enormously. Sanitation’s name changed, its budget multiplied, and its completion rate — the thing that actually matters — changed more than either.
So the honest finding is neither that everything was continued nor that everything was transformed. The machinery was inherited and the throughput changed, and which of those two facts a reader thinks is the important one is a political judgement this book is not going to make for them. What this series can say, because it has said it eleven times, is that the instrument almost always predates the government using it — and Chapter Four is the sharpest case of that in this part.
Before drawing anything from that table, it is worth saying how it was built.
Continuity is easy to assert and hard to prove, because both sides have an interest in the answer. Three things settle it and none of them is a claim by anybody.
The statute book. The 2005 employment Act, the 2013 food Act and the 2013 land Act are all on it, unrepealed. That is checkable in a minute.
The budget documents. Allocations to each programme are published annually and the series is continuous across 2014.
The failed ordinance. Three attempts to amend the 2013 land law, abandoned in 2015. A government that had replaced its predecessor’s framework would not have needed to try.
This book has used none of the claims made by either party about any of these programmes, because on this particular question both sides are unreliable in the same direction: the government wants credit for what it kept and the opposition wants credit for what it built.
Reading that table correctly requires one distinction that almost every account of these programmes gets wrong.
Three different things, routinely confused, and the confusion is what allows both sides to be right at once.
A scheme is an administrative programme. It can be renamed, merged or discontinued by executive decision. A statute is an Act of Parliament creating an entitlement; it can only be changed by Parliament. A budget line is the money attached, which moves annually with no change to either.
So a programme can keep its statute, lose its name and double its money — which is what happened to several of the entries in Table 2 — and a government can truthfully say it transformed the thing while an opposition truthfully says it inherited it.
The only one of the three that tells you whether a person has a right is the statute. Part Fifteen’s word box made that distinction about schemes and rights, and it is the reason this table has three columns instead of one.
3.2 — Two continuities nobody claims
Two further areas belong here because they are the ones neither side finds it useful to describe as continuous.
The relationship with Washington
Part Fifteen’s Chapter Seven is about the civil nuclear agreement of 2005 to 2008 — three years, six international gates, a governing coalition brought down, and a confidence vote. Its central finding was that the achievement was diplomatic rather than electrical: what India obtained was the end of thirty years of exclusion and formal acceptance that it could hold nuclear weapons without penalty.
Everything since has been built on that. The defence agreements, the technology arrangements, the quadrilateral grouping and the trade relationship all sit on a foundation laid between 2005 and 2008 by a government whose own allies walked out over it.
The Look East policy of the 1990s — Part Fourteen — became Act East. The strategic opening to Washington of 2005 to 2008 — Part Fifteen — became a much larger relationship.
Neither renaming involved a change of direction. Both involved a great deal more activity in the same direction.
And the same is true in reverse of the thing that has not changed: India still has no boundary settlement with China, the line described in Part Six is still where the fighting of 1962 left it, and the Henderson Brooks report from 1963 is still not public. Sixty-three years, and Chapter Seven’s second table records the document in the same place.
The states
Part Three described Article 356 being turned into an instrument in 1959. Part Sixteen recorded the Supreme Court restricting it in 1994, and Chapter Four here records the count falling sharply after that and staying low.
What has replaced it is not a return of power to the states. The Planning Commission — the body Part Five described allocating Plan transfers by formula, with the dissenting notes nobody read — was wound up in 2015 and replaced by an advisory body with no allocative power. The allocation function moved to the Finance Commission and the ministries.
And the goods and services tax created something genuinely new: a Council in which the Union and the states vote together on rates, with the Union holding a weighted share.
That this is federalism strengthened: the states now sit in a body that sets a national tax, which they never did before. Dismissal of state governments under Article 356 has become rare. The Finance Commission’s transfers are formula-driven and its recommendations have been accepted.
That it is federalism weakened: the states surrendered the power to set their own indirect taxes and got a vote in a Council where the Union’s share is decisive on any question the Union cares about; the compensation promised for that surrender became a dispute; and the body that used to negotiate Plan allocations with states no longer allocates anything.
What this book will say: Part Five’s finding was that a formula moves the politics upstream to the design stage and then hides it, which is why the dissenting notes mattered and nobody read them. A Council with a weighted vote does the same thing one layer further up. Whether that is an improvement on a Planning Commission depends entirely on which set of dissenting notes a reader thinks was worse, and nobody has read either.
3.3 — Where the continuity is closest
One area deserves a paragraph of its own because it is the least discussed and the most complete.
Part Thirteen described the Armed Forces (Special Powers) Act, its descent from a colonial ordinance of August 1942, the 1972 amendment allowing the centre to declare an area disturbed without state consent, and the Jeevan Reddy Committee’s recommendation in 2005 that it be repealed, which was not acted on.
That Act remained in force in parts of Nagaland, Manipur, Arunachal Pradesh and Assam into March 2026.
Its area of application has been reduced repeatedly since 2014 — a genuine change, and it goes in the credit column of Chapter Two by reference. Withdrawal has continued the pattern begun in Mizoram in the 1980s, Tripura in 2015 and Meghalaya in 2018.
The statute itself is unamended. Section 6, which bars prosecution of personnel without central sanction, is the same section Part Thirteen described. In December 2021, after fourteen civilians were killed at Mon in Nagaland, the state cabinet recommended prosecution and sanction was refused.
Sixty-eight years, six Prime Ministers, four parties in office at the Centre, one committee recommending repeal in 2005, and the section that matters most is untouched.
Which is the sharpest continuity in this part and the least discussed.
The employment guarantee, the food law, the identity system, the transfer mechanism and the auction rules were all built before 2014 and kept after it. Several were expanded considerably.
One law was attacked and survived: the 2013 land acquisition Act, which an ordinance tried three times to weaken and could not.
And the closest continuity of all is not a welfare programme. Section 6 of the Armed Forces Special Powers Act — the sanction bar — is unamended after sixty-eight years, and sanction was refused at Mon in December 2021.
4The Instruments
Part Two of this series listed four forward promises and named the parts that owed them. Three were paid by Part Nine. The fourth said that Part Eighteen owed the account of Article 370’s removal in 2019. This is it.
This chapter is not about whether the removal was right. That is a live political dispute on which this book takes no position and Chapter Ten says so. It is about how it was done, because the how is what the four tests measure.
4.1 — What happened
On 5 August 2019 the President issued Constitutional Order 272, which amended Article 367 — the interpretation clause of the Constitution — so that the words “Constituent Assembly of the State” in Article 370 would be read as “Legislative Assembly of the State”. Constitutional Order 273 followed, and Article 370 ceased to operate. The Jammu and Kashmir Reorganisation Act divided the state into two Union Territories.
Jammu and Kashmir was at that time under President’s Rule. It had no Legislative Assembly. Parliament acted in place of the state legislature.
Article 370 contained its own removal clause. The President could declare it inoperative — but the clause required the recommendation of the Constituent Assembly of Jammu and Kashmir.
That body had dissolved itself in 1957. So on the face of the text, the condition for removal had become impossible to satisfy: the only body that could recommend it no longer existed.
The 2019 orders solved that by redefining the words. Article 367 is the clause that tells you how to read the rest of the Constitution, and amending it changed what “Constituent Assembly” meant in Article 370 — to “Legislative Assembly”, a body which existed in law and which, at that moment, was under Parliament’s powers because the state was under President’s Rule.
4.2 — What the Supreme Court held
On 11 December 2023 a five-judge Constitution Bench under Chief Justice D. Y. Chandrachud upheld the abrogation, unanimously.
It held that Article 370 was a temporary provision, that the President’s power under it survived the dissolution of the Constituent Assembly, and that the State of Jammu and Kashmir retained no element of internal sovereignty distinguishing it from other states. It declined to rule on the constitutionality of the reorganisation into Union Territories, on the Union’s undertaking that statehood would be restored, and directed that elections be held.
Part Two of this series recorded that Article 370 was headed “temporary”. Part Seven traced its hollowing-out by Presidential Order across four decades. This book’s own account of the provision, written before the judgment was examined for this part, is closer to the Court’s than to that of the provision’s defenders, and that is recorded here because it would be dishonest to discover it now.
For the removal: the provision was headed temporary, was described as such by its own architects, had been progressively emptied by Presidential Orders under governments of every party since 1954, and a five-judge bench of the Supreme Court has now held its removal lawful. A special constitutional status maintained for seventy years by a provision everybody agreed was temporary is not a settled arrangement; it is a deferred decision.
