The Indian government is working exactly as it was designed to.

Start with the uncomfortable premise, because everything else follows from it: the Indian state is not malfunctioning. It is doing, with great reliability, exactly what its structure is built to do. Corruption, apathy and negligence are not bugs in the machine. They are its outputs.

This sounds like cynicism. It is actually the opposite. The cynical view is that India is governed badly because Indians in office are bad people - greedy politicians, lazy babus, callous police. That theory has a fatal problem: we have been replacing the people for seventy-eight years. Every party has held power somewhere. Entire generations of officials have retired and been replaced by idealistic young ones. The behavior survives every transplant. And in systems dynamics there is a name for that observation - it is the founding insight of the whole field: when behavior persists across a complete change of actors, the behavior belongs to the structure, not the actors.

The purpose of a system is what it does. Not what it says, not what it was meant for - what it does.

Stafford Beer’s POSIWID principle, the sharpest diagnostic tool in systems thinking

So this piece asks a different question. Not “why are they like this?” but “what structure would produce exactly this behavior from ordinary people?” - and then checks whether India has that structure. (Spoiler: it does, in almost textbook form.)

You need only five ideas to follow the whole argument. A stock is anything that accumulates - money, trust, pending court cases, the shared sense of what is “normal.” A flow fills or drains a stock. A reinforcing loop (marked R) is a circle of cause and effect that feeds itself - each turn amplifies the next, which is how you get compounding and “it keeps getting worse.” A balancing loop (marked B) pushes back toward a goal, like a thermostat - it is how systems self-correct. And a delay is the time between a cause and its consequence: a balancing loop with a long enough delay behaves as if it doesn’t exist.

One reading rule for everything that follows: every loop below is populated by people behaving rationally inside their local incentives - the voter selling a vote, the constable taking a hundred rupees, the officer who stops signing files, the family that flees to a private school. Nobody needs to be a villain for the machine to produce villainy at scale. That is the whole point - and, as we’ll see at the end, the only real source of hope.

Part 1 · The inherited operating system

Institutions are stocks. They accumulate slowly - procedures, precedents, reflexes, the folk knowledge of “how things are done here” - and they drain even more slowly. Which is why the single most important fact about the Indian state is its birth certificate.

The administrative machine India runs today was not designed to serve Indians. It was built, mostly between 1858 and 1935, to do three things: extract revenue, maintain order, and keep a suspicious eye on the governed. The Police Act of 1861 - drafted in the panicked aftermath of 1857 - built a force whose explicit job was protecting the regime from the population. The district collector was, as the name still says, a collector. The “steel frame” of the services was engineered for obedience up the chain and authority down it, with the citizen appearing in the design mainly as a subject to be managed.

In 1947, India executed one of history’s great handovers - and kept the machine. This was not stupidity; a newborn state facing Partition needed a working apparatus immediately, and this was the apparatus that existed. The Constitution wrote breathtaking new goals on top: liberty, equality, dignity, welfare. But in systems terms a goal is just one component - and a weak lever if the structure underneath stays intact. The daily procedures, the incentive gradients, the postures of authority carried over almost untouched. Many states still police themselves under descendants of the 1861 Act; the Supreme Court’s 2006 directives on police reform (Prakash Singh) remain substantially unimplemented. Article 311 still makes a civil servant nearly impossible to remove - a protection designed to insulate a colonial service, now insulating everyone, including the indifferent.

So hold this image: a machine engineered for extraction and control, handed new owners and a new mission statement, but never re-engineered. Everything that follows is what happens when you run modern electoral democracy through that machine.

We changed the drivers. We repainted the bus. We never once looked at the engine.

Part 2 · The engine - six reinforcing loops

Six reinforcing loops do most of the work. Each is a flywheel: push it once and it pushes itself harder on the next turn. They interlock - several share variables - which is why the system feels less like a set of problems and more like a single organism defending itself.

R1 · The Election ROI Loop

Winning elections in India costs staggering, still-rising amounts: the 2024 general election was estimated at around ₹1.35 lakh crore - the most expensive election ever held anywhere on Earth. Legal spending limits are a fraction of what serious campaigns cost, which means the system has quietly made a structural decision: a competitive candidate must raise unaccounted money. Not may - must. The honest candidate isn’t merely at a moral disadvantage; they are at a hopeless resource disadvantage before the game begins.

