The Part I Was Most Afraid To Write
Young people died in 1990. Not metaphorically, not in the aggregate sense in which people die of policy — individually, by name, in public, and on camera.
They were overwhelmingly students. They were overwhelmingly from communities that had held a disproportionate share of India's public employment for a very long time. And they believed, correctly or not, that their futures had just been taken from them by an announcement in Parliament.
Writing about that requires a decision, and I want to state mine before Chapter 6.
The anti-Mandal protests are often treated as a shameful episode by privileged people defending privilege. Some of it was. And a nineteen-year-old who takes his own life in public is not a political abstraction, whatever you think of the cause.
Equally, there is a version of this story in which the deaths are the whole point and the reason for the policy vanishes. That is not honest either. Nobody died for the four decades in which the Kalelkar report sat unread, and nobody protested about that.
So Chapter 6 states what happened and does not editorialise over it. I have deliberately not described how anybody died. The scale is the fact that matters, and it is bad enough.
Three other things surprised me. The first commission's own chairman disowned his own report — Kaka Kalelkar wrote to the President in 1955 repudiating the caste basis of the recommendations he had just signed, a fact almost never mentioned.
The 27% figure is not a measurement. It is what was left after subtracting the existing quotas from a court-imposed ceiling. Mandal's own estimate of the OBC population was 52%. Chapter 2 explains the gap, and it is not what either side thinks.
And India already tried an economic quota in 1991. The Supreme Court struck it down in 1992. EWS in 2019 was the second attempt, and it needed a constitutional amendment to survive.
— Lovepreet Singh
July 2026
1The Commission That Disowned Itself
India's first attempt to identify its backward classes ended with its own chairman writing to the President to repudiate the report he had just signed.
Article 340, described in Part 8, empowers the President to appoint a commission to investigate the conditions of socially and educationally backward classes. It was used almost immediately.
Appointed on 29 January 1953 and chaired by Kaka Kalelkar, a Gandhian, writer and social worker. It submitted its report on 30 March 1955.
It identified 2,399 castes and communities as socially and educationally backward, of which 837 were classified as most backward, and recommended measures including reservation in government service and educational institutions.
And then the chairman changed his mind
Having presided over a two-year exercise that identified backwardness by caste, he wrote that he had come to doubt the whole approach — that treating caste as the unit of backwardness would entrench caste rather than dissolve it, and that the remedy risked perpetuating the disease.
Several members of the Commission also recorded dissent.
The Government of India declined to act on it. The Home Ministry's position was that an all-India list of backward classes on caste lines was not acceptable, and the matter was left to the states.
The abolition side treats this as the case closed: the first man asked to do the job concluded it could not be done honestly. That is a real point and it deserves more attention than it gets.
But look at what he was objecting to. Kalelkar's doubt was not that backwardness was imaginary — his report documents it at length. It was that caste as the administrative unit would entrench caste. He was making the argument the Reservation Hatao Andolan makes in 2026, in 1955, from inside the machinery.
And the retention side has a reply that is rarely given: the alternative Kalelkar wanted was to identify backwardness by economic condition instead. India has now tried that — in 1991, struck down; and again as EWS from 2019. Chapter 8 covers the first attempt, and Part 10 the second. Neither has produced the caste-free outcome he hoped for.
The honest summary: the founding objection to caste-based identification was raised by the first person asked to do it, has never been answered, and the substitutes tried since have not worked either.
2The Mandal Method
How the second commission actually worked — and the fact that explains why the figure is 27% and not 52%.
The Second Backward Classes Commission was appointed on 1 January 1979 by the Janata Party government under Morarji Desai, and chaired by B. P. Mandal, a former Chief Minister of Bihar. It submitted its report on 31 December 1980.
The eleven indicators
Mandal's method was more systematic than it is usually credited with being. The Commission built a scoring system and applied it to castes across the country.
Four social indicators — worth 3 points each (12 points). These covered whether a caste was regarded as socially backward by others, depended mainly on manual labour, had unusually early marriage, and had women working outside the home at above-average rates.