Against the method: the objection is not to the outcome but to the route. A state’s special status was ended while that state had no legislature, by amending the interpretation clause of the Constitution to change the meaning of a word, with Parliament acting in place of the assembly whose consent the text required. Critics argue this establishes that a state’s constitutional position can be altered while its own legislature is suspended.
What the Court itself said about that is worth having: it held that under Article 3 the Union ought to have referred the reorganisation proposal to the state legislature for its views, while declining to decide the consequences.
This book takes no position on whether the removal was right. It is a contested political question, a court has ruled on its legality, and Chapter One said this part would not adjudicate matters of that kind.
What it can say about the mechanism, it can say with unusual confidence.
Everything in the two sections above is documentary and none of it depends on any account by a participant.
Constitutional Orders 272 and 273 are published instruments with dates on them. The Reorganisation Act is on the statute book. The proclamation under Article 356 that put the state under President’s Rule is a published order. The judgment of 11 December 2023 runs to several hundred pages and is reported.
That is unusual for this series and it is why this chapter can be confident about the mechanism while saying nothing about the merits. Part Fourteen had to build its central charge out of one affidavit and one lawyer’s memoir. Part Fifteen’s most-quoted figure comes from a report that has never been published.
Here the whole apparatus is public, dated, and can be read by anybody who wants to check whether this chapter has described it accurately.
4.3 — The instrument test
What this part can do is apply the second of Chapter One’s four tests: what tool was used, who built it, and had it been used before.
Everybody assumes that an instrument belongs to whoever built it — that a power created by one party is that party’s power, and that its use by another is a departure.
Set out what was used on 5 August 2019.
A Presidential Order under Article 370 itself. Part Seven of this series counted these across four decades and found the provision hollowed out by exactly this mechanism, under Congress governments, while the word “temporary” sat at the top of it.
President’s Rule under Article 356. Part Three described this power being turned into an instrument in Kerala in 1959. By 1992 it had been used around ninety times. Part Sixteen recorded the Supreme Court restricting it in 1994.
An amendment to Article 367, the interpretation clause. Part Three’s finding about the First Amendment of 1951 was that its durable product was a procedure rather than a policy — a demonstration that a judicial reading could be reversed by amendment — and that the procedure had no subject matter, which is to say it would work for anything.
Every one of the three instruments used in 2019 was built by Congress governments and charged against them in this series. Not one was created after 2014.
Part Nine asked the reader to name an instrument used after 25 June 1975 that did not exist before 12 June. Part Ten found not one Punjab instrument that was a Punjab instrument. Part Sixteen found the coal method dating from 1993 and struck down across five governments. This is the twelfth instance of the same finding and the first in which the government using the tools is not the one that made them.
Which is the whole reason this part exists. A finding that only ever appeared when Congress was the subject would have been a fact about the author. It appears here too, more sharply than anywhere else.
4.4 — Two other instruments
Two more, stated briefly, because the pattern is the same and repetition would be padding.
Ordinances. Part Sixteen described the ordinance power — a law made by the executive when Parliament is not sitting — and the episode of September 2013. It has been used since 2014, as it was before, including for the attempted amendment of the 2013 land law described in Chapter Three. The power dates from the Government of India Act 1935, which Part Two identified as the source of much of what survived least changed into the Constitution: the machinery of control.
Disruption and the shortened session. Part Sixteen found the 2009–2014 Lok Sabha the least productive full-term Parliament in Indian history, and found that disruption is the only instrument the rules give an opposition. That finding was explicitly about the institution rather than the party, and it has held: the tactic has been used by whoever is out of power, and the number of days Parliament sits has continued its long decline under governments of both descriptions.
The Lok Sabha sat for well over a hundred days a year in the 1950s. It now sits for a fraction of that.
That decline runs continuously across seventy years and every government in this series. It is the single clearest example of something that is nobody’s policy and everybody’s practice, and no party has ever proposed a minimum-sitting rule while in office.
Part Sixteen’s finding was that disruption is the predictable output of rules that give an opposition one lever. A shorter session is the government’s half of the same arrangement, and it requires no lever at all — only the power to decide when the House meets.
Both of those are properties of the building rather than of anybody in it, which is the finding this chapter keeps arriving at.
On 5 August 2019 Article 370 was ended by a Presidential Order amending the Constitution’s interpretation clause, while the state was under President’s Rule and had no legislature. The Supreme Court upheld it unanimously on 11 December 2023.
This book takes no position on whether it was right. What it can say is about the tools: the Presidential Order, Article 356 and the amendable interpretation clause were all built by Congress governments and all three are charged against them in this series.
Twelfth instance of the same finding. First one where the government using the instrument did not make it.
5The Money
Part Sixteen’s central difficulty was that every allocation question went against the government and every criminal question failed, so that the largest scandal in modern Indian history produced a conclusion this book called too honest to be useful.
This chapter has no such difficulty, and that is why it is short.
5.1 — What the scheme was
The Electoral Bond Scheme was introduced by the Ministry of Finance on 2 January 2018, having been announced in the Budget of 2017–18 as a means of cleaning up political funding.
A bearer instrument sold by the State Bank of India in fixed denominations. A donor bought one and gave it to a political party, which encashed it within fifteen days.
The bank knew who bought it. Nobody else did — not the Election Commission, not the tax authorities in the ordinary course, and not the public.
Accompanying amendments removed the cap on how much a company could donate, removed the requirement that a company disclose which party it had given to, and removed the requirement that a company have been profitable.
Only parties with more than one per cent of the vote at the previous general or assembly election could receive them.
Most Supreme Court cases are heard by two or three judges. A case raising a substantial question about the interpretation of the Constitution goes to a bench of at least five, called a Constitution Bench.
The number matters for a reason that has run through this whole series. A judgment can only be overturned by a larger bench than the one that gave it. Part Fourteen’s Sita Soren case took twelve years to reach seven judges, because the ruling it had to overturn had been given by five.
So the size of a bench is a statement about how hard the finding will be to undo. Three of the largest questions in this part went to five judges each — Article 370 in December 2023, the electoral bonds in February 2024, and demonetisation in January 2023.
And unanimity is the second half of it. Part Sixteen’s coal judgment was three judges; the spectrum judgment two. Part Eight’s Kesavananda Bharati was thirteen judges and seven to six — the narrowest possible margin on the most consequential question ever put to the Court.
Five judges agreeing without a dissent is, in Indian constitutional practice, about as settled as a finding gets short of a larger bench being convened to disturb it. That is what happened on 15 February 2024, and a review was dismissed seven months later.
5.2 — What the Supreme Court held
Court-found. A five-judge Constitution Bench, unanimously, struck the scheme down as unconstitutional. Bench: Chief Justice D. Y. Chandrachud, with Khanna, Gavai, Pardiwala and Misra JJ.
The ground: anonymity violated the voter’s right to information under Article 19(1)(a). Information about who funds a party is essential to voting effectively.
Also struck down: the removal of the cap on corporate donations, held to be manifestly arbitrary and in breach of Article 14.
Directions: sale to stop immediately; the State Bank to give the Election Commission the purchaser and recipient details of every bond since 12 April 2019; the Commission to publish them.
A review petition was dismissed in September 2024.
When the data was published, the total encashed was around ₹16,518 crore. In his concurring judgment Justice Khanna observed that the party-wise figures showed donations disproportionately favouring the party in power at the Centre over a continuous period, and that the majority of contributions had gone to parties in power at the Centre and in the states.
₹16,518 crore is not a presumptive figure. It is the sum that was bought, given and encashed, and every rupee of it is now attributable to a named purchaser and a named recipient because a court ordered it published.
Part Sixteen spent a whole chapter establishing that the ₹1.76 lakh crore attached to the spectrum allocation was a calculation of what a different method might have raised, and that the prosecuting agency’s own figure was ₹30,984 crore.
Compare the two kinds of number. One is an estimate of a road not taken. The other is a bank statement.
That is why Chapter Three’s table has a row for it, and why this book — which spent Part Sixteen defending an auditor against people who had misread him — treats this figure differently from that one.
