The Last One
I promised in Part 1 that when I gave a verdict it would sit in a box that says it is a verdict, and that you would be free to throw it away and keep the facts. That promise falls due here.
So Chapter 7 is my judgement, marked as opinion. Everything before it is evidence and everything after it is a proposal. If you want the facts without my conclusions, stop at the end of Chapter 6 — nothing in this series depends on you agreeing with me.
That is not a fudge and I want to defend it before you read it. A fudge is when you split the difference to avoid offending anyone. What follows does the opposite: it says one side is comprehensively wrong on the facts and the other is comprehensively wrong about its own machine, and then says that neither of their programmes would achieve much.
That will satisfy nobody, which is the price of having read the evidence rather than picked a team.
Three things I want on the record before the verdict.
First, I changed my mind while writing this. Several times, and Chapter 1 lists where. I began with a settled view and finished with a different one. If a fourteen-part investigation does not move the person conducting it, they were not investigating.
Second, the verdict is falsifiable. Chapter 7 states what evidence would overturn it. A conclusion that nothing could disprove is not a conclusion, and I have spent thirteen parts saying so to other people.
Third, Chapter 8 is the part I would most like you to argue with. It is a concrete design for 2047 — eight specific measures, most of which neither movement is asking for. It is easy to hold a position. It is much harder to write down what you would actually do, and much easier to be shown wrong once you have.
And one last thing. Whatever you conclude, you now know more about this than almost anyone shouting about it. Please use that carefully. The people at the bottom of this system do not benefit from another confident voice.
— Lovepreet Singh
July 2026
1Where I Was Wrong
Six things I believed at the start of this series that the evidence changed. If none of these had happened, the series would not have been worth writing.
Before the verdict, an accounting. These are positions I held in July 2026 and no longer hold, in the order I lost them.
1. I thought varna was straightforwardly a birth hierarchy from the start
Part 2 changed that. The word shudra appears exactly once in the entire Rig Veda. The four-varna hymn is in the latest layer, says nothing about heredity, and the genetic evidence shows widespread population mixing continuing for centuries afterwards.
The ranking was there from the beginning. The closure came later, and the people who say so are not simply making excuses.
2. I thought "the British created caste" was a nationalist talking point
Part 5 changed that too. They did not create the hierarchy — the eyewitnesses in Part 4 settle it. But they created the modern administrative object: the lists, the schedules, the census categories, the certificate in your file. That half of the claim is correct and well documented, and dismissing it wholesale was lazy.
3. I thought reservation was obviously working
Part 11's Group A table stopped that. Every reserved category below quota at senior level, seventy-six years in. It works — the causal evidence is real — but "obviously working" was not a defensible position and I had held it without checking.
4. I thought the anti-reservation case was mostly bad faith
Part 12 made that untenable. The strongest version of it — no defined exit condition, incoherent categories, judicially established elite capture, a state that keeps caste on file while asking people to forget it — is a serious argument that I could not answer as easily as I expected.
The public version is often bad faith. The argument itself is not.
5. I thought the two sides disagreed about a great deal
They do not. Under cross-examination in Part 12 both accept that the injury is real, the categories are crude, elite capture is genuine, the ceiling was breached incoherently, and the top is unrepresented. The entire dispute reduces to rebuild or remove.
6. And I thought this was the important argument
This is the one that changed most. Part 12, Chapter 8 listed eight things neither case addresses — marriage, land, capital, the private sector, support after admission, Dalit Muslims and Christians, denotified tribes, and measurement.
Every one has a larger claim on attention than exam cut-offs. None can be settled by winning this argument. I did not expect to finish a fourteen-part series on reservation concluding that reservation is not the main event, and I have not enjoyed it.
That the historical injury was real, systematic and enormous. Nothing in thirteen parts of evidence moved that even slightly, and the material in Part 6 made it considerably harder to doubt than when I started.
2The Demands, Assessed
The Reservation Hatao Andolan published four demands and one broader proposal. Each is taken seriously and answered on the evidence.