Three educational indicators — worth 2 points each (6 points). Non-attendance at school, drop-out rates, and rates of matriculation, each compared against the state average.
Four economic indicators — worth 1 point each (4 points). Family assets, dependence on loans, housing, and access to drinking water.
A caste scoring 11 or more out of 22 — half — was classified as socially and educationally backward.
Note the weighting, because it is the whole design. Social indicators carry three times the weight of economic ones. This was deliberate: the Commission's view, consistent with Article 16(4) as described in Part 8, was that the category is about social backwardness, with economic condition as evidence rather than as the test.
3,743 castes identified as socially and educationally backward.
Other Backward Classes estimated at approximately 52% of India's population.
Recommended reservation: 27% in central government services and public sector undertakings.
Now the question almost nobody asks
If the Commission estimated the OBC population at 52%, why did it recommend 27%?
The Supreme Court in Balaji (1963) had indicated that reservation should stay under 50%. Scheduled Castes already held 15% and Scheduled Tribes 7.5% — a total of 22.5%.
50 minus 22.5 leaves 27.5.
The Commission recommended 27% because that was the space remaining below the ceiling. It said so.
For the retention side: the figure most fought over in modern India is not a claim that OBCs deserve 27% of anything. On the Commission's own numbers they are 52% of the population receiving 27% of reserved posts. Attacking 27% as excessive means defending a ratio the Commission itself considered a compromise forced by a court.
For the abolition side: it means the number was never derived from evidence about how much correction was needed. It is the residue of a ceiling. Nobody ever established that 27% was the right amount to fix anything — it was simply what fitted.
A policy that has reorganised Indian politics for thirty-six years rests on a figure produced by subtraction.
3The Numbers Problem
The most serious criticism of Mandal is a methodological one, it is correct, and there is a defence that its critics almost never confront.
The Commission's central estimate — that Other Backward Classes are around 52% of India — has been carrying enormous political weight for four decades. It should be examined properly.
The three objections, and they are all valid
India has not conducted a full caste enumeration since 1931 — a point established in Part 5. So the Commission had no contemporary count of how many people belonged to the castes it was identifying.
It worked from the 1931 figures, with reference also to 1891, arriving at 52% by subtraction: total population, minus Scheduled Castes and Tribes, minus non-Hindu communities, minus forward Hindu castes.
That is arithmetic on data roughly fifty years old at the time of the report, and roughly ninety-five years old today.
The 1931 census was of undivided India. Its population included the territories that became Pakistan and Bangladesh, and Burma had been administratively separated only in 1937.
Using it to estimate proportions in the India of 1980 requires assumptions about partition, differential migration and differential population growth over half a century.
India has more than seven lakh villages. The Commission's own survey sample was in the region of 0.1% of them.
From that base it classified 3,743 castes across an enormous and heterogeneous country.
All three objections are accurate. Anyone defending Mandal has to accept them.
And now the defence, which is stronger than it sounds
1. The data gap was created by the state, not by the Commission. India stopped counting caste after 1931 as a deliberate post-independence policy. A commission appointed to identify backward castes was therefore instructed to do so by a government that had spent thirty years refusing to collect the information required. The Commission did not choose to use 1931 data; it was the only data in existence.
2. The alternative was to do nothing. Article 340 required an investigation. If the absence of a caste census makes such an investigation impossible, then the constitutional provision is inoperable by design — and the government that refused to count is the party that made it so.
3. The classification and the population estimate are separate exercises. The eleven-indicator scoring identified which castes were backward. The 1931 data was used to estimate how many people that covered. An error in the second does not automatically invalidate the first.
4. And the recommendation did not depend on the 52% figure. As Chapter 2 showed, 27% came from the 50% ceiling minus existing quotas. If the OBC share were 40% or 60%, the recommendation would have been the same, because it was constrained by law rather than by population.
The criticism is right that the numbers are weak. The defence is right that the weakness was manufactured by a state that refused to measure.
And both are about to become obsolete. Census 2027 includes full caste enumeration, with field work having begun in April 2026. For the first time since 1931, India will have an actual count.