5.3 — Measured on the same ruler
Now apply Part One’s grading scale, which is the whole point of this part.
| Spectrum, 2008 | Coal, 1993–2010 | Electoral bonds, 2018 | |
|---|---|---|---|
| Who found it | CAG, then Supreme Court | CAG, then Supreme Court | Supreme Court |
| What was found | Allocation method arbitrary; 122 licences cancelled | Allocations arbitrary and illegal since 1993; 214 blocks cancelled | The statute and the scheme themselves unconstitutional |
| Unanimity | Two judges | Three judges | Five judges, unanimous |
| Who benefited | Companies; no minister convicted | Companies; officials convicted, ministers not | Political parties, disproportionately the governing one, on the Court’s own finding |
| Presumptive or actual | Presumptive loss; Part Sixteen Chapter Three | Presumptive loss | Actual sums, actually received, now published |
| Criminal finding | All acquitted 2017; appeal pending | Convictions of officials; several later acquittals | None. No prosecution has followed |
Read the fifth row. Part Sixteen spent an entire chapter explaining that a presumptive loss is not money anybody received. The sums in the third column are money that was received, by named recipients, and the figures are public because the Court ordered them published.
Everybody assumes that a scandal is something done in breach of the law. That is what the word means, and it is why the previous seventeen parts are full of audits, prosecutions and inquiries.
The electoral bond scheme broke no law. It was the law. It was passed by Parliament as part of a Finance Bill, operated by a state-owned bank, and lawful in every particular for six years — until five judges held that the law itself was unconstitutional.
Part Nine of this series found that almost nothing about the Emergency was unconstitutional — every instrument was in the document, every threshold met, every procedure followed — and said that this was worse than the alternative, and was a finding about the drafters as much as about the government.
This is that finding again, forty-three years later, and it is the sharpest single vindication of this series’ method in eighteen parts. A book that measured only crimes would have nothing to say about either the Emergency or the bonds, because in each case the thing complained of was lawful when done.
Which is why Part One built a scale with Court-found on it and not only Convicted. The most consequential things governments do to their own accountability are usually legislated, not concealed.
There is a real defence of the scheme and it has to be put at its strongest, because it was made openly and it is not frivolous.
The defence of the scheme, at full strength: before 2018, Indian political funding was overwhelmingly in cash, untraceable and unbanked. The bonds moved a large share of it into the banking system, where every rupee was recorded by a bank, paid by cheque and reported. Donor anonymity was the price of getting money out of suitcases, and it was a price the government said openly it was paying. The Union Home Minister argued after the judgment that the effect of striking it down would be to return political funding to cash.
That argument is serious and Part Fourteen’s Chapter Eight is evidence for it — one crore rupees appeared on the floor of the Lok Sabha in 2008 and the state could never trace where it came from.
Where it fails, on the Court’s reasoning rather than this book’s: the Court did not hold that anonymity was undesirable. It held that the voter’s right to know outweighed it, and — this is the part the defence does not answer — that a less restrictive means was available, since money can be banked and recorded without being hidden from the electorate. The scheme achieved traceability for the bank and opacity for everybody else, and only the second of those was constitutionally objectionable.
And on the removal of the corporate donation cap, which had nothing to do with cash, the Court’s finding of manifest arbitrariness is unanswered by any argument the defence has offered.
Which leaves the finding where the Court left it, and it is the largest single item in this part.
5.4 — And what replaced it
A scheme that carried a very large share of national political funding was ended by court order in February 2024. The obvious next question is what took its place, and this book cannot answer it.
What is known: the bonds stopped. The purchaser and recipient data going back to April 2019 was published. Parties reverted to the pre-2018 channels — declared donations above the disclosure threshold, electoral trusts, and cash below it.
What is not known, and this is the point: whether the total volume of political money fell, stayed the same, or moved somewhere less visible.
The Union Home Minister argued after the judgment that the effect would be to send funding back into cash. If that is right, then the Court removed an opaque legal channel and the money went to an opaque illegal one, and the transparency gained is smaller than it looks.
If it is wrong, the disclosure regime is working and the judgment did what it was meant to.
Nobody has published the comparison — declared party income by source, before 2018, during the bond years, and since February 2024 — and every component of it is in Election Commission filings that parties are already obliged to make.
This is Chapter Ten’s kind of item and it goes there too. It is worth flagging here because it is the single most important open question about the largest finding in this part, and because it is the fifth time in this series that a question everybody argues about could be settled by arithmetic on data that is already public.
15 February 2024. Five judges, unanimously, struck down the electoral bond scheme as unconstitutional. A review was dismissed in September 2024. About ₹16,518 crore had been encashed.
Measured on this series’ own scale, that is a stronger finding than either 2G or Coalgate: unanimous, on the statute itself rather than on a method, and about actual money received by named recipients rather than a presumptive loss.
And the deepest point: the scheme broke no law. It was the law. Part Nine found the same thing about the Emergency, and said it was worse than the alternative.
6What Was Said
Part Seventeen ended by promising that this part would carry the equivalent list of statements under the identical rule — date, venue and operative wording — and said in terms that if it did not, Part Seventeen’s fifth defence point was a rhetorical concession rather than a real one. This chapter pays that.
6.1 — The same three tests
Part Seventeen began with roughly forty candidates and eleven survived. This chapter began with a comparable working list and the survival rate was almost identical: the great majority failed for want of a date, a venue, or a wording that was not somebody’s translation of somebody’s paraphrase.
Roughly three in four failed, which is what Part Seventeen found.
The reasons were the same reasons. A phrase circulating detached from its speech. A remark known only from a screenshot. A Hindi sentence available in four English versions, none of them sourced. A statement everybody is certain about and nobody can date.
That is the finding of this chapter and it arrives before any statement does. The failure rate for political quotations in India is not a property of one party’s supporters. It is a property of a country with no public archive of political speech, which is what Part Seventeen’s Chapter Nine was about and why its only recommendation was a filing job.
How exactly they failed is worth setting out, because the pattern is the argument.
The failure rate is not an impression. The candidates failed in four identifiable ways and they are the same four Part Seventeen found.
No date. A phrase attributed to a named person with no day attached. Roughly half the working list.
No venue. A dated remark with no record of where it was made or to whom — which matters, because a sentence in Parliament, at a rally and to a reporter are three different acts.
Translation of a translation. A Hindi or regional-language sentence available only in English versions that disagree with one another, with no audio.
Clipped. A sentence whose meaning turns on a pronoun or a referent that is not in the circulating version — the exact defect Part Seventeen’s Chapter Two found in the most-quoted political sentence in modern India.
Not one candidate failed because it was shown to be false. They failed because nobody had written down when, where or in what language. That is the finding, and it is a fact about the country rather than about anybody’s supporters.
6.2 — The one that meets every test
One statement from these twelve years satisfies all three tests at the highest standard available outside Parliament, and it is the one this chapter is built on — partly because it is well evidenced and partly because it is the exact counterpart of Part Seventeen’s Chapter Two.
Date: 21 April 2024. Venue: an election rally at Banswara, Rajasthan, during the general election campaign. Speaker: the Prime Minister, Narendra Modi.
Characterising the Congress manifesto, he said that when the Congress-led government was in power it had said Muslims have the first right over the country’s assets; that this meant the wealth would be distributed to “those who have more children” and to “infiltrators”; and, of a proposed survey of property, that such a mindset would not spare women’s mangalsutras.
Established to the highest standard available outside Parliament: on camera, at a named venue, on a fixed date, circulated by a news agency the same day, and never disputed by the speaker or his party.
The premise of the passage is the statement Part Seventeen’s Chapter Two examined at length: the sentence of 9 December 2006 about first claim on resources.
Part Seventeen found that sentence real, and the reading now attached to it not supportable on the full paragraph — a finding this book made in the course of dismantling the best material it had against Congress.
6.3 — What was left out, and why that matters here
Part Seventeen’s Chapter Ten named a statement it had excluded — one the author was confident had been made and could not pin to a day and a town — and said that describing a rule is cheap and this was the rule costing something. The same discipline is owed here and this section is it.
The working list for this chapter contained remarks attributed to senior figures of this government which are as widely known as the one Part Seventeen dropped. Several were the subject of formal complaints to the Election Commission or to police. They are not in this chapter.
They failed the same way. A clip circulating without the surrounding minutes. A remark whose date is given as a month. A sentence spoken in Hindi at a state rally and known nationally in an English rendering nobody can trace. A phrase whose meaning turns entirely on what preceded it, and the preceding sentence is not in the version anybody has.
A complaint to the Election Commission is not a substitute for a date and a venue. It establishes that somebody objected. It does not establish what was said.
If those remarks went in on the strength of everybody knowing them, then Part Seventeen’s Chapter Two collapses — because that is the identical basis on which every fabrication in Indian politics travels, and this book spent that chapter dismantling the best material it had against Congress on exactly that ground.
Which produces the honest position of this chapter, and it is uncomfortable for anybody hoping for a tally.
Part Seventeen: roughly forty candidates in, eleven out. Part Eighteen: a comparable list in, one out.