Selection by examination performance alone, with no caste category.
The evidence: India already runs a caste-blind quota — EWS since 2019. Its threshold was copied from the OBC creamy layer with no study behind it, covers the overwhelming majority of eligible households, and the government publishes no data on who has received it. Removing caste does not remove categories; it relocates them.
Separately, entrance scores measure performance after purchased preparation, so "merit alone" selects partly for family resources. And where selection is already caste-blind — the private sector — identical CVs produce ten interview calls, six and three.
VERDICT: Rejected. Not because the principle is wrong, but because India has tested it and the result is a differently arbitrary system with less measurement.One qualifying and admission standard for all candidates.
The evidence: for all-India government MBBS seats in 2025, General closed at 660 and OBC at 658 — two marks apart, and level by the final stray round. The gap this demand targets is largely not where it is believed to be.
The SC gap of roughly 85 marks is real and is a legitimate subject of argument. It is a different argument from the one being made.
VERDICT: Largely misdirected. The premise is wrong for the largest category and right for one.An end to category-based fee concessions and upper-age relaxations.
The evidence: these are administrative relaxations, distinct from seat reservation, and they are the most reasonable things the movement has asked for. There is a real case that age relaxation compounds advantage across repeated attempts, and a real case against on the ground that candidates from families without resources start later and interrupt more.
Fee concessions are means-testable and could be moved to income criteria without touching the quota at all.
VERDICT: Legitimate and specific. These deserve a serious policy discussion, and arguing about them separately would be far more productive than the argument being had.The state should cease requiring or recognising caste in public life.
The evidence: caste is sustained by marriage, by neighbours, by land and by employers. Four religions removed the doctrine — Buddhism, Islam, Christianity and Sikhism, one of them born in Punjab specifically to break it — and caste is inside all four today. There is no basis for expecting a form column to succeed where that failed.
What would end is measurement. The 97% manual-scavenging figure, the Group A shortfall, the sub-classification exercises in Telangana and Karnataka all exist because somebody counted by caste.
VERDICT: Rejected, and it is the weakest thing they propose. It would delete the record and leave the practice — and the movement's own best arguments depend on data the proposal would abolish.The movement's strongest argument is one it rarely leads with: that the categories are incoherent. Arbitrary at the threshold, unstable in membership, uncapped in total, and unevenly distributed inside.
Everything in Parts 10 and 11 supports that. The ₹8 lakh line with no study behind it. Jats added and removed within a year. 59.5% centrally and 62% in Maharashtra past a "constitutional" ceiling, while Bihar's 65% was struck down. 3,743 OBC castes sharing one undivided quota.
They are right about the machine. They are wrong about what to do with it.
3What Is True In The Case Against
Stated plainly, with nothing withheld. Every item here is established somewhere in Parts 1 to 13.
- There is no exit condition, and there never has been. Article 16(4) is triggered by inadequate representation. In seventy-six years no government, court or commission has defined what adequate representation would be. A condition nobody has specified cannot be satisfied.
- Elite capture is real and judicially established. Warned about in 1963, built into the creamy layer in 1992, extended in 2018, and confirmed by a seven-judge bench in 2024 when it permitted the Scheduled Caste category to be split. It is not a talking point; it is sixty years of judicial finding.
- And it is universal, not an Indian failure. Malaysia after fifty-five years, South Africa after twenty-eight, Nepal within fifteen. Three continents, one pattern.
- The categories are incoherent. A threshold copied without study. A community added to the OBC list and removed by a court within a year. A ceiling that binds Bihar and not Parliament. Nearly four thousand castes in one undivided quota.
- The state keeps caste on file. Registered at birth, declared at school, certified at admission, carried into employment. Whatever else this achieves, it is not the annihilation of caste that the movement's own founding figure demanded.
- Ambedkar's limit has been crossed. He told the Assembly on 30 November 1948 that reservation must be confined to a minority of seats. Central reservation is 59.5%.
