Every few years, Indian public argument returns to the same fork in the road. A scam surfaces, a file is suppressed, a contract is inflated, and two prescriptions compete for the citizen’s loyalty. The first says the answer is more light: give every person the right to demand the record, and wrongdoing will be exposed faster than any institution could expose it. The second says the answer is more consequence: build one powerful body that can investigate the powerful and put them on trial, because exposure without punishment changes nothing. The RTI vs Lokpal question is the statutory form of that argument, because India, unusually, enacted both prescriptions. The Right to Information Act, 2005 armed every citizen with a small, repeatable power. The Lokpal and Lokayuktas Act, 2013 concentrated a large power in a single institution. The two statutes ran side by side for over a decade, which means the argument no longer has to be conducted in the abstract. There is a record, and the record discriminates between the designs.

The comparison matters to different readers for different reasons, and it is worth naming them at the outset because the verdict is not identical for each. A citizen with a specific grievance, a pension file buried in a district office or a ration entitlement diverted on paper, needs to know which door to knock on first and what each door demands of the person knocking. A legislator or a legislative drafter needs the design lesson, because accountability bills continue to be written and the choice between distributing a power and concentrating it recurs in every one of them. A researcher needs the evidence stated carefully, with figures attached to their sources, because this comparison is usually conducted with unsourced numbers and inherited slogans. This article serves all three by doing one thing properly: setting the two statutes against each other across the dimensions that actually decide between them, reading the parallel timelines as evidence, and then defending a verdict with the deciding factor named. The verdict, stated in advance so the argument can be tested against it, is that the distributed model is the sounder foundation and the centralised model is a necessary but badly sequenced second storey, and that the factor which decided the contest was not the sincerity of either movement or the drafting skill of either text but institutional dependency: one Act needed nothing to be built before a citizen could use it, and the other could not move until the state it was meant to police chose to constitute it.

RTI vs Lokpal: the two models stated plainly

The first model distributes a power. Under Section 3 of the Right to Information Act, all citizens hold the right to information, and the machinery that honours the right sits inside the very bodies being watched. Every public authority designates its own Public Information Officers, receives applications through them, answers within the period the Act prescribes, and defends a refusal before Information Commissions whose decisions bind it. The citizen’s power is small in each instance: a single application buys a single record or a single answer. Its strength is repetition and ubiquity. It can be exercised from any district, against any office, on any working day, without permission, and the state cannot switch it off in one place without amending the statute in public. The full working of that machinery, its exemptions, its appeal structure and its case law, belongs to the complete guide to the Right to Information Act 2005, which owns those questions; what matters here is the shape of the design. The power is held by the crowd, and the crowd does not need to be appointed.

The second model concentrates a power. The Lokpal and Lokayuktas Act, 2013 creates a single institution at the Centre, a Chairperson and members supported by an Inquiry Wing and a Prosecution Wing, with jurisdiction over the Prime Minister in qualified terms, over ministers and members of Parliament, over central government servants of every group, and over a defined class of publicly funded bodies. Its power is large in each instance: on a complaint it can direct a preliminary inquiry, order an investigation by an agency including the Central Bureau of Investigation, and set a prosecution in motion before a special court. Its weakness is the mirror image of the first model’s strength. Everything it can do waits on the institution existing, staffed and willing, and the institution exists only when a selection committee acts and the President appoints. The statute’s composition, jurisdiction and safeguards are treated in full in the guide to the Lokpal and Lokayuktas Act 2013, which owns them; what matters here is again the shape. The power is held by a bench, and a bench can be left unconstituted.

Stated at their strongest, the two designs answer different failures. The information model answers the failure of the closed file: the citizen who suspects that a muster roll is padded or a tender rigged but cannot prove it, because proof sits in records the state keeps and will not show. The ombudsman model answers the failure of the unpunished official: the case where the facts are broadly known, reported in newspapers and admitted in audit paragraphs, and still no one with authority moves against the person responsible. A comparison that pretends one of these failures is imaginary will reach a worthless verdict, and much popular comparison does exactly that, treating the two Acts as rival brands of the same product. They are not the same product. One manufactures evidence. The other is supposed to manufacture consequences. The honest comparison therefore proceeds on two levels at once. It asks which design performed better as designed, on the dimensions where they can be set side by side, and it asks what the performance teaches about the order in which accountability machinery should be built.

Two movements, and the designs they carried into law

Both statutes were carried into law on the shoulders of mass movements, and the movements explain the designs, because each movement asked for the instrument its own experience had taught it to want. The demand for information grew out of village-level struggles over wages and rations in Rajasthan, where the Mazdoor Kisan Shakti Sangathan’s public hearings turned on a single discovery: the records of expenditure, read aloud, convicted the officials who had written them. The campaign that became the National Campaign for People’s Right to Information generalised that discovery into a statutory demand, and the state-level information laws of the late 1990s tested the mechanism before the central Act of 2005 generalised it. The passage story, the drafts and the dilutions belong to the legislative history of the RTI Act; the point for the comparison is that the movement’s method was itself distributed. Thousands of people had each used a small disclosure to settle a small account, and the statute they won preserved that shape.

The Lokpal demand had the opposite biography. The institution had been recommended by the first Administrative Reforms Commission in the 1960s and bills had failed in Parliament repeatedly across four decades, so the movement of 2011, India Against Corruption mobilised behind Anna Hazare’s fasts, was demanding something the state had promised and withheld for two generations. Its draft, the Jan Lokpal Bill, concentrated power deliberately: a strong, independent bench with its own investigative control, jurisdiction reaching the higher judiciary’s conduct machinery in some versions and the entire bureaucracy in all of them, because the movement’s diagnosis was that existing agencies were captured and only a new, separate centre of power could be trusted. The enacted Act of 2013 was a diluted descendant of that draft, as the account of the Anna Hazare movement and the Lokpal Bill records in detail. The point for the comparison is that the movement’s method was itself concentrated. It sought a single dramatic transfer of authority, and the statute it won preserved that shape too, including the shape’s vulnerability: a transfer of authority can be delayed by the authority that must perform the transfer.