It is worth noticing that neither side is confident about what it will show. If OBCs turn out to be well below 52%, the case for 27% weakens considerably. If they turn out to be above it, the demand to breach the ceiling becomes far harder to resist.
Part 10 covers this. It is the single most consequential dataset in the entire argument, and it will land while this series is still being written.
4Ten Years On A Shelf
The report was submitted in December 1980 to a government that had not asked for it, and nobody touched it for a decade.
The Mandal Commission was appointed by the Janata Party government in January 1979. That government collapsed. Indira Gandhi returned to power in January 1980.
The report was submitted on 31 December 1980 — to a Prime Minister whose party had not commissioned it, did not want it, and had no political interest in acting on it.
It sat. Through Indira Gandhi's return, through her assassination in 1984, through Rajiv Gandhi's enormous majority and his entire term. Nine and a half years.
It shows the policy was never inevitable. Two Congress governments with large majorities declined to implement a report that had been formally submitted to them. There was no constitutional compulsion and no political cost to inaction.
It shows what changed in 1990 was not the evidence. The report in August 1990 was the same document it had been in January 1981. What differed was that a Prime Minister needed something in a way that his predecessors had not.
And it tells you something about the shape of Indian politics before Mandal. A report identifying half the population as backward could be ignored for a decade because that half did not yet vote as a bloc. After 1990 it did, and no government has been able to ignore it since.
Which is the real answer to why reservation has proved so difficult to reverse — and Part 14 will have to build on it.
57 August 1990
The announcement, and the argument about why it was made — with both readings given properly.
On 7 August 1990, Prime Minister V. P. Singh announced in Parliament that his government would implement the Mandal Commission's recommendation: 27% reservation for Other Backward Classes in central government jobs. The office memorandum giving effect to it followed on 13 August.
The position he was in
V. P. Singh headed a National Front minority government that survived on support from two irreconcilable directions — the BJP on one side and the Left on the other. Within his own coalition he faced a direct challenge from Devi Lal, his deputy, who had significant support among north Indian agrarian communities.
V. P. Singh had a long record on the question and the report had been ignored for a decade by governments with far more power to act. Somebody had to implement it eventually, and no moment would ever have been convenient.
He knew the cost. He implemented it anyway, and it destroyed his government within three months. Whatever else that is, it is not the behaviour of a man doing the easy thing.
The timing is the argument. The announcement came days after a leadership challenge from a rival whose base was precisely the communities Mandal would benefit, in a minority government whose survival depended on holding a coalition together.
Implementing Mandal built a new constituency at a stroke and cut the ground from under the challenger. Ten years of inaction followed by action in a week of crisis is a pattern with an obvious explanation.
Politicians act on conviction when circumstances make it survivable and often not otherwise. That is not cynicism; it is how democratic politics works everywhere.
The more useful observation is what the timing tells you about the previous decade. If a beleaguered minority Prime Minister could implement Mandal in a week, then two majority governments could have implemented it at any point in the 1980s.
They did not, because there was no pressure to. The policy waited not for evidence, or for readiness, or for a court, but for a government weak enough to need it.
6The Fire
What happened in the autumn of 1990, stated plainly and without commentary.
Within weeks of the announcement, north India — and Delhi above all — was in sustained public disorder. University campuses shut. Roads and railways were blockaded. There were riots, and there were deaths in police action.
And there was a form of protest that had no real precedent on this scale in independent India.
Rajiv Goswami, a student at Deshbandhu College, Delhi University, attempted to take his own life in public as a protest against the implementation of Mandal. He survived with severe injuries. He died in 2004, aged 33.
A photograph of the incident became one of the defining images of the period, and it was followed by similar acts by students across the country.
The historian Ramachandra Guha puts the number of such attempts at around 200, with more than 60 people dying.
Other accounts give figures of approximately 159 attempts and 63 deaths. Contemporary reporting places attempted self-immolations at over 150.
The participants were overwhelmingly young, urban, students, and from communities not covered by any reservation.
I am not going to describe any of it further. The numbers are the fact that matters.