- The prize is shrinking. Government employment is a small and falling share of Indian jobs, and the sharpest measured discrimination is in the private sector, which reservation has never entered.
- And the poor of the general category exist. Parliament acknowledged it in 2019. They are used as a rhetorical prop by a movement whose base is urban and educated, and they are still real.
Anyone who tells you the case against reservation is entirely bad faith has not read it. Most of what is said in public is bad faith. The case underneath it is not, and treating it as such is how the retention side loses arguments it should win.
4What Is True In The Case For
Same treatment, same standard.
- It works, and there is causal evidence. Caste groups newly added to the Scheduled Caste list gained seven to eight rank points of upward mobility over the following twenty years — a natural experiment, peer-reviewed, published in 2024. Against a national background of flat and low mobility, the groups that received the policy moved.
- The injury is current. Twenty-seven per cent of Indian households admit practising untouchability — a perpetrator self-report, so a floor. Fifty-two per cent among Brahmin households. Around five per cent of marriages are inter-caste.
- The labour market is not caste-blind. Identical CVs produced ten interview calls for an upper-caste name, six for a Dalit name and three for a Muslim name — in the private sector, where reservation has never applied.
- Degrading work is still allocated by birth. Around ninety-seven per cent of identified manual scavengers are Scheduled Caste. Over sixty-six per cent of central government sanitation staff come from reserved categories. People die in sewers every few days.
- The constitutional condition remains unmet. Group A of the central government: Scheduled Castes 14.2%, Scheduled Tribes 6.54%, Other Backward Classes 19.14% — all below quota. The trigger for the exception still obtains.
- Removing belief has been tried four times and failed. Buddhism, Islam, Christianity, Sikhism. Caste is inside all of them in India today.
- Removal has fast, visible effects. After the 2023 United States ruling, Black enrolment at Harvard fell from 18% to 11.5% within two admissions cycles, at Princeton from 9% to 5%.
- And reserved places sit empty. More than 2,750 reserved teaching posts vacant in central universities; 4,297 reserved academic positions closed as "none found suitable." Those places were taken from nobody.
Anyone who tells you reservation is a failure that survives only on politics has not looked at the mobility evidence. Anyone who tells you the discrimination is historical has not looked at 2011, 2007, 2024 or 2025.
5What Each Side Is Lying About
Including, in both cases, to itself.
The abolition side
- That reserved categories have taken over. The government publishes the opposite. Group A is below quota for all three, and OBCs are eight points short. This is the central grievance of the movement and it is false.
- That there is a large score gap. Two marks between General and OBC for all-India government MBBS seats.
- That reservation was meant to last ten years. Article 334 clocked reserved legislative seats. Articles 15(4) and 16(4) never had an expiry.
- That historical caste atrocities are invented propaganda. Said publicly by prominent voices in the movement. Part 6 is a forty-page answer, and it is not a close question.
- And the one they tell themselves: that removing state recognition would weaken caste. Four religions removed the doctrine entirely and caste survived all of them. There is no evidence for this belief and considerable evidence against it, and it is the load-bearing assumption of the entire programme.
The retention side
- That representation has been achieved. Not at Group A, not after seventy-six years, not for any of the three categories.
- That nothing has improved. Scheduled Caste mobility is measurably rising, Group C representation meets quota, and overall central government representation exceeds the SC and ST shares. Denying this erases the people who won it.
- That elite capture is a manufactured grievance. A seven-judge bench found otherwise in 2024, and two states have legislated on it since.
- That rising atrocity figures prove worsening conditions. Registered-case totals measure the reporting system.
- And the one they tell themselves: that reservation addresses caste inequality. It addresses employment and education. Scheduled Castes hold about 8.5% of land, own about 11.4% of enterprises and account for about 2.6% of billionaire wealth. Seventy-six years of the policy has not moved ownership at all, and defending it as though it were an equality programme prevents anyone from building one.