There is a temptation, writing about these two movements, to grade them by sincerity or by crowd size, and the comparison should refuse it, because the evidence that matters is not which movement was nobler but what each design did once the crowds went home. The information movement left behind a statute that any individual could operate alone, the morning after commencement, without asking anyone’s leave. The ombudsman movement left behind a statute that required its adversary, the executive of the day, to perform a series of constitutive acts, constituting a search committee, convening a selection committee, issuing warrants of appointment, before the first complaint could be entertained. Movements disperse. Designs persist. The decades after enactment are governed by the design, and it is the design, not the movement, that the rest of this article compares.

Who can start, and what starting demands

The first dimension is initiation, and on initiation the two statutes are not close. Under the information model, any citizen may apply. The application is a request in writing, or through electronic means, in English or Hindi or in the official language of the area, accompanied by the prescribed fee, which the central rules fix at ten rupees, with applicants below the poverty line exempted from the fee altogether. Section 6(2) strips the transaction of everything that usually deters a complainant: the applicant is not required to give any reason for requesting the information, or any personal details other than those necessary for contacting the applicant. The citizen does not have to allege that anyone did anything wrong. Suspicion is not the price of admission; citizenship is. The request goes to an officer who already exists, designated inside the authority holding the record, and the statutory clock begins on receipt.

Under the ombudsman model, starting requires a complaint in the prescribed form, supported in the manner the rules require, alleging an offence under the Prevention of Corruption Act, 1988 against a named public servant. The complaint is screened. A bench considers whether a prima facie case exists before a preliminary inquiry is directed, and complaints that do not conform to the prescribed format have been disposed of in large numbers on that ground alone, a pattern visible in the institution’s own record of how few received complaints survived into inquiries. The complainant must also accept a risk the information applicant never faces: the Act punishes a false, frivolous or vexatious complaint, with a fine that can extend to one lakh rupees. Whatever the merits of that deterrent against abuse, its effect on initiation is structural. One statute fines nobody for asking. The other prices a failed accusation.

What must a person already know before using each route?

A person using the information route needs to know almost nothing in advance: roughly which office holds the record, and how to describe it closely enough to be found. A person using the Lokpal route must already know the identity of the public servant, the transaction complained of, and enough supporting material to survive screening.

That knowledge asymmetry deserves to be dwelt on, because it inverts the usual logic of a grievance system. Complaints machinery normally exists for people who know an outcome was wrong but cannot see inside the process that produced it. The information Act is built for exactly that person. Its premise is that the citizen’s ignorance of the file is the problem the statute exists to cure, so it demands no knowledge as the price of curing it. The Lokpal’s procedure, by contrast, is built for a person who has already assembled much of the case: the specific allegation, against the specific servant, in relation to a specific act of corruption within the limitation period of seven years that the Act prescribes. Where is such a person to obtain that material? Very often, from the records the first statute discloses. The two Acts thus stand in a relation the framers of neither fully articulated. The information Act is the evidence-gathering front end of the accountability system, and the Lokpal is a back end that presupposes evidence somebody else gathered. A citizen deciding where to begin is therefore rarely choosing between equivalents. Beginning with the ombudsman, without the record, is beginning at the stage that requires the record already to exist.

The screening stage deserves a further word, because it is where the concentration model’s gatekeeping becomes concrete. A bench of the Lokpal, considering initiation against the Prime Minister, must be the full bench of the Chairperson and all members, and at least two thirds of the members must approve, with the proceedings held in camera and the records sealed if the complaint is dismissed, a safeguard package stated in Section 14 whose precise terms the Lokpal pillar sets out. Safeguards of that weight are defensible, indeed necessary, for jurisdiction over a head of government. But their structural meaning should not be missed. At the very top of the system, the design multiplies the number of people who must agree before anything begins, which is the opposite of the information model’s design, where a single officer must answer and the burden of justification sits on refusal. One model defaults to motion and makes obstruction justify itself. The other defaults to rest and makes the accusation justify itself. Everything else in this comparison flows from those two defaults.

Where the burden sits

The second dimension is the burden, and it is the dimension lawyers notice first because it decides cases before they are argued. The Right to Information Act places the burden, at the point where it matters most, on the state. An applicant need not justify the request at all. If the Public Information Officer refuses, the officer must give reasons, must inform the applicant of the appeal route, and, when the matter reaches the Commission, Section 19(5) places the onus of proving that the denial was justified on the officer who denied it. Section 20 then attaches a personal penalty to default: a penalty of two hundred and fifty rupees for each day of delay or obstruction, up to a total of twenty-five thousand rupees, imposed on the officer personally after a hearing, with the burden of proving reasonable and diligent conduct again on the officer. The design choice is deliberate and unusual. In most legal processes the person seeking something must persuade the forum. Under this Act, at the appellate stage, the person withholding something must persuade the forum. The citizen’s silence is not a weakness in the case; the officer’s silence is.

The Lokpal Act allocates the burden the other way, as criminal process generally does. The complainant must state the allegation, support it in the prescribed form, and survive a prima facie screening, and the institution’s wings must then build, through preliminary inquiry and investigation, a case capable of sustaining a prosecution before a special court, where the ordinary criminal standard will apply. None of this is a drafting defect in itself. A body that can set a prosecution in motion against a minister or a secretary cannot proceed on demand in the way a disclosure regime can, because the consequences it triggers are graver and irreversible in a way a released file is not. The comparison is nevertheless instructive, because the burden allocation explains the volume difference between the two systems better than any claim about public enthusiasm. A process that presumes in favour of the applicant will be used millions of times. A process that presumes nothing and prices error will be used by the organised, the advised and the desperate. Published estimates cited in the transparency literature, including the report cards of the Satark Nagrik Sangathan, put applications under the information Act at between four and six million a year. The Lokpal’s registered complaints, by contrast, were stated in a Lok Sabha answer to number in the low hundreds a year across the years 2021-22 to 2024-25, with 149, 320, 166 and 292 complaints registered in those four years respectively. The figures are not measuring the same act, and this article does not pretend otherwise. They are measuring what two burden allocations do to the willingness of ordinary people to begin.

Who carries the burden of proof under the RTI Act and the Lokpal Act?