First, the fear was real and it was about a real thing. India in 1990 was a closed, slow-growing economy in which a government job was, for an educated young person without capital, close to the only secure route to a middle-class life. The private sector that exists now did not. Removing a quarter of those posts from open competition was, in that economy, an enormous change to a very small set of opportunities.
Second, that is a fact about scarcity, not about caste. The same announcement in 2026, in an economy with a vastly larger private sector, would not produce the same reaction — and the current movement, as Part 1 noted, is conducting itself in comment sections rather than on the street. The intensity of 1990 is a measure of how little else there was.
Third, and this is the hardest one: the deaths of 1990 are used, in a great deal of writing, as an argument. They are not an argument. They are a cost. A policy can be justified and still have cost sixty young lives, and pretending otherwise — in either direction — is where this subject usually goes wrong.
This chapter deals with suicide, and it is a difficult subject. If any of it touches you personally, please talk to someone you trust. I can help you find appropriate support resources if that would be useful.
7Mandal Against Kamandal
Seventy-eight days in the autumn of 1990 that set the shape of Indian politics for the next thirty-six years.
What happened next is the most consequential political sequence in modern India, and it is usually told as two separate stories. It is one story.
The framing that came out of it
Indian commentary has described the period ever since as Mandal versus Kamandal — the commission against the water pot carried by ascetics. Two mass mobilisations launched within weeks of each other, offering the same electorate two different primary identities.
| MANDAL | KAMANDAL | |
|---|---|---|
| The unit | Caste | Religion |
| The claim | You have been kept out. Take your share. | You are one people. Stop being divided. |
| Who it splits | Hindus, by caste | Indians, by religion |
| Who it unites | OBCs, SCs and STs across regions | Hindus across castes |
The two mobilisations are frequently described as though one caused the other. The honest position is narrower.
What is established: the Rath Yatra was launched seven weeks after the Mandal announcement, by a party then supporting the government that had made it, and its consequences brought that government down.
What is not established: that the Ayodhya campaign existed because of Mandal. That movement had its own long history predating 1990.
What is undeniable: that from 1990 onward, Indian national politics has been organised around the competition between a caste-based and a religion-based account of who "we" are — and that both accounts were placed before the electorate within the same eight weeks.
Every election since has been fought on some version of that contest.
And the electoral consequence
The decade after 1990 saw the rise of parties built on OBC and backward-caste consolidation — the Samajwadi Party, the Rashtriya Janata Dal, the Janata Dal (United) — and of leaders including Mulayam Singh Yadav, Lalu Prasad Yadav and Nitish Kumar, who between them dominated the politics of Uttar Pradesh and Bihar for a generation.
The political scientist Yogendra Yadav has described the period as a "second democratic upsurge": a large expansion in political participation and representation by groups previously outside power.
This is the answer to a question the abolition side rarely asks itself: why has no government reversed reservation? Chapter 4 gave half the answer — the constituency did not exist before 1990. This chapter gives the other half. After 1990 it did, and it decides elections.
8Indra Sawhney
16 November 1992. Nine judges. Six to three. And an economic quota that almost nobody remembers being struck down.
The Mandal order was immediately challenged. The Supreme Court constituted a nine-judge bench — one of the largest ever assembled on a single question — and delivered judgment on 16 November 1992.
Part 8 summarised what it held. This chapter is about the part that gets left out.
What the Court decided
| Question | Held |
|---|---|
| 27% for OBCs | Upheld, by 6 votes to 3. Caste may be the starting point for identifying a backward class, because in India caste and social backwardness substantially overlap. |
| Total reservation | Capped at 50% in normal circumstances — the ceiling traced in Part 8 to Ambedkar's own words in 1948. |
| Creamy layer | Must be excluded. Advanced members of a backward class are to be identified and taken out. |
| Promotions | Excluded. Article 16(4) does not extend to promotion. Overruled by Parliament three years later. |
| Economic-only reservation | Struck down. See below. |
The forgotten holding
In September 1991, the government of P. V. Narasimha Rao issued an office memorandum modifying the Mandal order. It added a further 10% reservation for other economically backward sections — people not covered by any existing scheme.