The abolition side's errors are about the present — what the composition is, what the gap is, what the law says. Those are checkable, and government data settles them. On the facts, they are simply wrong.
The retention side's errors are about its own success — how much has been achieved, how well the instrument is aimed. Those are errors of self-assessment by people defending something that works less well than they say.
One list is factually wrong. The other is defending a machine it will not examine. Those failures are not equivalent, and the response to each has to be different.
6The Whole Series In Fourteen Facts
One fact from each part — the single thing from it that, if you remembered nothing else, would leave you better informed than almost anyone arguing about this.
Before the verdict, the evidence base in fourteen lines. Everything in the next chapter rests on these.
Four of them damage the abolition case badly. Four damage the retention case badly. Three damage both. And three — the 1902 policy, the 1950 Order, and the undefined condition — are simply absent from the public argument on either side.
That distribution is not something I engineered. It is what fell out of the evidence, and it is the strongest single reason for the verdict that follows.
7The Verdict
My judgement. This is opinion, not evidence. The reasoning is shown so you can disagree with it precisely.
Part 1 set out five questions. Everything since has been an attempt to answer them. Here they are, answered.
| Question | Answer | On what basis |
|---|---|---|
| Was the injury real, and how bad? | Yes, severe | Part 6 — documents, proclamations, court records |
| Is it still happening? | Yes, measurably | 27%, 10 : 6 : 3, ~97%, ~5% |
| Does this medicine work? | Yes — partly | 7–8 rank points; but not at the top, not on assets |
| Is there something better? | Not as a replacement | EWS record; US 2023; Malaysia; South Africa |
| Who decides when it ends? | Nobody, ever | The deepest failure in the entire policy |
On the facts, the retention side wins decisively. Almost every factual claim the Reservation Hatao Andolan makes in public is false, and the government's own data disproves the central one. Reserved categories have not taken over the services; they are below quota at every senior level. The score gap for the largest category is two marks. The policy demonstrably works. The ten-year claim is a misreading. And the suggestion that historical atrocities were invented is answered by forty pages of documents, proclamations and perpetrator self-reports.
On the machine, the abolition side wins substantially. There is no exit condition and never has been. The categories are arbitrary at the threshold, unstable in membership, uncapped in total and unevenly distributed inside. Elite capture is judicially established and internationally universal. The ceiling set by the policy's own architect has been crossed. And the state that asks citizens to stop thinking in caste requires them to prove it at every threshold of their lives.
On the remedy, neither wins, and this is the important part.
Abolition would remove a working instrument from people the evidence shows are still being discriminated against, and the American data from 2023 tells us the composition effect would be immediate and large. It would also delete the measurement that makes the discrimination visible. That is not a reform; it is turning off the light and declaring the room empty.
Indefinite continuation without a defined target is not sustainable either. A measure that cannot say what success looks like, in a shrinking sector, past its own architect's limit, loses legitimacy — and the movement that appeared in July 2026 is the first instalment of the bill. The next one will be larger.
So the verdict is this. The Reservation Hatao Andolan is wrong about the facts and right about the machine. Its critics are right about the facts and wrong to defend the machine as it stands. And the question they are fighting over — abolish or keep — is the cheapest question available, because it is the only one that requires nobody to specify anything, fund anything or lose anything.
The honest answer to "who is right" is that the retention side is right about whether, and the abolition side is right about how — and that a policy nobody will redesign will eventually be destroyed by people who were wrong about it.
A conclusion that nothing could disprove is not a conclusion. Here is what would change mine.
Towards abolition: replication showing no mobility gain for newly listed groups. Group A composition reaching quota and holding. The untouchability self-report falling towards zero. Inter-caste marriage rising sharply. Evidence that the manual-scavenging composition had normalised. Any one of these would weaken the case for continuation; three together would end it.
Towards unqualified retention: evidence that the categories are in fact well calibrated — a defensible basis for the ₹8 lakh threshold, an audit showing "none found suitable" reflects a genuine pipeline shortage, and data showing benefits distributed evenly within categories rather than concentrated. That would remove most of Chapter 3.