Under the RTI Act, the burden rests on the official: the applicant need give no reasons for asking, and Section 19(5) requires the officer who refused to prove the refusal was justified, with a personal penalty for default. Under the Lokpal Act, the burden rests on the complainant, who must frame and support an allegation that survives screening.

The burden dimension also governs what each system produces when it fails. When an information request fails, it fails visibly and appealably: a refusal order, a deemed refusal by silence, a first appeal, a second appeal, each stage generating paper that can itself be requested and examined. The failure leaves a record, and the record can become the story. When a Lokpal complaint fails at screening, it fails quietly, and where the complaint concerns the Prime Minister it fails under seal by design. Opacity at the point of failure is sometimes justified, and the article does not argue that screening should be public. It observes only that a system whose failures are auditable will accumulate public knowledge about itself, including knowledge of its own obstruction, while a system whose failures are sealed asks to be taken on trust, which is the one currency a new institution has not yet earned.

What each route delivers at the end

The third dimension is the remedy, and here the advantage reverses, which is precisely why the comparison is worth conducting honestly. The information Act delivers a record. Its appellate machinery can order disclosure, can direct a public authority to appoint officers, publish information, change its record-keeping practices and compensate an applicant for loss suffered, and its decisions are binding under Section 19(7). It can penalise the defaulting officer personally. What it cannot do is prosecute anybody, remove anybody, or recover a rupee of misappropriated money. Its most powerful product is a fact, established on paper, which the authority must then live with. The Supreme Court’s decision in Chief Information Commissioner v. State of Manipur drew the boundary exactly: a Commission entertaining a complaint cannot itself direct disclosure, because the complaint and appeal routes serve different purposes, a distinction that underlines how carefully the Act separates its own functions. Even at full stretch, the information model’s end product is knowledge with an official stamp.

The Lokpal’s end product is a prosecution. Where a preliminary inquiry and investigation establish a prima facie case, the institution can direct that a public servant be proceeded against, file the charge sheet through its Prosecution Wing before a special court, and pursue the trial of offences under the Prevention of Corruption Act, the same offences the Prevention of Corruption pillar in this series examines. It exercises superintendence over the investigating agency in cases it refers, including the Central Bureau of Investigation, a power no disclosure regime possesses. It can also recommend transfer or suspension of a public servant connected with an allegation while the process runs. These are real powers of a different order from anything in the information Act, and the article’s verdict does not minimise them. A system that can only expose depends, for consequence, on somebody else choosing to act: a vigilance department, a police force, a government deciding whether to sanction a prosecution, a press corps deciding whether the story runs. A system that can prosecute carries its own consequence inside itself, once it exists and chooses to move.

Can either body punish a corrupt official by itself?

Neither body convicts anyone, and the difference lies in what each can start. An Information Commission can impose a penalty on the officer who defaulted, capped at twenty-five thousand rupees, and can recommend disciplinary action, but it cannot prosecute corruption. The Lokpal cannot convict, but it can order an investigation, direct a prosecution and superintend the agency conducting it.

The remedy dimension therefore cuts against any simple scoreboard. If the measure is the gravity of the outcome a single successful use can produce, the concentrated model wins outright: a conviction and a prison sentence outweigh a released file on any scale of consequences. If the measure is the probability that a use will produce its outcome at all, the calculation changes, because the prosecution route must run the full course of criminal process, with sanction questions, trial delays and appellate risk, while the disclosure route produces its outcome, the record, at the end of a thirty-day statutory clock and two appeal stages. The information Act also contains a provision that quietly widens its remedy beyond the file in the individual case: Section 8(2) empowers disclosure in the public interest notwithstanding the Official Secrets Act, 1923, and Section 22 gives the Act overriding effect over inconsistent laws, so the record it releases is released against the grain of a legal culture built on official secrecy. The Lokpal has no equivalent lever against secrecy as such. It can summon records for its own proceedings, but its proceedings against the highest offices are confidential by design, and the citizen who set them in motion may never learn what the file contained. One model publishes. The other sequesters. For a polity whose corruption problem is substantially a problem of hidden transactions, which of those is the remedy is not a rhetorical question, but it is the question on which the two designs genuinely divide, and honest defenders of the ombudsman answer it without flinching: sequestered investigation is what makes prosecution possible, and publication is what makes prosecution unnecessary often enough that the state can afford to ignore it.

How high and how wide each Act reaches

The fourth dimension is coverage, and coverage has two axes that are usually confused: how high the reach extends, toward the top of the state, and how wide it spreads, across the mass of daily administration. On width, the information Act is without rival in the Indian statute book. It reaches every public authority, a term the courts have read to cover bodies owned, controlled or substantially financed by government, with the office of the Chief Justice of India held to fall within it by a Constitution Bench in the Subhash Chandra Agarwal case, and it reaches them all in the same way, through their own designated officers, without any institution having to be created for the purpose. The village panchayat office and the Prime Minister’s Office are, for the purpose of an application, the same kind of counter. The exemptions in Section 8 and the excluded organisations in Section 24 qualify that reach, as the pillar article sets out, but the default is universality.

The Lokpal’s reach is narrower at the base and more dramatic at the summit. At the base, its design concedes the width problem: complaints against Group C and Group D employees, the grades a citizen actually meets, are routed for preliminary inquiry through the Central Vigilance Commission under the machinery the Act inserted into the Central Vigilance Commission Act, which is an admission inside the statute that the single institution cannot itself process the mass of lower-level grievance. At the summit, its reach is the boldest in Indian law: the Prime Minister is within jurisdiction, subject to the exclusions for international relations, external and internal security, public order, atomic energy and space, and to the full-bench safeguard described earlier. No information commissioner can do anything to a Prime Minister. The Lokpal, on paper, can investigate one.

Does each Act reach the highest offices in the same way?

The two Acts reach the top of the state by opposite methods. The RTI Act treats the highest offices as ordinary record-holders, whose officers must answer or justify refusal, subject to exemptions. The Lokpal Act singles the Prime Minister out, brings the office within jurisdiction, then surrounds the inquiry with exclusions and a full-bench safeguard.