That is EWS, twenty-eight years early.
Indra Sawhney struck it down. The Court's reasoning was that Article 16(4) permits reservation for a backward class, and that economic criteria alone cannot constitute a class for that purpose. Poverty is not the test the Constitution set; social and educational backwardness is.
An economic quota was attempted by executive order in 1991 and held unconstitutional in 1992.
In 2019 the same idea was enacted again — this time by amending the Constitution, inserting Articles 15(6) and 16(6), which is precisely what was required to get around the 1992 holding. It was upheld in 2022 by 3 votes to 2.
What this means for the Reservation Hatao Andolan's central proposal: replacing caste-based reservation with economic criteria is not a fresh idea. It is a proposal India has now attempted twice, which required a constitutional amendment to become lawful at all, and which the Supreme Court has said cannot substitute for the constitutional test of backwardness.
Whether it works is a separate question, and Part 10 examines the record of EWS since 2019. But anybody arguing that nobody has ever tried this should know that the first attempt is thirty-five years old and is in the law reports.
9The Creamy Layer
The filter the Court ordered in 1992, how it was built, and the number that has not moved since 2017.
The Indra Sawhney judgment required that advanced sections within backward classes be excluded from the benefit. Somebody then had to decide what "advanced" meant.
Constituted to give effect to the Court's direction. It set out the criteria that still govern OBC creamy-layer exclusion, combining an income threshold with categorical exclusions.
The categorical exclusions cover children of holders of constitutional posts, of officers of specified senior ranks in the civil and armed services, and of certain professionals — regardless of income.
The income line, over three decades
| Year | Limit | Note |
|---|---|---|
| 1993 | ₹1 lakh | The original threshold |
| 2004 | ₹2.5 lakh | |
| 2008 | ₹4.5 lakh | |
| 2013 | ₹6 lakh | |
| 2017 | ₹8 lakh | Raised by order dated 13 September 2017 |
| 2026 | ₹8 lakh | Unchanged for nine years |
The line has not been raised since 2017, and that is unusual. Between 1993 and 2017 it was revised five times, roughly every five years, broadly tracking inflation. It has now been static for nine, despite repeated demands from state governments and from the National Commission for Backward Classes.
The effect of freezing a nominal income threshold is that inflation gradually pushes more families above it. A creamy layer limit that does not move is a quota that quietly shrinks.
And the number is the same as the EWS threshold. As Part 1 noted, EWS eligibility also uses ₹8 lakh. Which produces the observation that neither side likes: at exactly the same household income, a person from an OBC family is too rich to qualify for OBC reservation, and a person from a general-category family is poor enough to qualify for EWS.
Whatever the merits of either scheme, that is not a coherent way to draw a poverty line.
10The Second Round
2006 to 2008 — Mandal reached the classroom, there were protests again, and this time somebody found a compromise.
Mandal in 1990 applied to jobs. It did not apply to admission to central educational institutions. That took another eighteen years, and it produced the second great anti-reservation agitation.
The compromise nobody remembers
Rather than converting existing seats, the government substantially increased overall intake in the affected institutions — an expansion of the order of 54% — phased over several years, so that the absolute number of seats available to unreserved candidates was preserved.
Reservation was delivered by making the pie larger rather than by re-cutting it.
Compare the two rounds.
1990: a fixed number of government posts, redistributed. Zero-sum. Sixty deaths.
2008: reservation introduced alongside an expansion of capacity. Not zero-sum. Protests, litigation, and no comparable violence.
The difference in outcome is enormous, and the mechanism is not mysterious. When the reserved share rises and the absolute number of unreserved places stays constant, the arithmetic that produces panic is removed.
For the abolition side: this concedes something important. The 2006–08 round shows that the intensity of opposition tracks scarcity rather than principle. If the objection were purely to caste as a criterion, expansion would not have defused it.
For the retention side: it concedes something too. The policy was politically survivable in 2008 because it was made non-zero-sum, which is an admission that the zero-sum version imposes a real cost on identifiable people.