A reader could take that as evidence that this government’s leaders have said less that is objectionable. That reading is not available and this book will not permit it.
Part Seventeen had the advantage of decades. Its eleven survivors include a parliamentary exchange from 1961, a speech from 1984 whose footage surfaced in 2015, and a press conference from 2011 preserved in a magazine parenthesis. Time is what turns a remark into a record, and Chapter One’s asymmetry box said this would happen.
The correct conclusion is the one Chapter Eight reaches about everything else: the difference is in the archive, not in the speakers. And there is no archive, which is Part Seventeen’s Chapter Nine and the only recommendation either part makes.
6.4 — What followed
The Congress, the CPI(M) and the CPI(ML) Liberation complained to the Election Commission. Complaints were also made to police and to the state chief electoral officer.
On 25 April 2024 the Election Commission issued a notice — to the president of the Bharatiya Janata Party, not to the speaker — referring to speeches by “some of the party’s star campaigners” and requiring a response. A near-identical letter went the same day to the president of the Congress over a complaint about a speech by Rahul Gandhi at Kottayam.
Both letters are dated 25 April 2024, are signed by the same officer, and are near-identical in form. That is not a detail; it is the evidence.
The Commission’s practice in this instance was to address the party rather than the individual, and to issue paired notices to both national parties on the same day. Its critics argued that this avoided naming a named person against whom named complaints had been made. Its defenders argued that the Model Code binds parties and that even-handed treatment of both is the correct posture for a referee.
Both readings are available on the same documents, which is why this book gives the documents and not a conclusion. No finding was made against anybody and no proceeding followed.
What the speech itself is worth as evidence needs both cases put at full strength, and neither is weak.
The defence of the speech: the Congress manifesto did propose a nationwide socio-economic survey, and redistribution was under discussion in the campaign by senior figures of that party. A politician characterising an opponent’s manifesto in hostile terms is doing ordinary electoral work. No religion was named in the passage; “infiltrator” is a term for illegal immigration, on which this government has legislated; and no court or commission has found the speech unlawful.
The case against: the premise was a claim about what a former Prime Minister said, and Part Seventeen established that the claim does not survive a reading of the paragraph. The phrase about having more children is, in Indian political usage, a well-understood signifier requiring no naming. And the complaint was made by three parties and by civil liberties organisations under specific sections of the Representation of the People Act.
What this book will say, and it is narrow: the words are established, the premise is one this series has itself shown to be unsupportable, and no body with the power to make a finding made one. Alleged as to unlawfulness. Established as to what was said.
6.5 — And the same finding as Part Seventeen
A reader arriving at this chapter will assume its purpose is to establish equivalence — that a bad sentence on one side answers a bad sentence on the other and the ledger balances. Part Seventeen’s Chapter Eight already destroyed that idea and the destruction applies here: words and acts do not correlate. The cruellest sentence in Part Seventeen came from a government that then appointed a commission of inquiry; the most honourable from one that then did nothing for four decades.
So this chapter cannot show that this government is better or worse than its predecessor, and it is not trying to. What it shows is that the method transfers. Apply three tests to one party’s statements and three-quarters fail; apply them to another’s and three-quarters fail. Find one statement that meets every test and the analysis available is identical in shape.
Part Seventeen said its own chapter was the most quotable and least probative in a nineteen-part audit. That was true of it and it is true of this one, and a reader who takes only this chapter from this part has taken the least useful thing in eighteen.
Which is a smaller conclusion than a statements chapter usually reaches, and it is the honest one.
Part Seventeen promised this chapter and this chapter exists. Same three tests, same failure rate: roughly three in four candidates could not be dated, placed or sourced.
One meets every test: 21 April 2024, Banswara. Its premise is the 2006 sentence Part Seventeen showed cannot bear the reading now put on it.
The Election Commission wrote to the party rather than the speaker, and sent a near-identical letter to the other party the same day. No finding was made against anybody.
7What Has Not Moved
Seven parts of this series ended with a table of ordinary laws that would have prevented the failure just described, and recorded that none of them had been made. This chapter takes every one of those rules and asks a single question: what has happened to it since 2014?
This is the third of Chapter One’s four tests, and it is the one that does most of the work in this part.
7.1 — The ledger
| The rule | From | Status |
|---|---|---|
| A statutory duty on named officials to act during communal violence; command responsibility; removal of the sanction requirement | Parts Eleven, Fifteen, Sixteen | Not enacted. Drafted 2005 and 2011 under the previous government; not brought forward since |
| Compensation as of right for wrongful prosecution after prolonged custody | Part Fifteen | Not enacted. Recommended by the Law Commission in 2018 |
| Prosecution sanction decided by a body independent of the executive, with reasons and appeal | Parts Thirteen, Fifteen, Sixteen | Not enacted. Section 6 of the Armed Forces Special Powers Act unamended; sanction refused at Mon, December 2021 |
| Statutory outer limit on pre-trial custody under special-powers statutes | Part Fifteen | Not enacted |
| Repeal of confession provisions in state organised-crime statutes | Part Fifteen | Not attempted. State statutes unchanged |
| Terms of reference of an inquiry laid before Parliament and enlargeable by the commission | Part Fourteen | Never proposed by anybody |
| Statutory limit on the life of a commission of inquiry; extensions approved by Parliament | Part Fourteen | Not enacted |
| Time limits for corruption trials involving public office | Parts Fourteen, Fifteen | Not enacted. Part Sixteen’s spectrum appeal took six years to be admitted |
| A named officer statutorily responsible for aggregate threat assessment | Part Fifteen | Partly built by executive action after 2008; never legislated |
| A statutory deadline for constituting any body created by an Act | Part Sixteen | Not enacted — though the Lokpal itself was constituted in March 2019 |
| Audit reports to state the counterfactual used and the range produced | Part Sixteen | Not enacted |
| A statutory default requiring auction of finite natural resources | Part Sixteen | Substantially done. Auctions the default for coal from 2015 and for spectrum since 2012 |
| A published national register of terrorist incidents and their judicial outcomes | Part Fifteen | Does not exist |
| A public verbatim archive of political speech by officeholders | Part Seventeen | Does not exist |
| A statutory threshold before an official is charged over a policy decision | Part Sixteen | Partly attempted by an amendment in 2018; effect contested |
Fifteen rules. Two substantially done, two partly, eleven not. Every one of them is ordinary legislation and none requires a constitutional amendment.
A table asserting that fifteen things did not happen is the hardest kind of claim to source, and a reader is entitled to know how it was built.
Each row was checked two ways: against the statute book, for whether such a law exists; and against the record of bills introduced in either House, for whether one was ever brought and failed. The distinction matters enormously. A bill introduced and defeated is a political failure. A bill never introduced is something else, and every one of the eleven is the second kind.
The weakness is obvious and this book states it: proving that nothing happened is proving a negative, and a single missed bill would make a row wrong. Chapter Ten repeats the invitation to correct it.
What raises confidence is the pattern rather than any row. Eleven separate proposals, from seven different parts, across two governments of opposite politics, and not one of them was brought and lost. Fifteen coincidences would be a coincidence. Eleven identical outcomes are a finding.
7.2 — What the four that moved had in common
Four of the fifteen went somewhere. Two substantially, two partly. Asking what those four share turns out to be more useful than asking why the eleven did not move.
| The rule | What made it move |
|---|---|
| Auction as the default for natural resources | A Supreme Court order. The coal blocks were cancelled in September 2014 and something had to replace the method within six months |
| The Lokpal constituted | Supreme Court pressure, applied twice, on a body that already existed in law and was conspicuously empty |
| A named officer for threat assessment | A catastrophe. Built by executive action in the months after November 2008 |
| A threshold before charging an official over a policy decision | A constituency. The civil service, which had watched a Coal Secretary convicted and stopped signing things |
A court order, a court order, a catastrophe, and an organised group with a direct interest. Not one of the four was made because it was a good idea.
Now look at the eleven that did not move and ask which of those four things they have.
Compensation for wrongful prosecution: no court order, no catastrophe, and its constituency is people who have just been released from jail.
The communal violence law: no court order, and its constituency is the families in Part Eleven, who are a small minority in a small number of constituencies.
A register of terrorist incidents: nobody’s interest at all. An archive of political speech: an anti-constituency, since it constrains everybody who would have to create it.
Every rule that got made had somebody powerful behind it. Every rule that did not was for people with no way of insisting.
Part Fifteen’s Chapter Eight found that an electorate can deliver a fair verdict on how a government treated most people and no verdict at all on how it treated a few. This table is the legislative half of the same fact, and it holds across both governments identically.