Neither set is currently close. Census 2027 will move several of these numbers for the first time in ninety-five years, and I expect to have to revise something.
8A Design For 2047
Eight measures. Concrete enough to be argued with, and none of them is "abolish" or "keep".
India will be a hundred years independent in 2047. Here is what I think should be in place by then, in order of how much difference it would make.
Parliament should define, by statute, what "adequate representation" under Article 16(4) means: proportional representation for a category sustained across Groups A, B and C for a stated period — ten consecutive years is a defensible figure. Publish the measurement annually. A category that meets it graduates out.
Why first: it removes the single strongest argument against the policy, at no cost to anyone currently benefiting, and it converts a permanent structure back into a remedy. Nothing in the Constitution prevents it.
Not quotas. Every employer above a size threshold required to record and publish workforce composition by category, review recruitment and promotion practices periodically, and act where under-representation is found — with enforcement machinery behind it.
Why: this is Northern Ireland's mechanism, and it is the only programme in Part 13 that closed its gap and was wound down. It is also where India's sharpest measured discrimination sits — ten, six, three — and where nothing is currently counted.
Extend to Other Backward Classes what the Supreme Court permitted for Scheduled Castes in 2024 and Telangana and Karnataka have implemented. Nearly four thousand castes cannot share one undivided quota once the data exists to split it.
Why: elite capture is the most robust international finding in this entire series. It is not an Indian flaw; it is a property of these programmes everywhere, and sub-classification is the only demonstrated architecture against it.
Not caste or income. Both, layered, as Brazil did: caste weighted heavily, plus first-generation-learner status, parental education, medium of schooling, rural residence and household assets.
Why: it targets far more precisely than either criterion alone, it dissolves the false choice the entire 2026 argument is built on, and India already has the framework — Mandal's own eleven weighted indicators.
Where capacity can be increased, increase it, so that the number of unreserved places never falls. Fund the faculty and infrastructure that the 2008 and 2019 expansions did not.
Why: it is the only method that has ever defused this conflict in India, it worked twice, and the delimitation proposal before Parliament is already doing it at constitutional scale without anyone calling it reservation policy.
Bridge courses, language support, mentoring and hostel provision, funded as a standing entitlement rather than a scheme.
Why: more than 13,500 reserved-category students dropped out of central universities, the IITs and the IIMs. Both explanations of that figure imply the same remedy. It is cheap, uncontroversial, and the only intervention in this book that requires nobody to lose an argument.
Remove the religious test from Scheduled Caste status. A person of Scheduled Caste origin who converts to Islam or Christianity should be treated as Sikh and Buddhist converts already are.
Why: the current position is internally incoherent on its own stated logic, it has stood for seventy-six years, and the mobility evidence shows Muslim mobility declining over exactly the period in which Scheduled Caste mobility rose.
EWS beneficiary composition. A continuous published series of Group A composition. An audit comparing the frequency of "none found suitable" for reserved and unreserved posts at the same institutions. A working definition of administrative efficiency under Article 335.
Why: each is cheap, each would settle a live dispute, and the fact that neither movement has demanded any of them tells you something about how much either wants the argument resolved.
It is not a compromise between abolition and retention. Six of the eight measures are things neither side is asking for, and the two that touch the quota itself — defining a finish line and sub-classifying — would be resisted by the retention side.
It also does not solve caste. Nothing in it touches marriage, which Part 3 identified as the mechanism and Ambedkar named as the only solvent in 1936. I do not know what would, and I distrust anyone who says they do.
What it would do is make the policy defensible: targeted, measured, bounded, and finally aimed at the sector where the harm actually is.
9What Each Side Would Have To Give Up
One thing each. Both are painful, and neither movement has shown any sign of contemplating it.
Accept a defined, measurable finish line — and therefore accept that one day categories graduate out.