Unaccounted money does not come from supporters. It comes from investors, and investors expect returns - contracts, land-use changes, clearances, transfers, regulatory mercy. Office becomes a revenue instrument, which raises the value of holding office, which raises what everyone will spend to win it, which raises the cost of the next election. The loop closes, and every turn tightens it. It also selects who enters: as costs climb, parties prefer candidates who bring their own money and muscle. Milan Vaishnav’s When Crime Pays documents the result - self-financing candidates with serious criminal cases are not tolerated but recruited. In 2024, a record 46% of elected MPs had criminal cases, and a candidate with cases had roughly three times the chance of winning. The loop is not failing to filter criminality; it is filtering for it.

R2 · The Posting Market

R1 lives at the top. R2 is its transmission system, and it runs on the one lever politicians hold over every bureaucrat: the transfer. Because postings differ enormously in rent potential, control over postings becomes control over livelihoods. Where transfers are discretionary and frequent, posts acquire prices: officers pay or lobby to obtain lucrative seats, must then recoup, recoup through extraction, and a share flows upward to the politicians who control the posting - conveniently funding R1.

The loop’s most corrosive product is not the money. It is the filter. An officer who refuses to play is not fired (Article 311) - they are rotated to irrelevance. Ashok Khemka, who kept flagging irregular land deals, retired in 2025 after 57 transfers in 33 years - one every seven months, across a career. Every young officer watches and updates accordingly. The system doesn’t need to corrupt anyone directly; it just needs everyone to see what integrity costs.

R3 · The Great Exit

Here is the quietest loop, and possibly the deepest. Ask why public services can decay for decades without triggering an explosion from the people most capable of making noise. The answer is that they left - not the country, the systems. Over forty years, everyone who could afford it executed a private secession, service by service: private schools, private hospitals, borewells and generators and inverters, private security, air purifiers. Albert Hirschman named the choice in 1970: when a system fails you, you can use voice (complain, organize, demand) or exit (quietly buy your way out). Every exit removes exactly the person whose voice carried the most weight.

The cruel elegance is that exit looks like a solution - the private school genuinely educates the child. Locally rational, systemically fatal. By now the loop has produced an elite that experiences the Indian state almost exclusively through two interfaces - the airport and the tax notice - and whose politics floats free of service quality altogether. The state stopped feeling pain from its failures because the people with the sharpest nerve endings unplugged.

R4 · Learned Helplessness

“Why are citizens so apathetic?” is the wrong question. Apathy is usually framed as a cause of bad governance. In this model it is an output - the correct conclusion of millions of natural experiments run by ordinary people against their own state. File an FIR: the station may refuse. File a civil case: India’s courts hold 5.29 crore pending matters, and a dispute can outlive the person who filed it. File an RTI: it genuinely works - which is why civil-society trackers have counted around a hundred RTI users murdered since 2005 for asking the wrong question. Each channel teaches the same lesson: effort and outcome are barely coupled, and effort is sometimes fatal.

Update on that evidence honestly and you get the observed behavior: disengagement, or the bribe - the one lever that visibly, reliably works. And the loop closes with a click: as belief drains, effort drains; as effort drains, the system faces less pressure to respond; as responsiveness falls, the belief becomes more accurate. The tragedy isn’t that citizens are wrong about the system. It’s that they’re right, and being right makes it righter.

R5 · The Normalization Ratchet

Kaushik Basu reframed corruption as not many individual moral failures but an equilibrium - a stable, self-enforcing pattern of mutual expectations. Where everyone expects everyone to be corrupt, the corrupt move is safe and the honest move is expensive; with the opposite expectations, the same people produce the opposite behavior. Two equilibria, same people. The variable isn’t character; it’s what everyone believes everyone else will do.

That belief is a stock, filling for decades: every act of visible impunity - the acquitted strongman, the unpunished scam - adds to the shared conviction that “this is how it works,” which lowers the moral cost of joining in, which widens participation, which produces more visible impunity. The ratchet turns and does not turn back on its own. Each generation calibrates “normal” to a baseline the previous one would have called scandal - which is why your grandfather’s stories of one corrupt official in the taluk sound like fairy tales, and why your children may hear yours the same way.