7.3 — And the documents
The fourth of Chapter One’s tests. Five parts of this series ended by naming something the state holds and will not release.
| The document | From | Withheld since | Status |
|---|---|---|---|
| The Henderson Brooks–Bhagat report on 1962 | Part Six | 1963 | Not released. Part I published abroad in 2014 |
| Home Ministry file notes, 27 November – 6 December 1992 | Part Fourteen | 1992 | Not released |
| The Pradhan Committee report on 26 November 2008 | Part Fifteen | 2009 | Not published. The stated reason expired in 2012 |
| The freight equalisation calculation | Part Five | — | Never done. Not withheld; nobody has attempted it |
| The audit of the audits — presumptive estimates against realised outcomes | Part Sixteen | — | Never done |
Three withheld, two never attempted. Sixty-three years, thirty-four years and seventeen years respectively for the first three. Governments of every description have held all three.
Twenty items across the two tables. Fifteen have not moved at all.
Of those fifteen, not one was blocked by an opposition, defeated in a vote, or struck down by a court. Nothing stopped them. Nobody brought them.
Part Thirteen described an addition nobody had made, where every component existed, nothing was concealed, and no decision existed that anybody had to defend. Part Fifteen found seventeen intelligence alerts forwarded and never assessed, because assessing them was nobody’s job.
This table is that finding at the scale of a whole state, and it is the single most important thing in this part.
And there is a distinction inside that sentence which is the whole of why the failure is invisible.
There are three ways a law does not get made and they are not the same thing.
Defeated: a bill was brought and lost a vote. That is democracy working and somebody can be held to the result.
Lapsed: a bill was brought, referred to a committee or left pending, and died when the House was dissolved. That is a process failure and the record shows who let it happen.
Never introduced: nothing was brought at all. There is no vote, no debate, no committee report and nobody to ask.
Every one of the eleven unmade rules in Table 4 is the third kind. That is the single most important fact about the table, and it is the reason the failure is invisible: a defeat produces a record and an absence produces nothing.
7.4 — One item that moved in the wrong direction
Honesty requires this row to be given its own section rather than a line in a table.
Part Fifteen described the Right to Information Act of 2005 as the largest single transfer of power to ordinary people in this series, and noted that the best evidence it had teeth was what later governments tried to do to it.
In 2019 the Act was amended. The tenure, salaries and conditions of service of the Information Commissioners — the officials with the power to fine a department for refusing information — were removed from the statute and made subject to rules prescribed by the central government.
The defence: the 2005 Act had equated Information Commissioners with Election Commissioners in status and tenure, which was an anomaly, since one body conducts elections and the other adjudicates disclosure requests. Rationalising service conditions is ordinary administrative tidying and does not touch a single power of the Commission.
The reply: Part Fifteen’s Word Box said it in advance and without knowing this would happen — without a penalty on a named person, a duty to disclose is a suggestion. The penalty is imposed by an official whose tenure and salary are now set by the body most often on the receiving end of it.
Part Thirteen’s finding, made about tribal land: a right is worth the independence of whoever enforces it. Part Sixteen’s Chapter Six found a prosecution sanction that did not test evidence but relocated a decision to somebody political.
This is the third instance of the same mechanism and the only one where a government amended a statute to produce it.
Where the defence keeps something: the Act’s substantive rights are unamended and it is still used at very large scale. This is a change to the enforcer, not to the right.
Which is the whole ledger, and it is the part of this book that will be read least and matters most.
Twenty items — fifteen rules and five documents — that this series said would prevent a failure or settle a question. Fifteen have not moved.
Not one was blocked, defeated or struck down. Nobody brought them.
Two moved forward: auctions became the default for natural resources, and the Lokpal was finally constituted in 2019. One moved backwards — the 2019 amendment putting the tenure of Information Commissioners in the government’s hands.
8What the Ruler Actually Shows
Chapter One said this part exists to test the ruler rather than to judge the government, and that the only way to test a method is to point it at something else. This chapter reports what came out.
8.1 — The four tests, scored
| Test | Applied to Congress, Parts 1–17 | Applied to 2014–2026 | Does the ruler transfer? |
|---|---|---|---|
| Grade | Produced findings ranging from convictions to fabrications; forced concessions the author did not want to make | Produced one unanimous Supreme Court finding stronger than any in Part Sixteen, and a credit chapter larger than expected | Yes, cleanly |
| Instrument | Found eleven times that the tool predated the government using it | Found it a twelfth time, and the tools were built by the party this book prosecutes | Yes, and more sharply |
| Rule | Fifteen rules proposed across seven parts; none made | Two made, two partly, eleven not | Yes |
| Document | Five named; three withheld, two never attempted | All five unchanged | Yes |
Four tests, four transfers. That is the answer to the question this part was written to ask, and it is the reason the rest of the series can be read as something other than a campaign document.
8.2 — What that means, stated carefully
It does not mean the two governments are the same. It does not mean their records are comparable in weight, and this book has said four times that it has no denominator with which to compare them.
It means something narrower and more useful.
This series has assumed for seventeen parts that it was auditing a party. The title says so. The selection says so. Every chapter is about something a Congress government did or failed to do.
Apply the ruler to a different government and three of the four tests return the same answer. The instruments predate everybody. The rules are not made by anybody. The documents are released by nobody.
Which means that for three of the four tests, the party was never the variable. Seventeen parts of evidence about Congress turn out to be, in substantial part, evidence about the Indian state — about a legislature that does not legislate its own accountability, an executive that does not release its own files, and a set of instruments that every government inherits and none dismantles.
That is uncomfortable in both directions and this book will not soften either. It weakens every charge in this series that depends on the party being the cause. And it strengthens, enormously, every finding about the machine — because a finding that survives a change of government is a finding, and a finding that does not is a coincidence.
Part One’s founding box said that treating Congress as one continuous actor across seventy-nine years was a choice made before the analysis began, and that it cut both ways. Eighteen parts in, here is the fuller version: the continuous actor is not the party. It is the state, and the party built most of it.
Which is the only sentence in this series that is simultaneously a defence of this government, an indictment of Congress, and a reason to think the whole exercise was worth doing.
All of which invites an obvious question about who is grading the grader.
The claim that a method transfers is exactly the sort of claim an author would make about his own method, so it needs an external check.
Here is one. The method produced results the author did not want and did not predict, in both directions.
Chapter Two is longer than intended, and the author’s note records that it was lengthened twice after the first draft came out thin. Chapter Five found a Supreme Court judgment that, measured on Part One’s own scale, is stronger than the central case of Part Sixteen — which is a worse result for this government than this book expected to reach. And Chapter Eight’s finding damages Parts One to Seventeen more than it damages anything here.
A method that returns only what its user wanted is not a method. Three results in three directions, none of them the intended one, is the best available evidence that these four tests are doing something other than expressing an opinion.
8.3 — The one test where the party was the variable
The first test — the grade — is the exception, and it is worth saying exactly how.
The specific findings differ. Parts One to Seventeen contain an Emergency, a demolition, five days in Delhi, a wave of wrongful arrests, two audit reports and a series of prosecutions that mostly failed. This part contains a tax reform, an insolvency law, a constitutional change upheld by five judges, and a political funding statute struck down by five judges.
Those are not the same events and nothing in this chapter says they are.
What the grade test measures is whether the method can find things it did not go looking for.
It found a credit chapter here larger than this author expected, which is evidence the method is not simply hostile.
It found, on the scale this book built in Part One, that the electoral bond judgment is a stronger finding than either 2G or Coalgate — unanimous, against the statute itself, and about money actually received rather than presumptively forgone. That is evidence the method is not simply friendly either.
A ruler that produced only good news here, or only bad, would have failed. It produced both, and the author’s own expectation was wrong about which chapter would be longer.
8.4 — What this does to Part Nineteen
Part Nineteen consolidates this series: the full evidence-grade table, the state ledger, the folklore column, and the strongest case the defence can still make. Chapter Eight changes what that part has to be.
Part Nineteen was designed on the assumption that a consolidation is an addition — that the way to end a nineteen-part audit is to gather every finding into one place and present the total.
Chapter Eight makes that impossible. If three of the four tests were measuring the state rather than the party, then a consolidated list of Congress failures is not a total of anything. It is two different things stacked on top of each other: a set of specific acts by specific governments, and a set of structural facts about a country that outlived all of them.
Those cannot be added. Adding them is what a campaign document does.
So Part Nineteen has to have two columns rather than one, and the division between them is not the evidence grade. It is: does this finding survive a change of government? Part Eighteen has now supplied the test for that on twenty separate items and the answer for fifteen of them is yes.