Why this is hard: because a threshold, once written, can be reached. And a movement built on the conviction that the injury is deep and continuing experiences any stated endpoint as a betrayal in advance.
Why they should: because the alternative is not permanence. It is legitimacy collapse. A remedy that cannot describe its own success stops being read as a remedy, and the July 2026 movement is what that looks like at five million followers. Defining the finish line is how you protect the policy, not how you lose it.
Accept that caste discrimination is real, current and measurable — 27%, ten-six-three, ninety-seven per cent — and argue about design instead of existence.
Why this is hard: because conceding the injury concedes that something is owed, and the emotional force of the movement comes from a sense of being wronged rather than of owing.
Why they should: because the case that concedes the diagnosis is far stronger than the case that denies it. Part 12 built it, and it is formidable. Every time a prominent voice says the atrocities were invented, they trade a winnable argument about mechanism for an unwinnable one about facts.
The retention side would get a policy that is defensible in public for the first time in decades — targeted by evidence, bounded by a stated goal, and extended into the private sector where most Indians actually work. It would also get an answer to the question that has embarrassed it since 1950.
The abolition side would get the redesign it actually wants. Every specific complaint they make — arbitrary thresholds, elite capture, no exit, incoherent categories — is addressed by Chapter 8. What they would not get is the deletion of the record, which was never going to end caste anyway.
Both would have to stop enjoying the argument, which is the real obstacle. It is a good argument. It is free, it is loud, it costs nothing, and it can be had forever.
10What You Should Actually Do
For the reader, whatever you concluded.
You have read fourteen books about a policy you probably cannot change. Here is what that is actually worth.
1. Be harder to fool
This was the promise in Part 2 and it is the main deliverable. When somebody tells you 70% is reserved, you know the figure and you know why horizontal quotas do not add. When somebody says reservation was only for ten years, you know which Article and which subject. When somebody says the atrocities were invented, you know about the purified tank at Mahad and the 27%.
That is a small thing that compounds. Most arguments are won by whoever has an actual number.
2. Correct your own side first
It is easy to fact-check people you disagree with and useless. Nobody changes their mind because a stranger corrected them.
The only corrections that land come from inside. If you are broadly against reservation, the most valuable thing you can do is tell your own people that the Group A figures exist. If you are broadly for it, tell yours that reserved posts sit empty and that elite capture is a finding of the Supreme Court.
3. Notice what is missing rather than what is loudest
When you see this argument, ask which of the eight items from Part 12 is being discussed. Usually none. Marriage, land, capital, the private sector, student support, Dalit Muslims and Christians, denotified tribes, measurement.
Anyone who raises one of those is being more useful than anyone with a hashtag.
4. Watch three numbers, not the noise
- Inter-caste marriage rates. Currently around 5%. If this rises steeply, caste dissolves regardless of policy. Nothing else predicts as well.
- Group A composition. Published annually. If it climbs steadily towards quota, the retention case strengthens; if it stays flat, the abolition critique does.
- Census 2027. The caste count in February 2027 will reset the terms of everything in this series.
5. And if you are going to act, act on something specific
Of the eight measures in Chapter 8, two require no legislation and no money: publishing data the government already holds, and funding student support. Both are winnable. Neither has a campaign.
A movement of five million people demanding an audit of "none found suitable" and a published EWS beneficiary series would get them, and would learn something either way. A movement of five million people demanding abolition will get a general election.
11The Last Word
Closing note.
In the last week of July 2026 I watched an account go from nothing to nearly six million followers in seven days, and noticed that almost nobody arguing seemed to know what they were arguing about. That was the whole reason for this.
Fourteen parts later, the thing I did not expect is how much both sides were owed.
The people demanding abolition are, in the main, not bigots. They are young, they are frightened, and they are competing for a diminishing number of secure positions in a country that has not created enough of them. Their diagnosis is wrong and their proposal would fail. Their frustration is not manufactured.