R6 · The Red-Tape Ratchet

Every corruption scandal produces the same reflex: a new rule, a new sign-off, a new oversight body. It feels responsible. It is, in fact, the loop’s food. Distrust of officials → more rules → more compliance checkpoints - but every checkpoint is a point of discretion, and every point of discretion is a toll booth where approval can be delayed until consideration appears. More toll booths → more extraction and delay → more distrust → more rules. Liberalization dismantled the License Raj at the top for big firms and left it fully intact at the street level, where the shop license and the building permit still live.

The same ratchet has a second jaw. An officer faces asymmetric risk: sign a file that later attracts the vigilance machinery and your career can end over a good-faith decision; sign nothing and nothing happens to you, ever. So the rational officer stops deciding. What looks from outside like sloth is often meticulous self-defense. The anti-corruption apparatus, aimed at the corrupt, ended up teaching the honest to do nothing - negligence as armor.

Part 3 · The missing brakes

A healthy system pairs every engine with a brake. Democracies are supposed to carry four great balancing loops: elections, courts, watchdogs, and internal accountability. India has all four - on paper. Each has been delayed, jammed, or captured, and a balancing loop that doesn’t bite is not a weak brake; it is functionally no brake at all.

B1 · The Electoral Thermostat, jammed

The textbook loop is beautiful: govern badly → voters notice → they vote you out → your replacement corrects course. Now count the ways the signal is jammed between “governed badly” and “voted out.” Identity sorting: where votes express belonging - caste, religion, party-as-tribe - the vote stops carrying performance information. Patronage as substitute: a targeted transfer or a gas cylinder, delivered visibly, can outvote public goods delivered invisibly. Money and muscle, courtesy of R1, decide who is even on the ballot. And blame diffusion: the pothole is municipal, the road budget is state, the scheme is central, and the officials are transferred before consequences mature. Accountability that cannot find an address is not accountability.

B2 · Justice - a brake with a twenty-year delay

Deterrence is a formula: expected cost = probability of being caught × speed of consequence × severity. India’s justice system has quietly zeroed the first two terms. There are 5.29 crore cases pending; India runs on roughly 15 judges per million against a recommended 50, with a third of High Court seats vacant. A trial that matters can be stretched past two decades. Meanwhile the jails run at 131% capacity, three-quarters undertrials - a perfect inversion: punishment before trial for the poor, trial without end for the powerful.

Systems dynamics has a cold theorem: a balancing loop whose delay exceeds the decision horizon of the actors it is meant to restrain does not restrain them. A consequence that arrives in twenty years, discounted to the present, rounds to zero. This single broken loop quietly underwrites all six engines above - every one runs at the speed it does because nothing arrives in time to stop a turn of the wheel.

B3 · Watchdogs on a leash

The third brake - investigative and audit institutions - fails differently: not delay, but capture with retargeting. The CBI, the ED, the state vigilance machinery report, in practice, to the executives they are meant to police. The Supreme Court called the CBI “a caged parrot speaking in its master’s voice” in 2013 - said it about the government of the day, and it has fit every government since. A captured watchdog is worse than a dead one, because it still has teeth; they just point sideways. Cases accelerate against opponents and soften for defectors-who-complied - which converts the anti-corruption apparatus itself into a political instrument inside R1.

B4 · Internal accountability - the inverted incentive

The last brake is the state’s own HR system. A civil servant is nearly impossible to fire (Article 311), is promoted substantially by seniority, and can be transferred instantly at political displeasure (R2). Assemble those and look at the incentive field: no career upside to serving citizens well, no downside to serving them badly, and severe, immediate downside to displeasing political masters. The only live feedback loop in the building trains obedience - not service, not performance, obedience. The machine does not need cynical people; it manufactures them from idealists, at scale.

Part 4 · The thin state

One giant misdiagnosis has to be cleared away, because it inverts the common prescription. Ask what’s wrong with the Indian state and you’ll hear “bloated bureaucracy.” The data says nearly the opposite: where the state touches citizens, India is one of the most understaffed serious countries on Earth.

A third of High Court judgeships sit vacant. Devesh Kapur’s work on why the Indian state both fails and succeeds makes the point: relative to population, India’s public employment is far below the rich democracies its middle class compares it to. Lant Pritchett’s name for the creature is a flailing state - a sophisticated head attached to limbs it can barely move. Connect understaffing to the model and negligence stops being a character flaw and becomes arithmetic: an overloaded interface must ration; rationing under discretion produces queues; queues produce queue-jumping markets. “Speed money” is exactly what its name says. Maximum presence, minimum capacity - extraction-era architecture with a skeleton crew. That is the terrain on which all six engines run.