Which means the last part of this series will be shorter on charges than it was going to be, and longer on machinery. That is a worse book for anybody who came here to read a prosecution and a better one for anybody who wanted to know why nothing changes.
8.5 — What this does to Parts One to Seventeen
Three consequences, and this book would rather name them than have a reader find them.
Every “they did this” becomes weaker. Where a part charged a Congress government with using an instrument, Chapter Four shows the instrument outliving the party. The charge that survives is narrower: not that they used it, but that they built it and nobody has dismantled it.
Every “nothing was done” becomes stronger. The seven rule tables were the least-read pages in this series. Chapter Seven shows that eleven of fifteen rules remain unmade under a second government with different politics and, for a decade, a larger majority. A failure that survives a change of government is a structural finding rather than a partisan one.
And the parts that rest on specific acts are untouched. Nothing in this chapter affects Part Nine, Part Ten, Part Eleven or Part Fourteen’s Chapter Four. Those describe things that were done, on dates, by named governments, and no comparison alters them.
Four tests. All four transfer. The ruler works on something other than the thing it was built for, which is the only evidence that it is a ruler.
Three of the four return the same answer whichever government is measured. For those three, the party was never the variable — the state was.
Which weakens every charge in this series that needed the party to be the cause, and strengthens every finding about the machinery. A failure that survives a change of government is structural. One that does not is a coincidence.
9What Would Prevent It
Every part of this series ends with this chapter. This is the first one addressed to a government still in office, and that changes what it can honestly say.
9.1 — The defence, at full strength
One. The building record. Chapter Two. A tax reform that had defeated every previous government, an insolvency regime, delivery of basic services at a scale not seriously disputed, and an anti-corruption office finally filled after five years empty.
Two. It inherited the instruments and did not build them. Chapter Four. Every tool used on 5 August 2019 was made by a Congress government and charged against it in this series.
Three. It legislated auctions into permanence. Chapter Seven’s twelfth row. Of the fifteen rules this series proposed, the one with the largest financial consequence has been substantially done.
Four. Twelve years is not fifty-four, and 2026 is not the vantage point from which any of this will eventually be judged. Chapter One’s In Real Terms box, and it applies to the charge column with exactly the force it applies to the credit column.
Five, and it is the one this book has to concede hardest. The most serious finding in this part — the electoral bond judgment — was produced by a court sitting under this government, against this government, and enforced. The scheme stopped. The data was published. The review was dismissed.
The fifth point is the strongest thing any government in this series can say for itself, and it is worth setting against the record of the others.
Part Nine found a government that placed its own Prime Minister’s election beyond the courts by constitutional amendment, and that the amendment was struck down under the basic structure doctrine by a Chief Justice who had opposed that doctrine.
Part Sixteen found a government whose largest scandals produced two Supreme Court judgments against it, which it accepted, and then legislated the remedy.
Part Eighteen finds a government whose political funding statute was struck down unanimously and which complied. The data went to the Election Commission. It was published.
That is a real thing about the Indian constitutional order and it has now survived three governments of very different character. This book records it here rather than in Chapter Ten because it is a defence and belongs where the defence is stated.
9.2 — Where the defence loses
On two things, and both are in Chapter Seven rather than anywhere more dramatic.
Eleven of fifteen rules. A government with a single-party majority for a decade did not bring eleven ordinary laws that this series says would have prevented specific, documented failures. Not one was blocked. Nobody brought them. The most conspicuous is the communal violence law, which is now unenacted across twenty-two years and three governments — Parts Eleven, Fifteen and Sixteen having each carried it forward.
Three documents. The 1963 report, the 1992 file notes and the 2009 committee report have all been held by this government for twelve years and none has been released. The reason given for the third expired in 2012.
To which this part adds the one item that moved backwards, in Chapter Seven’s third section: the 2019 amendment placing the tenure and conditions of Information Commissioners in the hands of the government whose refusals they adjudicate.
Set the two halves of this chapter against each other in the plainest possible terms.
What was done: a tax reform in 2017, an insolvency law in 2016, an office filled in 2019, auctions made permanent from 2015. Four substantial acts in twelve years, and every one of them required legislation, negotiation with states, or both.
What was not done: eleven laws, none of which required negotiation with anybody, none of which was opposed, and none of which was introduced.
This is not a government that could not legislate. It legislated a national tax through a constitutional amendment requiring the consent of every state, which is the hardest thing in the Indian statute book to do. Then it did not bring a bill requiring a police officer to register a case.
The same sentence, with different dates, is true of the government in Parts Fifteen and Sixteen.
9.3 — The rules, addressed to a government in office
This chapter cannot do what the previous seventeen did, which was to list remedies for a failure already complete. Here the list is prospective, and that is a different kind of writing.
| The rule | Political cost of making it | Who it would constrain |
|---|---|---|
| Release the three withheld documents, redacted where genuinely operational | Almost none. All three describe governments of the other party | Nobody now living in office |
| Compensation as of right for wrongful prosecution | Low. Recommended by the Law Commission in 2018 | State police forces, of every party |
| A verbatim archive of political speech | Low in money, high in habit | Every officeholder, equally |
| A national register of terrorist incidents and outcomes | Almost none. The data exists | Investigating agencies |
| Statutory deadline for constituting a body created by an Act | Low. This government constituted the Lokpal | Every future government |
| Prosecution sanction decided independently of the executive | High. It removes a power every government has used | The Union, directly |
| The communal violence law | Very high, and it is the reason nobody has passed it in twenty-two years | State police and their political superiors |
Read the middle column downwards. The items nobody has done are not the difficult ones. Five of the seven cost almost nothing, and four of those five would embarrass a previous government rather than this one.
Everybody assumes that reforms do not happen because they are politically costly. It is the standard explanation and it is applied to every unmade law in this series.
Test it against the table above. The three withheld documents describe the conduct of governments of the other party. Releasing them costs this government nothing and would hand it three genuine indictments of its opponents. It has not done it. Neither did the previous government, which held them for ten years and could have released the 1963 report at no cost at all.
So the cost explanation fails on precisely the cases where it should be easiest.
What is left is worse and duller. These things are not done because doing them is nobody’s job, nobody’s promotion depends on it, and no minister is ever asked about them. A file that is not released produces no news. A law that is not introduced produces no headline. An archive that does not exist has no constituency.
Part Thirteen called this an unmade addition where every component exists and no decision exists that anybody has to defend. Eighteen parts in, that is the most transferable finding in this entire series — it has now appeared in the tribal belt, in an intelligence system, in a courtroom, and in the legislative programme of two consecutive governments.
9.4 — The question turned on this book
Chapter One promised that this chapter would report on whether §3.3 held — the instruction, written before Part One, that if this part read as an advertisement it would retroactively discredit everything before it.
A reader is better placed than the author to judge that, and Chapter Ten records what the author expects. What can be said here is what was done to try.
The credit chapter was placed second, as in every part since Fourteen. The strongest finding against this government was given a chapter of its own and measured explicitly against Part Sixteen’s, on the same scale, in a table, and found stronger. The one contested constitutional question was given both cases at full strength and no verdict. The statements chapter delivered the promise from Part Seventeen and reached the same conclusion Part Seventeen had reached about its own material — that such chapters are the most quotable and least probative in the book.
And Chapter Eight, which is the finding of this part, weakens Parts One to Seventeen more than it weakens anything here. That was not the intention and it is not what this author expected to write.
The defence is real: a large building record, instruments it inherited rather than made, auctions legislated into permanence — and a Supreme Court judgment against it that it complied with.
It loses on eleven unmade rules and three withheld documents, and on the one item that moved backwards in 2019.
And the cost explanation fails. Five of the seven remedies cost almost nothing and four would embarrass somebody else. They are not done because they are nobody’s job.
10An Honest List of What We Do Not Know
This is the longest honest list in the series, because this part covers a government still in office and most of what will be known about it is not knowable yet.
10.1 — What this part did not adjudicate, and why
Chapter One said this part would not rule on live political disputes where no court or commission has. Those are named here so that the omissions are visible rather than silent.
Whether the removal of Article 370 was right. A five-judge bench has held it lawful. Whether it was wise, and what it has meant for the people of Jammu and Kashmir, is a contested political question on which this book takes no position. Chapter Four gives both cases.
Citizenship legislation and the associated register. Contested, litigated, and substantially unresolved at the time of writing. This book has not been able to grade it and has left it out rather than assert.