The coverage comparison yields a pattern that repeats across every dimension of this article. The distributed model is strong exactly where the state’s contact with the citizen is densest, the counter, the file, the local office, and weak exactly where concentrated power protects itself, because information about a protected decision can be exempted, delayed or released only after the decision has hardened into history. The concentrated model is strong exactly at the protected summit, where its jurisdiction is genuinely historic, and weak at the base, where its own statute routes the everyday case elsewhere. Neither coverage map is a drafting accident. Each is the footprint of the movement that demanded the statute: a movement of wage workers and ration card holders built a tool for the base, and a movement focused on ministerial scandal built a tool for the summit. A citizen whose grievance lives at the base and a citizen whose grievance lives at the summit are holding different maps, and the recommendation section of this article will return to that difference, because it is where the general verdict stops applying mechanically.

The parallel timelines: what the calendar proves

The fifth dimension is time, and it is where the comparison stops being a matter of design preference and becomes a matter of record. The Right to Information Act received assent on 15 June 2005. A small group of its provisions, including the sections constituting the Information Commissions and requiring the designation of Public Information Officers, came into force at once, and the remainder of the Act commenced on 12 October 2005, the one hundred and twentieth day. Public Information Officers were to be designated within one hundred days of enactment, inside authorities that already existed, with staff who already drew salaries. The Central Information Commission was constituted with effect from that October commencement, its first Chief Information Commissioner, Wajahat Habibullah, took office on 26 October 2005, and citizens were filing applications, and receiving replies and refusals, from the commencement itself. From assent to a functioning system used by ordinary people: a single season, most of it consumed by the Act’s own deliberate hundred-day preparation window.

The Lokpal and Lokayuktas Act received assent on 1 January 2014 and came into force on 16 January 2014. It then waited. The selection committee could not be completed in the form the Act contemplated because the Sixteenth Lok Sabha had no recognised Leader of the Opposition, the largest opposition party falling short of the strength the Speaker’s directions require for that recognition. Parliament amended the Act in 2016 to admit the leader of the single largest opposition party to the committee, and the institution still waited. A search committee was constituted only in September 2018, chaired by Justice Ranjana Prakash Desai. The first Chairperson, Justice Pinaki Chandra Ghose, and the first members were appointed by warrant on 19 March 2019, the Chairperson took oath on 23 March 2019, and the members assumed charge on 27 March 2019. From commencement to a functioning institution: five years and two months, during which the Act existed, was cited, was promised, and could not entertain a single complaint, because a complaint requires a bench and there was no bench. The institution then lost its head again when Justice Ghose demitted office in May 2022, a judicial member was authorised to act as Chairperson, and the next Chairperson, Justice A. M. Khanwilkar, was sworn in only in March 2024. The vacancy pattern is not an embarrassment incidental to the design. It is the design, operating exactly as its dependency structure predicts.

Two features of this timeline deserve emphasis because they answer the standard defence of the delay. The first is that the delay was not caused by the Act’s opponents losing an argument about whether the institution should exist. Parliament had enacted it, the executive had brought it into force, and the Supreme Court, in Common Cause v. Union of India in April 2017, held the Act to be an eminently workable piece of legislation, rejected the position that its operation should wait upon further amendment, and held that appointments were not invalid merely because of a vacancy in the selection committee. Even with the Court saying the Act could work and the amendment of 2016 removing the stated obstacle, three more years passed before the search committee met and the appointments followed. A gate with several keepers opens slowly even when every keeper claims to favour opening it. The second feature is the contrast in what each statute needed from the executive it policed. The information Act needed the executive to notify designations inside its own offices, a step so administratively trivial, and so visible in its omission office by office, that non-compliance could be attacked piecemeal, authority by authority, by the very citizens the Act empowered. The Lokpal needed the executive to perform a small number of large, discretionary, easily postponed acts at the centre. Non-compliance there was total and invisible in its particulars: nothing to appeal, nothing to request, no officer in default against whom a penalty could run. The dependency did not merely delay the institution. It delayed it in a way the design itself made impossible to challenge from below.

How did the courts end up deciding whether each institution would function?

Both institutions needed the courts, but for different defects. The Commissions needed the directions in Anjali Bhardwaj, requiring vacancies to be anticipated and filled in time, because governments let benches empty. The Lokpal needed Common Cause, where the Court held it workable and refused to let a vacancy suspend it, because the institution did not exist until the executive moved.

The judicial dimension therefore qualifies the comparison without overturning it, and this article states the qualification at full strength because the brief of honest comparison requires it. The information model is sometimes praised as though it ran itself, and that is false. Its Commissions have been left understaffed and leaderless by the same executive reluctance that delayed the Lokpal, and the Satark Nagrik Sangathan’s Report Card for 2024-25 recorded some 4.13 lakh appeals and complaints pending across twenty-nine Information Commissions as on 30 June 2025, with several commissions having ceased to function for want of members. A backlog of that size is a form of denial, slower and less visible than refusal but not always less effective, and information delayed past the decision it was meant to inform is information wasted. The difference the timeline establishes is narrower and harder than the slogan version of this comparison. The information model degrades; the concentrated model can be prevented from existing. A degraded system still answers some applications every day, disciplines some officers, and produces the record by which its own degradation is measured and litigated. An unconstituted system produces nothing, measures nothing, and cannot even generate the evidence of its own failure. That asymmetry, degradation against non-existence, is the empirical heart of this article, and the verdict section will rest on it.

The two accountability models compared

The dimensions argued above are drawn together in the comparison table that follows, which is this article’s findable artifact. It is deliberately confined to the dimensions that decide between the designs, rather than the feature lists by which the two Acts are usually compared, and its final row carries the payload: the elapsed time between each statute’s commencement and the first moment a citizen could effectively use it. Every other row can be argued about, because design choices trade off against each other and reasonable readers weight them differently. The final row is a pair of dates, and dates do not have a weighting problem.