Part 5 · Why it compounds

Why does it feel like it’s getting worse? In a systems frame this is the least mysterious part. A system whose reinforcing loops are strong and whose balancing loops are broken doesn’t sit still at “bad.” It accelerates. Worsening isn’t an anomaly the model must explain away; it is the model’s central prediction. Five accelerants deserve naming.

First, the stakes grew with the economy. Since 1991 the economy expanded roughly eightfold, and the rents available at state discretion scaled with it. The accountability apparatus did not. When the prize grows tenfold and the guard does not grow at all, the ratio that governs behavior moves an order of magnitude in corruption’s favor - without a single person becoming worse.

Second, the election spiral has no brake of its own. Each cycle’s spending becomes the next cycle’s floor.

Third - the accelerant nobody had until recently - scandal metabolism collapsed below consequence time. Institutions run on decade-clocks; attention now runs on 72-hour clocks. Outrage that once accumulated into movements now discharges harmlessly as engagement. Polarization sorts every scandal into team jerseys, so scandals stopped carrying information. Each cycle of loud outrage followed by visible zero consequence runs another confirmation trial for R4: see? nothing happens.

Fourth, normalization compounds generationally. A voter born in 1995 has never observed a functioning alternative equilibrium.

Fifth, exit keeps getting easier. Every year the private substitutes improve, so R3’s exit ramp gets smoother while the public systems’ remaining constituency gets poorer and quieter.

Stable corruption would itself need explaining. Six flywheels, no working brakes: the only trajectory this structure can produce is the one you’re living in.

Part 6 · The whole map, and the stocks

Here is the machine assembled - simplified to its load-bearing core, engines in orange and broken brakes in blue. Trace any path with your finger and you end up back where you started, moving faster.

A model in this tradition is really a claim about stocks - the slow accumulations that define the system’s state. Watch these seven and you are watching the whole story; every loop above is just a pump attached to one of them.

Stock	Direction	What’s pumping it
Money required to hold power	Rising ↑	R1’s arms race; economy-scaled rents; opaque finance
Judicial backlog (5.29 crore)	Rising ↑	15 judges/million; vacancies; adjournment economics
“Everyone does it” norm	Rising ↑	R5’s ratchet; visible impunity; generational drift
Infrastructure of exit	Rising ↑	R3; ever-better private substitutes for public failure
Trust that effort works	Draining ↓	R4’s confirmations; scandal-without-consequence cycles
Voice invested in public systems	Draining ↓	Elite exit; danger to whistleblowers; futility learning
Frontline state capacity per citizen	Draining ↓	Flat hiring against rising population and complexity

Note what the table implies about time. Stocks change slowly - that is what makes them stocks. Nothing here reverses inside one election cycle, one movement, one honest officer’s career, or one viral exposé. Everyone who has promised otherwise has been defeated not by opponents but by hydraulics. That is not a counsel of despair; it is a specification of the minimum seriousness any real fix must have.

Part 7 · Where the model says to push

Donella Meadows left behind a famous hierarchy of leverage points - places to intervene in a system, ranked weak to strong. Run India’s reform history through it and a pattern jumps out: almost all reform energy has gone into the weakest levers, and almost all actual progress has come from the strong ones.

The weak levers, where the energy goes. Punishing individuals. Raids. New agencies. Awareness campaigns. These are parameter tweaks and actor swaps - the bottom of the ladder - and the model predicts their failure precisely: remove one rent-collector from an unchanged structure and it hires his replacement by Thursday. The 2011 anti-corruption movement is the clean experiment: one of the largest mobilizations in the republic’s history, aimed at actors and a new watchdog. The Lokpal it won sat unstaffed for six years and has convicted approximately no one. The loops never noticed. Anger at persons is the system’s preferred failure mode: it discharges pressure while leaving structure intact.

The strong levers, where the proof is. Wherever a structural intervention removed a discretion point or shortened a feedback delay, behavior flipped - fast, at scale, with the same officials and the same public. Railway bookings were a canonical bribe-market until IRCTC put the queue online; that corruption didn’t decline, it evaporated, because the toll booth was demolished, not policed. Passport issuance became a tracked pipeline. Direct Benefit Transfer routed welfare around the skimming hands. UPI made the everyday movement of money legible. None of these made anyone more moral. They changed the structure, and the “corrupt Indian” at that interface simply ceased to exist. Same people, different system, different behavior. (Honestly: the rents didn’t all die; some migrated up-stack to where discretion still lives - land, contracts, postings. The loop adapts. Reform is not an event; it is a pursuit.)