The condition of institutions. Claims about the independence of investigating agencies, the Election Commission, the press and the courts are made in both directions with great confidence and, on this book’s own standard, are mostly Alleged. Chapter Six’s treatment of one Election Commission notice is a deliberate demonstration of how little a single document establishes.
The farm laws of 2020 and their repeal in 2021. Enacted, protested against for a year, repealed. This book records the sequence and does not grade the merits, which remain contested among agricultural economists who are not aligned with either party.
Demonetisation, November 2016. Upheld by a Supreme Court bench in January 2023 by four to one on the question of the decision-making process. Its economic effects are disputed by serious people on both sides and no official finding settles them.
The objection to this section, and it is the strongest criticism this part will face: that refusing to adjudicate is not neutrality. A book that grades seventeen parts of Congress conduct and then declines to grade the most contested acts of the current government has produced a specific result by omission, and dressed the omission up as rigour.
The reply, which does not fully answer it: every item above fails the tests set out in Part One, and this book has excluded material on that basis before. Part Six graded the Henderson Brooks material Alleged throughout because the state would not confirm it. Part Fourteen refused to assert central direction of the demolition. Part Seventeen dropped roughly twenty-nine statements it believed were real.
Where the objection lands anyway: those exclusions were made about a government out of power for a decade, whose files had partly surfaced. Applying the same rule to a government in office produces systematically more exclusions, because less has surfaced. The rule is neutral and its effect at this moment is not.
This book has no answer to that and does not pretend to. What it has is Chapter Ten of the next series, and the commitment in section 10.5.
10.2 — Where this part may be wrong
The credit chapter may be premature. Chapter Two treats the tax reform and the insolvency code as structural achievements. Part Fourteen took twenty-two years and a Supreme Court judgment to establish what the coal method had been. A reader in 2040 may find Chapter Two naive.
The charge column may be premature in the same direction. Part Sixteen’s central case is still on appeal after eighteen years. Nothing in this part has had time to be tested the way the material in Parts Nine to Sixteen was.
Chapter Seven’s ledger is the strongest thing here and rests on absence. It says fifteen rules were not made. Proving a negative about the legislative record of two decades is harder than it looks, and if a reader finds one of those rows wrong, the table is wrong.
Chapter Six rests on one statement. Part Seventeen rested on eleven. Section 6.3 explains why and argues that the difference is archival rather than substantive, but a reader who thinks a single example cannot carry a chapter is making a fair criticism and this book cannot fully answer it.
The selection of what to compare. Chapter One set four tests, but the choice of which events to apply them to was the author’s. A different writer using the same four tests would produce a different Part Eighteen.
10.3 — The one new calculation
Chapter Five said this belonged here and it does, because without it the next section would overstate the case.
The electoral bond scheme was struck down in February 2024. Nobody has published what happened to political funding afterwards.
Declared party income by source, in three periods: before 2018, during the bond years, and since February 2024.
Every component already exists. Political parties file annual audited accounts and contribution reports with the Election Commission. The bond data itself was published on the Commission’s website by court order in March 2024. Electoral trust filings are public.
What that comparison would show is whether the judgment made political funding more transparent or merely moved it. If declared income fell sharply after February 2024, the money went somewhere the filings do not reach, and the Union Home Minister’s objection to the judgment was correct. If it did not, the disclosure regime is working.
That is the single most consequential open question about the largest finding in this part, and it can be answered by anybody with a spreadsheet and a fortnight.
Part Five ended by naming the calculation that would settle freight equalisation. Parts Fourteen, Fifteen, Sixteen and Seventeen each named another. This is the sixth, and like four of the five before it, it is arithmetic on data that is already public.
10.4 — The documents that would settle the rest
Nothing new. That is the finding.
Every one of the five documents in Chapter Seven’s second table already exists in a government building. Two of the calculations require no document at all — only arithmetic on data that has been public for a decade.
Apart from the calculation in section 10.3, this part names no new missing document — because the missing documents are the same ones named in Parts Five, Six, Fourteen, Fifteen and Sixteen, and they are all still in the same place.
The 1963 report is sixty-three years old. The 1992 file notes are thirty-four. The 2009 committee report is seventeen and the reason given for withholding it expired fourteen years ago.
Three governments of two very different characters have held all three and released none.
10.5 — What the second series is now committed to
This series has promised, in the text of nearly every part, a second series auditing the government of 2014 onwards on the same scale as this one. That promise is now specific enough to be checked, and this section is where it is written down.
Same scale. Not one part. Something comparable in length to the nineteen parts of this series, or the promise was never serious.
Same method. The four tests in Chapter One, the six-grade scale from Part One, the credit chapter placed second, the other side given its strongest case, a rule table and an honest list in every part.
Same treatment of acquittals and of unproved allegations. Part Sixteen refused to treat a failed prosecution as a proved charge. Part Eleven declined to name individuals against whom no body had found. Part Fourteen refused to assert what it could not establish. Those rules protected Congress figures repeatedly, and they will protect this government’s figures identically or the whole exercise was fraudulent.
And the items in section 10.1. Every question this part declined to adjudicate goes into that series and is adjudicated there, on whatever evidence exists by then.
One further commitment, and it is the one that will be hardest to keep.
The second series will have to contain a part like this one — a comparison, applying the same ruler back to the Congress governments of 2004 to 2014 and, by then, to whatever has happened in between. If it does not, it is a campaign document, and this part will have been the last honest thing in either.
10.6 — The four debts, accounted for
The front matter of this part named four things earlier parts had committed it to. This series has always recorded the settling of a debt in a numbered section rather than leaving it to be inferred, and this is that section.
| The debt | From | Where paid | Status |
|---|---|---|---|
| The account of Article 370’s removal in 2019 | Part Two, Ch 9.5 | Chapter Four | Paid. Mechanism, judgment and instrument test given; merits expressly not adjudicated and the omission named in 10.1 |
| The equivalent list of statements, same three tests | Part Seventeen, Ch 10.5 | Chapter Six | Paid, at one statement. Section 6.3 names what was excluded and why; 10.2 records the weakness |
| The status of every rule this series proposed | Parts Nine, Eleven, Thirteen to Seventeen | Chapter Seven | Paid. All fifteen rules and all five documents accounted for in two tables |
| That this part must not read as an advertisement | §3.3 of the working brief | Chapters Five, Seven, and section 10.6 | Reported on rather than claimed. The author cannot certify his own compliance |
Part Two’s four forward promises were paid across Parts Seven, Eight and Nine and recorded in each. Part Twelve’s panchayat promise was paid in Part Fourteen. Part Thirteen’s naming of the 1894 Act was answered in Part Sixteen. Nothing in this series is now outstanding except what Part Nineteen consolidates and what the second series has been committed to in section 10.5.
Across nineteen parts this book has made nine forward promises in numbered sections and kept eight of them in the part that owed it.
The ninth is the second series, and it is the only one that cannot be checked yet.
That is not offered as a virtue. It is offered because Chapter Seven of this part is a table of fifteen promises made by governments and not kept, and an author who applies that test to others and not to himself has written the wrong book.
The difference is that a broken promise in this book costs a reader sixty pages. The eleven in Chapter Seven’s table cost something else, and Part Eleven is about the people who paid it.
10.7 — Whether §3.3 held
Chapter Nine said a reader is better placed than the author to judge that. Here is what the author expects.
Supporters of this government will find this part hostile, because it contains Chapter Five and Chapter Seven and because no amount of Chapter Two offsets a unanimous Supreme Court judgment. Critics of this government will find it soft, because Chapter Two exists at all, because Chapter Four declines to say the removal of Article 370 was wrong, and because section 10.1 leaves out the things they consider most serious.
Both of those reactions are the intended result and neither is evidence that the part is balanced. A document can annoy everybody and still be wrong.
The only test that means anything is the one in Chapter Eight: whether the ruler transferred. It did, on all four tests, and the largest finding it produced — that three of the four tests were never measuring the party at all — damages this series far more than it damages the government this part is about.
An author who set out to write a prosecution and produced, in the eighteenth part, the strongest available argument that his own framing was wrong, has either been honest or has been outmanoeuvred by his own method. This book cannot tell which, and Part Nineteen is not going to resolve it either.
Five contested questions are left unadjudicated and named, so the omissions are visible: Article 370’s wisdom, citizenship legislation, institutional independence, the farm laws, demonetisation.
The objection — that refusing to grade is not neutrality when applied to a government whose files have not surfaced — this book concedes and cannot answer.
And for the first time in eighteen parts there is no new missing document. The missing documents are the same five, in the same buildings, held now by a third government.