Dimension	Right to Information Act, 2005	Lokpal and Lokayuktas Act, 2013
Who can initiate	Any citizen, for any record, without giving reasons	Any person, by a complaint alleging a Prevention of Corruption Act offence against a named public servant
What initiation requires	A short application and a fee of ten rupees under the central rules, with exemption for applicants below the poverty line	A complaint in the prescribed form, screened for a prima facie case, with a fine of up to one lakh rupees for a false or frivolous complaint
What the initiator must know in advance	Which office holds the record, and little else	The identity of the public servant, the transaction, and enough material to survive screening
Where the burden lies	On the refusing officer, who must justify denial under Section 19(5) and faces a personal penalty for default	On the complainant, who must frame and support the allegation before any inquiry begins
Timeline the statute sets	A reply within thirty days, forty-eight hours where life or liberty is at stake, then two appeal stages	A preliminary inquiry measured in weeks, an investigation ordinarily within six months, then trial before a special court
What the route delivers	Disclosure of a record, binding directions, compensation for loss, a personal penalty on the defaulting officer	Investigation under the institution’s superintendence and prosecution of a public servant before a special court
Cost to the state	A distributed administrative burden carried inside existing offices, plus the Information Commissions	A standing institution with members, two wings, premises and an investigative establishment
What must exist before it works	Designated officers inside authorities that already exist	A selection committee’s recommendation and presidential appointments constituting the institution
How it can be weakened	By amending the statute or its exemptions, or by leaving the Commissions understaffed	By declining to constitute it, by leaving its head vacant, or by amending the statute
Time from commencement to first effective use	In use from commencement on 12 October 2005, the Commissions constituted the same month	In force on 16 January 2014, first Chairperson and members appointed in March 2019

Read down the columns and the two philosophies announce themselves. The information column is a list of small numbers and short distances: ten rupees, thirty days, one officer, no reasons. The ombudsman column is a list of large undertakings: a bench, two wings, a superintended investigation, a special court. Read across the rows and the trades appear. The disclosure route is easy to start and light at the finish, ending at a fact. The prosecution route is hard to start and heavy at the finish, ending at a trial. The rows also expose the asymmetry the opening promised to name. Every weakness in the information column is a weakness of degree, a fee that could rise, a backlog that can grow, an exemption that can widen. The decisive weakness in the ombudsman column is a weakness of kind: the second row of the dependency line, the requirement that the institution be constituted before it works, converts every other strength in its column into a promise conditional on the conduct of the very executive the institution exists to investigate.

What each model costs the state

The sixth dimension is cost, and it is the dimension on which the published comparison is least reliable, because figures for both systems circulate without sources and this article will not add to them. What can be compared without inventing numbers is the structure of the cost. The information model’s cost is distributed and largely hidden inside existing establishments. Every public authority absorbs the work of its Public Information Officers within its own staff, the appellate load falls on the Central and State Information Commissions, and no new empire of buildings and cadres had to be raised for the system to run. The cost per use is small and the total cost scales with use, rising as citizens file more and falling if they file fewer. Its inefficiency is real but diffuse: lakhs of official hours spent answering requests of very uneven value, and a Commission backlog whose size, as recorded in the Satark Nagrik Sangathan’s report cards, represents a queue of citizens paying in time. The bill is presented in instalments, across thousands of offices, and no single budget line announces it.

The concentrated model’s cost is the opposite shape. It is a standing charge: a Chairperson and members at salaries pegged to the higher judiciary, an Inquiry Wing and a Prosecution Wing, premises, and the investigative establishment of the agencies it directs, all maintained whether complaints number in the hundreds or the thousands, and all incurred before the first complaint can be heard. Cost of that shape is not wasteful in itself; courts are also standing charges, and their value is not measured by dividing the budget by the case count. But it changes what the institution must prove. A distributed system that is used four to six million times a year on published estimates justifies its diffuse cost almost by enumeration. A standing institution must justify its concentrated cost by output, and it is on output that the institution’s own figures, as recorded in the parliamentary process, are most damaging to it. The Lokpal’s submission to the parliamentary standing committee recorded 8,703 complaints received, of which three had been fully investigated, and that no prosecution had been launched on the basis of its inquiries at the time of that submission. Set beside the registered complaint figures in the low hundreds for the years that followed, the picture is of an expensive instrument playing very few notes. Supporters answer, fairly, that a prosecution machine’s early years are its weakest, that format rejections suppressed the intake the figures describe, and that deterrence is not visible in complaint counts. Those answers have force, and the verdict section weighs them. They do not alter the structural point. A model whose cost is concentrated must survive scrutiny of its output per rupee in a way a distributed model never has to, and an institution that begins five years late begins that scrutiny with its credibility already spent.

There is a second cost, borne not by the state but by the user, and the comparison should include it because initiation costs decide who accountability is for. The information applicant’s outlay is the fee, the drafting of a short request, and, on appeal, persistence. The Lokpal complainant’s outlay is the preparation of a formal complaint capable of surviving screening, the exposure to a fine if the complaint is judged frivolous, and in any serious case the cost of legal assistance across a process measured in years. The first outlay is within the reach of the wage worker whose movement designed the statute. The second prices the remedy toward organisations, political actors and the professionally advised, which is one reason the institution’s docket has been dominated by complaints of a political character rather than the everyday extortion at the counter that fills the information system’s queues. A model of accountability that the poorest citizen cannot afford to start has answered the question of whom it serves before it has answered any other.

How each model can be weakened

The seventh dimension is vulnerability, and the two models fail in characteristically different ways that map exactly onto their designs. The information Act, because it works through a statute that citizens operate themselves, can be weakened only in the open or by neglect. The open route is amendment, and it has been taken: the amendment of 2019, which removed the statutory term and salary parity of Information Commissioners and left both to be prescribed by the Central Government, the very executive whose records the Commissioners order released. The neglect route is the vacancy and backlog pattern already described. The amendment trail, the rules that followed, and the parallel narrowing of the personal information exemption by a single sub-section of the data protection statute, which substituted Section 8(1)(j) with a bare reference to personal information and took effect in November 2025, are the subject of the account of the RTI Amendment Act 2019 and its effect, which owns that story; the comparative point is that both routes leave evidence. An amendment is debated, divided on and reported. A vacancy can be counted, a backlog measured, and both were, by the report cards and in the Supreme Court. The system’s degradations are themselves information, available under the system.