And for 2026: this is the single most useful precedent in this book, and Part 14 will return to it. The one demonstrated way of implementing reservation without a crisis is to expand the thing being reserved. Nobody in the current argument — on either side — is talking about expansion.
11What Mandal Actually Changed
Thirty-six years on. What moved, and the three things that did not.
What it changed
- It created a political identity. Before 1990 "OBC" was an administrative category in a shelved report. After 1990 it was a bloc that decides national elections, and every party in India now competes for it.
- It changed who governs. The leaders and parties described in Chapter 7 came to dominate the two largest states in the country. Representation in legislatures shifted substantially.
- It put OBCs into the central services, where their presence had been minimal.
- It made caste sayable in national politics. Before Mandal, open appeal to caste was regarded as backward and slightly disreputable in mainstream discourse. After it, every party does it, including those that deny doing it.
What it did not change
1. It did not reach the bottom of its own category. The 3,743 castes Mandal identified are not similar to one another. Some were already substantial landowning communities; some were not. Thirty-six years on, there is no sub-classification of OBC reservation of the kind the Supreme Court permitted for Scheduled Castes in 2024. Some states have created most-backward sub-categories; the centre has not. The benefit within the 27% is not distributed evenly and nobody knows by how much.
2. It did not produce the numbers to check itself. Thirty-six years after implementation, India still has no contemporary count of how many OBCs there are. A policy of this scale has been running on an estimate derived from a 1931 census. That is not a criticism of Mandal; it is a criticism of every government since.
3. It did not touch the private sector. Mandal was implemented in 1990 into an economy about to be liberalised. Government employment as a share of Indian jobs has been falling ever since. The policy captured a share of a prize that has been shrinking for its entire operational life.
Which is the finding that should trouble both camps. The abolition side is fighting over a diminishing quantity. The retention side won a share of a market that has been leaving.
12Nine Questions Nobody Asks
The blind spots in this stage of the argument.
1. Why 27% and not 52%?
Because 50 minus 22.5 is 27.5. The most contested number in modern India is the remainder of a subtraction from a judicial ceiling, not a finding about how much correction anyone needs.
Once you know this, arguments that 27% is excessive and arguments that it is a full remedy both look different.
2. Why did the first commission's chairman disown his own report?
Kalelkar concluded in 1955 that caste-based identification would entrench caste. That is the Reservation Hatao Andolan's argument, made from inside the machinery seventy-one years earlier, by a Gandhian appointed to do the job.
Neither side quotes him. The abolition side would have to explain why his proposed alternative has since failed twice; the retention side would have to engage with the objection.
3. Why has India refused to count caste since 1931?
Every criticism of Mandal's numbers is downstream of this. The state declined to collect the data and then commissioned an investigation that required it.
The refusal was itself a policy choice, made by governments that also declined to implement Mandal. Nobody treats it as a decision, but it was one, and it lasted ninety-five years.
4. An economic quota was struck down in 1992. Why does nobody mention it?
The Narasimha Rao government's 10% for economically backward sections was held unconstitutional in Indra Sawhney. The abolition side proposes essentially the same thing without knowing it has a legal history. The retention side does not raise it because EWS was upheld in 2022 and the precedent now cuts against them.
5. Who inside the 27% actually got it?
The Supreme Court permitted sub-classification of Scheduled Castes in 2024 on the finding that benefits had concentrated unevenly. The same question applies with at least equal force to a category of 3,743 castes — and no equivalent national exercise has been done.
Dominant OBC communities and the most deprived ones compete in the same pool. This is the most obvious unexamined problem in Indian reservation policy.
6. Why was 2008 survivable when 1990 was not?
Because seats were expanded and the absolute number of unreserved places was held constant. It is the only demonstrated method of implementing reservation without a crisis, it worked, and nobody in 2026 is proposing it.
7. Did Mandal cause the rise of Hindutva?
The sequence is documented and the causal claim is not. The Rath Yatra began seven weeks after the announcement; the Ayodhya movement had its own longer history.
What is undeniable is that both accounts of Indian identity were placed before the same electorate in the same eight weeks, and that the competition between them has organised national politics ever since.