Sources & further reading — Part 18
The State Ledger
Part Five committed every subsequent part to carrying this table. In a comparison part it records where the material in these ten chapters landed.
| State | What appears in this part | Standing |
|---|---|---|
| Jammu and Kashmir · Ladakh | 5 August 2019; the Reorganisation Act; the Supreme Court judgment of 11 December 2023 and its direction on statehood. | Contested; upheld as lawful |
| Nagaland · Manipur · Arunachal Pradesh · Assam | AFSPA in force into March 2026. Area of application reduced; Section 6 unamended; sanction refused at Mon, December 2021. | Both columns. Part Thirteen |
| Rajasthan | Banswara, 21 April 2024. Chapter Six. | Statement established; no finding made |
| Jharkhand · Chhattisgarh · Odisha · Madhya Pradesh · West Bengal | The coal belt again — auctions made the default from 2015, the one rule of the fifteen with the largest financial consequence. | Credit. Fourth appearance of this map |
| Every state | GST from 1 July 2017; the insolvency code; delivery programmes; the 2019 amendment to the transparency law. | Both columns |
The fourth row is the mineral belt for the fourth time. Part Five found it losing its only advantage to freight equalisation from 1952; Part Thirteen at the overlap of the three maps; Part Sixteen as the location of the cancelled coal blocks; and here as the place the auction rule finally applies. It is the only entry in this ledger that has moved in the same direction across four parts.
Timeline
| Date | Event |
|---|---|
| May 2014 | BJP wins 282 seats, NDA 336. Congress reduced to 44. Part Sixteen. |
| Dec 2014 – 2015 | Ordinance to amend the 2013 land acquisition Act, re-promulgated twice, abandoned. |
| Feb – Mar 2015 | Coal blocks re-auctioned under the 2015 Act. Auctions become the default. |
| 2016 | Insolvency and Bankruptcy Code. The identity authority given a statute after seven years. |
| Nov 2016 | Demonetisation. Upheld 4:1 by the Supreme Court in January 2023 on the process. |
| 1 Jul 2017 | Goods and Services Tax in force. |
| Aug 2017 | Supreme Court sets aside instant triple talaq. |
| 2 Jan 2018 | The Electoral Bond Scheme notified. |
| Mar 2019 | The first Lokpal appointed — five years and three months after the Act. Part Sixteen. |
| 2019 | Triple talaq legislation. RTI Act amended: Information Commissioners’ tenure and conditions moved to central rules. |
| 5 Aug 2019 | Constitutional Orders 272 and 273. Article 370 ceases to operate; J&K reorganised. |
| 2020 – Nov 2021 | Farm laws enacted, protested for a year, repealed. |
| Dec 2021 | Mon, Nagaland: fourteen civilians killed. State cabinet recommends prosecution; sanction refused. Part Thirteen. |
| 11 Dec 2023 | Supreme Court upholds the abrogation of Article 370, unanimously. |
| 2023 | 106th Amendment reserving a third of legislative seats for women. Not yet in force. |
| 15 Feb 2024 | Electoral Bond Scheme struck down unanimously by five judges. About ₹16,518 crore had been encashed. |
| 4 Mar 2024 | Sita Soren: seven judges end legislators’ immunity for bribery. Parts Fourteen and Fifteen. |
| 21 Apr 2024 | Banswara. Election Commission notices to both national parties on 25 April. |
| 4 Jun 2024 | BJP 240, NDA 293, Congress 99. A coalition government for the first time since 2014. |
| Sep 2024 | Review petition against the electoral bonds judgment dismissed. |
| Mar 2026 | AFSPA still in force in parts of four north-eastern states. Part Thirteen. |
Words Used in This Part
| AFSPA | Armed Forces (Special Powers) Act. Section 6 bars prosecution of personnel without central sanction. Part Thirteen. |
| Article 356 | President’s Rule — direct rule of a state from Delhi. Parts Three and Sixteen. |
| Article 367 | The interpretation clause of the Constitution: it tells you how to read the rest of it. Amended on 5 August 2019. |
| Article 370 | The provision giving Jammu and Kashmir a special constitutional position. Headed temporary. Parts Two and Seven. |
| Constitutional Order | An order issued by the President under a constitutional provision, having the force of constitutional law without a parliamentary amendment. |
| Electoral bond | A bearer instrument for donating to a political party anonymously. Struck down 15 February 2024. |
| GST | Goods and Services Tax, in force 1 July 2017, replacing state-by-state taxation of goods in transit. |
| Insolvency and Bankruptcy Code | The 2016 statute creating a single time-bound process for resolving failed businesses. |
| Like-for-like | Comparing by holding the method constant. Chapter One defines the four disciplines this part used. |
| Model Code of Conduct | The Election Commission’s rules of campaign conduct. Not a statute; binding by convention and enforced by the Commission. |
| Ordinance | A law made by the executive when Parliament is not sitting. Lapses unless Parliament passes it. Part Seventeen. |
| Presumptive loss | The gap between what a government received and what an auditor calculates it could have received. Part Sixteen, Chapter Three. |
| Prosecution sanction | Permission required before certain public servants can be prosecuted. Parts Thirteen, Fifteen, Sixteen. |
| Star campaigner | A leader designated by a party to campaign nationally. Their expenses fall on the party rather than the candidate, and the party answers for their conduct. |
A Word on Sources
Everything dated in this part was checked against a source rather than recalled, and the standard was raised rather than lowered because the subject is a government in office.
The load-bearing material is judicial. In Re: Article 370, 11 December 2023, five judges, unanimous. The electoral bonds judgment of 15 February 2024, five judges, unanimous, and the dismissal of the review in September 2024. The demonetisation judgment of January 2023, four to one. All are reported and all are public.
The statutory material — the Goods and Services Tax, the Insolvency and Bankruptcy Code, the 2016 identity statute, the 2019 amendment to the transparency law, the 2019 triple talaq legislation, the 106th Amendment, the Constitutional Orders of 5 August 2019 — is on the statute book and can be read by anybody.
Chapter Six rests on a rally recorded and circulated by a news agency on the day, and on Election Commission letters dated 25 April 2024. The complaints that produced them are public documents filed by named parties and organisations.
Chapter Seven’s two tables are the weakest thing here and the most important, because they assert absences. An absence is harder to source than an event. Each row was checked against the statute book and against the record of bills introduced; a reader who finds one wrong should say so, and the table is wrong.
Five contested matters are named in Chapter Ten and left ungraded. That is a choice with a direction to it, and Chapter Ten’s argument box states the objection rather than burying it.
A note from the author
I expected to find this part the hardest to write honestly and I was right, but not for the reason I expected.
I assumed the difficulty would be resisting the pull of a second prosecution — that having spent seventeen parts building a case against one party, I would find it easy and satisfying to turn the same apparatus on another. That pull was there and it was manageable, mostly because §3.3 of my own brief, written before Part One, told me in advance what it would cost.
The real difficulty was Chapter Eight, and I did not see it coming.
I built four tests over seventeen parts without ever asking whether they were tests of a party or tests of a state. Applying them here answered that, and the answer is that three of the four were never measuring what I thought they were measuring. The instruments outlive everybody. The rules are made by nobody. The documents are released by no one. Those findings do not become weaker because a different party is in office — they become the only findings in this series that have been shown to be findings at all.
Which means the most damaging chapter in Part Eighteen is aimed at Parts One to Seventeen. I have written it that way and I am not comfortable with it, and I have decided that discomfort is not a reason to soften a conclusion the method produced.
On the specific places where the writing could have pulled, three declarations. Chapter Two is longer than I wanted it to be and I lengthened it twice after noticing I had written it thin. Chapter Four declines to say whether the removal of Article 370 was right, and I hold views about it which are not in this book and will not be. And Chapter Ten’s argument box states an objection to my own method that I cannot answer — that a neutral rule applied to a government whose files have not surfaced produces a systematically kinder result. That objection is correct. I have no answer. I have put it on the page in bold rather than in a footnote.
Part Ten declared that I am Punjabi and have written on that material before. Nothing here touches that directly. What I will declare instead is that I am a critic of the Congress party writing about the government that replaced it, that everybody reading this knows that, and that the only protection against it is a method visible enough to be checked. Chapter One exists for that reason and it is the first chapter of the part.
Lovepreet Singh
Part Eighteen of The Congress Record
The Living Archive · misterlove.in
Part Nineteen is The Honest List: proven, alleged, folklore and unknown, with the state ledger and the full evidence-grade table consolidated, and the strongest case the defence can still make. The second series auditing the government of 2014 onwards is committed to in Chapter Ten of this part, in terms specific enough to be checked.