The concentrated model can be weakened without any of that visibility, because its master switch is constitutive. A government that wishes the information Act harm must legislate or visibly starve it. A government that wishes the Lokpal harm need only do nothing, decline to convene, delay the search committee, let the Chairperson’s office sit with an acting arrangement, and plead procedure at every step. The five years between 2014 and 2019 required no vote, no amendment and no defence on the floor of either House. When appointments finally came, they came after a change in the political weather and sustained litigation, not because the statute contained any mechanism to compel its own constitution. And the design’s vulnerability does not end at constitution. An institution appointed through a committee on which the executive of the day holds the dominant voice carries its dependence into its working life, in a way a million individual applicants, who hold their power directly from the statute and owe their standing to no committee, do not.

The cluster contains a third statute that sharpens this point to an edge. The whistle blower protection law was built on the lightest dependency of the three: no new institution at all, only disclosures made to authorities that already existed, including the Central Vigilance Commission and, for corruption at the summit, the Lokpal itself. It was, in the phrase this comparison cannot improve on, designed to depend on nothing, and the Whistle Blowers Protection Act 2014 was never switched on anyway, its commencement left to a notification the executive never issued. The lesson generalises beyond any single statute and it is the darkest version of the dependency thesis. Even the smallest constitutive step, a one-line commencement notification, is a veto if the step belongs to the actor the law constrains. Design that leaves any switch in the constrained party’s hand has not removed the gatekeeper. It has merely made the gate smaller and easier to overlook.

The case against this comparison

An honest comparison must turn on itself at this point, because the argument so far flatters the information model, and the flattery conceals the strongest point on the other side. That point is simple and it is true. Information is not a consequence. An application can establish, on the record and beyond denial, that a work was paid for and never executed, that a beneficiary list was fabricated, that a file was moved for a consideration. The disclosure changes what is known. It does not change what happens to the person who did it, unless some authority, a vigilance wing, a police force, a prosecuting government, chooses to convert knowledge into process, and the daily experience of information users is that this conversion is the exception. The disclosure model’s admirers sometimes write as though exposure were itself a sanction. It is a precondition of sanction, and a precondition can be waited out. Officials who survive a disclosure learn the lesson the system teaches: the paper will move on, the penalty ceiling is twenty-five thousand rupees, and the Commission cannot come for the money or the job. This was precisely the gap the Lokpal was designed to fill, and the men and women who demanded it, who had watched disclosures accumulate without consequence for years, were not naive for demanding a body with teeth. They had correctly identified the point at which the first model stops.

The second objection is more fundamental, and this article concedes it. The two statutes are not, strictly, alternatives, and framing them as a contest is in one respect a category error. They operate at different stages of a single sequence. Detection and prosecution are not rival methods of accountability; they are consecutive operations in one method, and a polity that must choose between knowing and punishing has already organised its failure. The comparison is nevertheless instructive, for two reasons. The first is that legislatures do in fact choose, every time an accountability bill is drafted, where to put the weight: in a right exercisable without leave, or in an institution exercisable by appointment. That choice is real even if the sequence point is sound, and it deserves to be made with the evidence of these two statutes in view. The second is that the sequence itself has a direction, and the direction is discoverable from the record. The stage that depends on no one’s permission must come first and must be protected as the foundation, because the stage that depends on permission can be delayed indefinitely, and a sequence whose second stage never begins is just the first stage, working. India’s experience is that the foundation was built and the superstructure was stalled for five years and then thinly used. The sequence verdict is therefore not that the second stage is unnecessary. It is that no second stage should ever be designed so that the constrained party controls whether it begins.

The allegation that the Lokpal was designed to fail deserves separate treatment, because it is the point where this comparison most often becomes partisan. The charge was made at the moment of enactment by Arvind Kejriwal, who called the bill passed in December 2013 a Jokepal and argued that it omitted the movement’s core demands, among them an autonomous investigating agency free of government control. Anna Hazare, in whose name the movement had mobilised, accepted the enacted bill and ended his fast when it passed, so the movement itself divided on whether the statute was a betrayal or a beginning. This article treats the designed-to-fail charge as an allegation the subsequent record partly supports and partly does not. What the record supports is narrower and more useful: whatever the intentions of 2013, the statute as designed contained no defence against the simplest obstruction, which is inaction, and inaction is what it received. A legislature that had wanted the institution to exist on a date certain could have written a default into the Act, a date on which the committee’s failure would trigger an alternative appointment route. It did not. Whether that omission was design or oversight is a question about motives the evidence cannot settle, and the article leaves it there. The drafting lesson does not depend on the answer.

One more dimension: who gets to know

Running beneath the formal dimensions is one the table could not hold, because it concerns not what each route does for its user but what each route does to the public stock of knowledge. The information model is a publishing machine. Every successful application takes a record out of a file and puts it into the world, where it can be copied, cited, printed, and used by people who never filed anything. Journalists build investigations on replies obtained by strangers. Opposition parties quote disclosures in legislatures. The next applicant starts where the last disclosure ended. The system’s product accumulates, and because Section 8(2) lets the public interest override even the Official Secrets Act where disclosure is warranted, the accumulation can reach material the state classified as secret. The model has no memory of its own, no registry of what has been established, and still the knowledge compounds, because it is held by everyone.

The ombudsman model is a sequestering machine, deliberately. A complaint sets in motion inquiries whose integrity is thought to require confidentiality; proceedings against the Prime Minister are in camera by statute and their records sealed on dismissal; the institution publishes no equivalent of the disclosed file for the citizen to build on. There are sound reasons. An investigation publicised at the wrong moment warns the investigated, moves the money and burns the witnesses, and a prosecution assembled in public may collapse before it reaches a court. The design trades public knowledge for prosecutorial possibility. But the trade has a cost that compounds in the other direction. When the institution is idle, as it was for five years, its confidentiality produces a perfect blank: not merely no prosecutions, but no record of what was alleged, what was examined, or why nothing followed. The information model’s failures, by contrast, are documented by the model itself, in refusal orders, appeal backlogs and penalty statistics that any citizen can obtain. A polity choosing between the designs is also choosing between a system that reports its own obstruction and a system whose obstruction is indistinguishable from its rest.