8. Why did the constituency not exist before 1990?
The report sat for nine and a half years because the people it would benefit did not yet vote as a group. After implementation they did.
Which is the actual answer to why reservation cannot be reversed — not the Constitution, not the courts, but that the policy created its own electorate. Part 14 has to reckon with this.
9. What happens if Census 2027 says something unexpected?
Both sides have built positions on a 52% estimate derived from 1931 data, and neither knows what an actual count will show.
If the figure is much lower, the case for 27% weakens. If it is much higher, the pressure to breach the 50% ceiling becomes very hard to resist. Nobody is preparing publicly for either result, which suggests nobody is confident.
13What The Evidence Can And Cannot Settle
The honest scorecard for Part 9.
| Question | Status | Best current answer |
|---|---|---|
| Did the first commission's chairman reject its basis? | Yes | Kalelkar's covering letter, 1955 |
| Was Mandal's method arbitrary? | No | 11 weighted indicators, 22-point scale |
| Was its population estimate weak? | Yes | 1931 census, pre-partition, 810 villages |
| Where does 27% come from? | Subtraction | 50% ceiling minus 22.5% for SC and ST |
| Why was it implemented in 1990 and not 1981? | Politics | A minority government under internal challenge |
| How many died in the protests? | ~60–63 | From roughly 150–200 attempts |
| Was 27% upheld? | Yes, 6–3 | Indra Sawhney, 16 November 1992 |
| Was an economic quota tried before EWS? | Yes | 1991 order, struck down in 1992 |
| Has the creamy layer line kept pace? | No | ₹8 lakh since 2017 |
| How is the 27% distributed within OBCs? | Unknown | No national sub-classification exists |
The three things nobody can tell you
- How many OBCs there are. Until Census 2027 reports, every figure in this argument is an extrapolation from 1931.
- Whether V. P. Singh acted from conviction or calculation. Both explanations fit the evidence and they are not mutually exclusive.
- What would have happened without Mandal. The counterfactual is unrunnable, and both sides assert it confidently.
Mandal was a weakly evidenced policy, implemented for mixed motives, at a terrible human cost, which then created the political constituency that has made it irreversible — and which nobody has since bothered to measure properly.
That sentence is uncomfortable for everyone, which is usually the sign that it is close to right.
14The Five-Line Nichod
Everything in this book, compressed. If you remember nothing else, remember these.
PART 9 — MANDAL AND FIRE
- 27% is a leftover, not a measurement. Mandal estimated OBCs at 52% of India and recommended 27% — because the Balaji ceiling was 50%, SC and ST already held 22.5%, and 50 minus 22.5 is 27.5. The most fought-over number in modern India is the remainder of a subtraction.
- The first commission's chairman disowned his own report. Kaka Kalelkar wrote to the President in 1955 arguing that caste-based identification would entrench caste — the Reservation Hatao Andolan's own argument, made from inside the machinery seventy-one years earlier, and never answered by anyone.
- The report sat for nine and a half years because nobody needed it. Two Congress governments with large majorities ignored it. It was implemented on 7 August 1990 by a minority Prime Minister facing an internal challenge — and his government fell within three months.
- Around 60 people died. Roughly 150–200 attempts at self-immolation, overwhelmingly by young urban students from unreserved communities, in an economy where a government job was close to the only secure route to a middle-class life. That is a cost, not an argument — and treating it as either a scandal or a talking point is where this subject usually goes wrong.
- India already tried an economic quota — and it failed. A 10% reservation for economically backward sections was introduced by executive order in 1991 and struck down by Indra Sawhney in 1992, because poverty alone cannot constitute a "backward class" under Article 16(4). EWS in 2019 needed a constitutional amendment to do the same thing lawfully.
Coming next
EWS in 2019 and the breaking of the 50% ceiling; the two dissents in 2022 and what they said. The irony chapter: Marathas, Jats, Patidars and Kapus — dominant communities demanding inclusion in quotas rather than abolition, and what that reveals. The 2024 sub-classification verdict, the biggest structural change in decades. Census 2027 and the caste count that will settle or explode everything in Part 9. Reservation in the private sector, in the judiciary and in the media — the three places it does not exist. And where the Reservation Hatao Andolan actually fits into all of it.