And the people defending reservation are defending something that works, that was won at enormous cost, and that answers an injury this series has established beyond reasonable dispute. They are also defending a machine they will not open, with a target they will not name, in a sector that is shrinking underneath them.
Caste is reproduced in weddings. It is held up by land, by credit, by who your neighbours will rent to and who an employer calls back. Every one of those is outside the argument that five million people are having, and every one of them is where the difference between Indians actually lives.
Reservation is a real remedy for one part of it. It is not the disease, it is not the cure, and it was never going to be either.
Three things to keep
The first is a fact. Twenty-seven per cent of Indian households told researchers that their own family practises untouchability. Not victims — the people doing it, describing themselves, about something unlawful, with nothing to gain by saying yes. Whatever else you take from this series, that number should stay with you, because it is the floor and not the ceiling.
The second is a sentence. Ambedkar, in 1936: "The real remedy for breaking caste is inter-marriage. Nothing else will serve as the solvent of caste." The man who designed the constitutional provisions said in writing that they were not the solvent. Ninety years on, neither side quotes him on it, because it points at families rather than governments.
And the third is a warning, which is the last thing this series has to say.
A remedy that cannot describe its own success will eventually be destroyed by people who were wrong about it. That is what July 2026 was — a movement built on false premises, gathering five million people in a week, because the thing it attacked had spent seventy-six years refusing to say what it was for.
The way to save reservation is not to defend it harder. It is to define it, measure it, target it properly, extend it where the harm actually is, and say out loud what would count as having finished.
Nobody on either side wants to do any of that. Which is why, on the evidence of everything in these fourteen books, I expect this argument to be had again in ten years, by different people, with the same bad numbers, and rather more heat.
I hope I am wrong. I have been wrong six times already in this series, and every one of them made it better.
— Lovepreet Singh
July 2026
Sources & further reading — Part 14
Part 14 contains no new research. Every factual claim is carried forward from Parts 1 to 13.
Where each claim was established
The 27% untouchability self-report, ~5% inter-caste marriage, the 10 : 6 : 3 correspondence study, the ~97% manual-scavenging composition, and the documented events at Mahad, Vaikom and Travancore — Part 6.
Group A composition (SC 14.2%, ST 6.54%, OBC 19.14%), Group C and sanitation composition, reserved faculty vacancies, the 4,297 "none found suitable" positions, the 7–8 rank point mobility finding, and the land, enterprise and wealth figures — Part 11.
Article 16(4)'s undefined condition, Article 335, Ambedkar's statement of 30 November 1948, Article 334 and the amendment history — Part 8. The 2025 NEET closing scores — Part 1.
The EWS threshold and its derivation, the Jat delisting, state reservation totals, the 2024 sub-classification judgment, Telangana and Karnataka, Census 2027 and the delimitation proposal — Part 10. The 1991 economic quota and the 2006–08 seat expansion — Part 9.
Caste inside four religions and the Constitution (Scheduled Castes) Order 1950 — Part 4. Endogamy as the mechanism and Ambedkar on intermarriage — Parts 3 and 7. The post-2023 United States enrolment data, Malaysia, Brazil, South Africa, Nepal and Northern Ireland — Part 13. The two steelmanned cases, the cross-examinations and the seven alternative positions — Part 12.
A note on the verdict — and on corrections
Chapter 6 is opinion and is marked as such. It states the evidence that would overturn it, because a conclusion that nothing could disprove is not a conclusion. Chapter 7 is a proposal, not a finding; each measure is drawn from something demonstrated somewhere in the series, but the package as a whole has never been tried and I make no claim that it would work as intended.
Chapter 1 lists six positions I abandoned during this series. I expect to abandon more. Census 2027 reports caste data in February 2027 and will be the first genuine measurement of several quantities this series has had to estimate from 1931 arithmetic; I anticipate having to revise at least one substantial claim when it lands.
If you find an error anywhere in these fourteen books, I want to know. A series whose central argument is that both sides are running on bad numbers has no business running on bad numbers itself.