This dimension also explains a puzzle in the politics of the two Acts. Governments of every colour have found the information Act easier to live with than its reputation suggests, and the ombudsman harder to constitute than its weakness on paper would predict. The reason is not far to seek. A million scattered disclosures embarrass a government serially and forgettably; each is a day’s story, deniable in detail and rarely organised into a single account. A single functioning ombudsman with jurisdiction over the Prime Minister is a standing threat concentrated at one address, capable of choosing its moment. The executive’s revealed preference across two decades, tolerating the disclosure regime while amending it at the edges, and postponing the prosecution regime while praising it in principle, is itself evidence about which design the constrained party fears, and it points the same way as the timelines. Power that is distributed is hard to refuse and easy to endure. Power that is concentrated is easy to refuse, at the single point where it must be switched on, and hard to endure thereafter. The rational constrained party therefore permits the first and postpones the second, indefinitely if the constitution of the institution is left in its hands.

The same experiment, run in the states

The comparison does not run only at the Centre, because both designs were replicated at the state level, and the replication supplies a control group of sorts. The information Act created State Information Commissions on the central model, appointed through a committee of the state executive, serving state public authorities, and the pattern of their functioning tracks the central one closely: immediate usability from commencement, because state Public Information Officers were designated inside existing state offices, followed by the same degradation by vacancy and backlog that the report cards document across twenty-nine commissions. No state had to wait years for its information regime to exist, because the regime’s existence did not depend on constituting the Commission first; the Commission is the appellate roof on a building whose ground floor, the designated officer in every office, was occupied from the start.

Section 63 of the Lokpal Act required every state to establish a Lokayukta by state law within one year of the central Act’s commencement, and the state record on that obligation is the dependency thesis running in twenty-nine jurisdictions at once. States entered the period with Lokayuktas that varied wildly in strength, some long-established and genuinely feared, others vacant, toothless or abolished in practice, and the one-year statutory deadline produced no uniform consequence for default, because the Act supplied none. The Lokayukta map of India remains a patchwork that reflects state political choice rather than the central command, which is what a command without a default mechanism always produces. Where a state government wanted a functioning Lokayukta, one exists in some form. Where it did not, the deadline passed like a resolution. The parallel with the state information commissions is instructive precisely because the same state executives behaved differently toward the two designs: toward the disclosure regime they were compelled by the statute’s self-executing structure to designate officers and answer applications whether they liked the regime or not, and toward the ombudsman regime they retained the constitutive choice and exercised it according to taste. Given the same governments, the design that removed the choice produced the institution, and the design that preserved the choice produced the patchwork. Few natural experiments in Indian legislation state their result as cleanly.

The information side of the experiment also survived a constitutional rearrangement that would have been a natural moment for quiet abolition. When the Jammu and Kashmir Reorganisation Act, 2019 took effect in October 2019, the state’s own information law was repealed and its State Information Commission abolished, but the accountability did not lapse with the institution: the central Act extended to the new Union territories and the pending work of the state commission transferred to the Central Information Commission. A distributed system absorbs the loss of one node by rerouting through the others, because the right lives in every office rather than in the commission house. No equivalent rerouting exists for an unconstituted ombudsman, at either level of the federation. The network survives its nodes; the bench is its only node.

Connecting the two: a proposal worth examining

If the two models are consecutive stages of one sequence, the obvious reform question is whether the sequence can be wired together, so that the first stage’s output becomes the second stage’s input by rule rather than by chance. The concrete proposal that recurs in the reform debate is this: where an Information Commission, in a decided appeal, records a finding that disclosed material establishes a prima facie case of corruption against an identifiable public servant, that finding should be transmitted to the Lokpal as a complaint the institution is obliged to take up through its screening stage, with the Commission’s record standing in place of the complainant’s supporting material. Variants of the proposal give the transmission other triggers, an auditor’s paragraph, a vigilance finding, but the information route is the natural source, because it is the only stage of the system that produces findings continuously and in public.

The proposal deserves examination on its merits rather than applause for its symmetry. In its favour, it attacks the exact defect this article has documented at the Lokpal’s front door: the complainant’s burden of arriving with a case already assembled. A transmitted finding arrives pre-assembled, tested in an adversarial appeal, and stripped of the frivolity the fine is meant to deter, so the screening stage could be shortened to the questions that matter, identity, jurisdiction and limitation. It would also give the disclosure system the one thing it lacks, a guaranteed reader with power, and would convert the Commission’s most damning orders from press material into process. Against it, three objections carry weight. The Commissions are appellate bodies for disclosure, not investigators, and their findings on a record are findings about what the record shows, which is a thinner thing than a prima facie case of a corruption offence; treating the one as the other could flood the Lokpal with referrals it must disappoint. The Lokpal’s defenders add that obligatory intake on another body’s trigger compromises the institution’s control of its own docket, the classic ombudsman prerogative. And the states’ experience warns that wiring two institutions together multiplies the points at which an unwilling executive can interrupt the current, since the transmission duty would itself need an enforcement mechanism. The article’s judgment is that the proposal survives its objections in a narrowed form: transmission as a right of the Commission rather than a duty, with the Lokpal required to record reasons where a transmitted finding is not taken up. That preserves the docket, removes the complainant’s impossible burden in the strongest class of case, and creates the public record of refusal that the concentrated model otherwise never generates. It is, in miniature, the design principle of the whole comparison: keep every stage’s output visible, and make every refusal explain itself.

The verdict: design for no gatekeeper

The deciding factor can be named without qualification. It is institutional dependency: the degree to which a statute’s operation depends on a constitutive act by the party the statute exists to constrain. Measured on that factor, the contest between the two models is not close. The information Act’s dependency approaches zero. Its officers are designated inside existing authorities, its Commissions were constituted at commencement, its power is exercised by citizens who hold it directly from the statute, and its degradation, though real, proceeds by inches that can be counted and litigated. The Lokpal Act’s dependency approaches totality. Before March 2019 it had no output whatsoever, not because its text was unclear, its jurisdiction contested or its offences undefined, but because the persons who alone could work it had not been appointed by the process the executive controlled. Common Cause established that the law was workable; the calendar established that workability was not the binding constraint. The binding constraint was a selection committee’s convening, and a design that makes accountability wait on that event has handed the constrained party a veto in everything but name, which is this article’s namable claim and the lesson it offers every future drafter: design for no gatekeeper.