A closing thought: the most useful thing in this part is the one nobody cites. In 2008 India implemented a 27% quota in its most competitive institutions by expanding total seats so that unreserved places did not fall — and it passed without a crisis. That is the only demonstrated method of doing this without people dying, and it is absent from the 2026 argument entirely. — L.S.
Sources & further reading — Part 9
Every factual claim in Part 9, and where it comes from.
The two commissions
First Backward Classes Commission (Kalelkar Commission), appointed under Article 340 on 29 January 1953, chaired by Kaka Kalelkar; report submitted 30 March 1955; 2,399 castes and communities identified as backward, 837 as most backward. Kalelkar's covering letter to the President expressing doubt about caste as the basis of classification, and the dissents recorded by other members. The Government of India declined to accept an all-India caste-based list and left the matter to the states. · Second Backward Classes Commission (Mandal Commission), appointed 1 January 1979 under the Janata Party government of Morarji Desai, chaired by B. P. Mandal; report submitted 31 December 1980; 3,743 castes identified as socially and educationally backward; OBC population estimated at approximately 52%; 27% reservation recommended.
Method and criticism
Eleven indicators: four social at 3 points each, three educational at 2 points each, four economic at 1 point each, out of 22; castes scoring 11 or more classified as backward. · Population estimate derived from the 1931 census, with reference to 1891, by subtraction; the 1931 census covered undivided India. · Field survey covered approximately 810 villages, against more than seven lakh in India — roughly 0.1%. · The 27% figure derives from the 50% limit indicated in M. R. Balaji v. State of Mysore (1963) less the 22.5% already reserved for Scheduled Castes and Scheduled Tribes.
1990
V. P. Singh's announcement of 27% OBC reservation in central government jobs, 7 August 1990; office memorandum, 13 August 1990. · Rajiv Goswami, student of Deshbandhu College, Delhi University, 19 September 1990; he survived and died in 2004. · Ramachandra Guha puts the number of self-immolation attempts at around 200 with more than 60 deaths; other accounts give approximately 159 attempts and 63 deaths; contemporary reporting places attempts at over 150. · L. K. Advani's Rath Yatra from Somnath, begun 25 September 1990; his arrest in Bihar on 23 October 1990 under the government of Lalu Prasad Yadav; withdrawal of BJP support and the fall of the V. P. Singh government on 7 November 1990.
Law and implementation
Indra Sawhney v. Union of India, judgment of 16 November 1992, nine-judge bench: 27% OBC reservation upheld 6–3; 50% ceiling; creamy layer to be excluded; Article 16(4) held not to extend to promotions; and the September 1991 office memorandum of the Narasimha Rao government providing 10% reservation for other economically backward sections struck down on the ground that economic criteria alone cannot constitute a backward class. · Ram Nandan Prasad Committee, 1993, on creamy layer criteria. · Creamy layer income limits: ₹1 lakh (1993), ₹2.5 lakh (2004), ₹4.5 lakh (2008), ₹6 lakh (2013), ₹8 lakh (order of 13 September 2017), unchanged since. · 93rd Amendment (2005) inserting Article 15(5); Central Educational Institutions (Reservation in Admission) Act, 2006; Ashoka Kumar Thakur v. Union of India (2008) upholding it for government-funded institutions with creamy layer exclusion; phased implementation with a substantial expansion of total seats — of the order of 54% — so that unreserved places were not reduced.
A note on what is missing — and on corrections
Part 9 covers the identification of Other Backward Classes and the events of 1990–2008. It makes no claim about whether reservation works (Part 11), about developments since 2019 (Part 10), or about Census 2027 beyond noting that field work began in April 2026. Casualty figures for 1990 differ between sources and all the main estimates located are given rather than a single preferred number. Descriptions of self-immolation have been deliberately limited to the fact and the scale.
If you find an error, I want to know. Corrections will be carried in later parts.