What the model would prescribe. Repair the justice delay (B2): filling judicial vacancies and enforcing statutory time limits re-arms deterrence for every other loop at once. Cut R1’s fuel line: real-time transparency of political money. Shield the transfer lever (R2): fixed tenures and transparent posting criteria. Push money and power down to the tier citizens can actually see - city governments, which currently raise on the order of 1% of GDP and rank among the weakest municipal systems in the world - so the pothole and the vote live at the same address. Staff the frontline. And keep demolishing discretion points the way IRCTC did: not by guarding the toll booth but by paving the road around it.

And the deepest lever - the paradigm. Every loop rests on a story: the state is a resource to be captured, and everyone knows it. Paradigms shift through demonstrated anomaly. The Delhi Metro, UPI, the passport center each quietly deposit the same subversive sentence in millions of minds - it doesn’t have to be like this - and expectation is a stock. Fill it, and it becomes pressure.

Part 8 · Honest caveats

It does not claim everything is worsening. Petty corruption at digitized interfaces has genuinely collapsed; roads, electrification and welfare delivery have improved. The model explains the persistence and compounding of governance rot specifically - and why it coexists with visible progress: growth raises the stakes faster than it builds the brakes.

It is deliberately party-agnostic - and that is a finding, not a hedge. Every loop here ran under every configuration of Indian politics. Whoever you blame, the model holds; whoever you champion, it holds. If reading it you felt it indicted your side, that feeling is R5 talking.

It compresses variation. Kerala is not Bihar. State-level divergence is real, and is itself more evidence: same civilization, different local structures, different outcomes.

And it is a map, not the territory. Six loops and four brakes cannot contain a subcontinent. The claim is only that these are the load-bearing circuits - and that any explanation which ignores them will keep producing reforms the system digests without noticing.

Coda · The one-paragraph version

India does not have a corruption problem the way a person has a bad habit. It has a corruption equilibrium the way a valley has a river - carved by structure, self-deepening, indifferent to who happens to be swimming in it. The machine was built for extraction and never re-engineered; elections became investments that office must repay; the repayment runs through a posting market that filters out the honest; the people who could object have exited, the people who can’t have learned that effort doesn’t couple to outcome, and the brakes - courts, watchdogs, the vote itself - are delayed past relevance, captured, or jammed. Six flywheels, no working brakes: of course it compounds. Nobody is driving. That is the terror of it, and also the strange mercy - because it means the rot was never made of Indian character, and everywhere the structure has actually been changed, at a railway counter or a payment app or a passport office, a hundred years of “this is just how we are” evaporated in one software update. The system is working exactly as structured. So the work - the entire work - is structure.

Sources

Data anchors. Election cost: Centre for Media Studies estimate of ~₹1.35 lakh crore for 2024, via Business Standard. Criminal cases among MPs: ADR (46% of 2024 winners) and win-rate comparison. Judicial backlog: 5.29 crore pending, via Madhyamam and the NJDG. Justice capacity (police, judges, vacancies, prison occupancy): India Justice Report 2025. Transfers: Ashok Khemka’s 57 in 33 years, via ThePrint. RTI killings: documented attacks on RTI users. Municipal finance: RBI on urban local bodies. State size: Devesh Kapur, “Why Does the Indian State Both Fail and Succeed?” (JEP, 2020).

Conceptual foundations. Donella Meadows, Thinking in Systems and “Leverage Points.” Stafford Beer’s POSIWID principle. Albert O. Hirschman, Exit, Voice, and Loyalty (1970). Milan Vaishnav, When Crime Pays (2017). Kaushik Basu on corruption as a self-fulfilling equilibrium. Akhil Gupta, Red Tape (2012). Lant Pritchett, “Is India a Flailing State?” (2009). Lakshmi Iyer & Anandi Mani, “Traveling Agents” (2012). Supreme Court of India: “caged parrot” (2013); Prakash Singh directives (2006).

Figures are the latest verifiable as of July 2026; the stocks they measure move slowly, but check current values before republishing any of them.