The verdict on sequence follows. Build the distributed stage first, protect it as infrastructure, and design any centralised stage so that its beginning does not lie in the gift of those it will investigate: commencement by the statute’s own force on a date the Act itself fixes, appointments that take effect by default if the committee defaults, interim arrangements that vest automatically rather than by authorisation, and published reasons required for every refusal at every gate. None of these devices is exotic. Each is a recognisable drafting technique, and the absence of all of them from the Lokpal Act, read beside their equivalents in the information Act’s self-executing structure, is the most instructive contrast in the two texts. India built the first stage well and the second stage badly, and the accumulated result is the picture this comparison began with: evidence in abundance, consequences in arrears.

A verdict in a comparison article must also state where it stops holding, and this one stops in three places. It stops, first, at the summit of the state. Where the wrongdoing is a ministerial or prime ministerial transaction, executed through layered intermediaries and protected by every exemption a skilled file can claim, disclosure will ordinarily arrive too late and too redacted to matter, and only an institution with the Lokpal’s summons, superintendence and prosecution powers answers the case at all; the distributed model has no substitute to offer, and pretending otherwise would repeat the flattery this article set out to resist. It stops, second, where the grievance is not evidentiary at all: the citizen who knows exactly what happened, holds the proof already, and needs only an authority to move, gains nothing from a statute that sells knowledge and should go where the consequence lives. It stops, third, in the limiting case of a captured centre. If the selection process ever produces a Lokpal that is constituted promptly and then serves its appointers, the dependency critique will have been answered in form and defeated in substance, and the comparison will need to be run again on a record this article does not have. Verdicts in institutional design are wagers on structure, and structure is a prediction, not a proof. The prediction this record supports is the one stated: of the two designs India enacted, the one that trusted the crowd has outperformed the one that waited for the state, and the reason is written in the five years when one of them did not exist.

Which route, for whom

The citizen with a specific grievance should ordinarily begin with the information route, and the reason is mechanical rather than ideological: the grievance usually turns on a fact the file holds, the pension calculation, the muster roll, the inspection report, the sanction order, and the information Act retrieves that fact at the cost of ten rupees and thirty days, without requiring the citizen to accuse anyone of anything. The reply, or the refusal, then defines the next step. Where the disclosed record establishes a corruption offence against an identifiable servant, the material for a Lokpal complaint exists in a way it did not before, and the complaint route becomes a realistic second stage instead of a leap. Where the grievance is against a decision by the highest offices and the file is certain to be exempt or endlessly appealed, the citizen should understand that the timeline stretches and the disclosure may arrive after its usefulness, and weigh the complaint route, political remedies and the courts accordingly. This ordering is general information about how the two systems are designed to be used, not advice on any individual matter; limitation periods under the Lokpal Act and the appeal periods under the information Act are unforgiving, and a person with a live grievance should take advice on the facts of their own case before a period expires.

The advocate’s answer differs by brief. For a client whose case needs a document, the information route is a discovery instrument that Indian civil procedure does not otherwise supply at this price, and it should be run early and in parallel with litigation rather than after it. For a client whose case needs a prosecution, the complaint route is the only statutory door, and the advocate’s work is the front-loaded assembly the design demands, allegation, servant, transaction, material, within the seven-year limitation. The legislator and the staffer are the readers for whom the verdict is written most directly: when the next accountability bill is drafted, the questions to ask of it are the dependency questions. What must be created before this works, who must create it, what happens if they decline, and does any switch remain in the hand of the constrained? A bill that cannot answer those questions in its text will answer them in its history, as the Lokpal Act did. The researcher and the journalist should treat the two systems as each other’s sources: the disclosure regime for the record, the complaint regime’s published figures, always attributed, for the output, and the gap between the two as the standing story of Indian accountability. The examination aspirant, finally, holds in this comparison the single most examinable design contrast in the polity syllabus, and candidates who keep their preparation as slide decks can work through the comparison in the ReportMedic PowerPoint viewer while testing themselves on the dimensions this article has used.

Independence by design: appointment, tenure and removal

A comparison of the two models would be incomplete if it examined only what each can do and not how each is protected against the power it polices, because independence is where institutional design either cashes or defaults on its promises. The two statutes protect their respective operators in strikingly different ways, and the difference again favours redundancy over concentration. The information model’s front line is not the Commission at all. It is the Public Information Officer, an ordinary official inside the authority, who needs no appointment, enjoys no tenure, and can be penalised personally for default. Above that front line sit the Commissions, whose members are appointed by the President or the Governor on the recommendation of a committee composed of the head of the executive, the leader of the opposition, and a cabinet minister nominated by the head of the executive, a composition in which the executive holds two of the three voices. Their removal is deliberately hard: proved misbehaviour or incapacity, established by a Supreme Court inquiry on a presidential reference, mirroring the protection of the higher judiciary. Their tenure and salary, however, were the point the 2019 amendment chose to loosen, replacing the statute’s own five-year term and Election Commission parity with terms the Central Government prescribes, which is why the amendment mattered out of proportion to its length: it moved the independence settlement of the appellate tier from the statute book into a rule book the beneficiary of secrecy controls.

The Lokpal’s independence architecture is more elaborate on paper, as befits a body aimed at the summit. Its selection committee seats the Prime Minister, the Speaker of the Lok Sabha, the Leader of the Opposition, the Chief Justice of India or a Supreme Court judge nominated by the Chief Justice, and an eminent jurist, and a search committee prepares the panel, so the judicial voice that is absent from the information commissions’ committee is present here. Removal follows the same judicial pattern, by the President after a Supreme Court inquiry on a reference, which may also be made on a petition signed by one hundred members of Parliament. Tenure is five years or until the age of seventy. The salaries are pegged to those of the Chief Justice of India and the judges of the Supreme Court. On every formal index, the ombudsman’s office is the better-armoured of the two. And the armour was irrelevant for five years, because armour protects an institution that exists. The comparison yields a principle that drafters consistently undervalue: independence provisions operate on a body in being, and no quantity of removal protection, salary pegging or committee balance compensates for a constitution process that can simply not be run. The information commissions, formally the weaker creation, were in being within weeks and have never since ceased to exist, even at their most understaffed, because the statute tied their existence to commencement rather than to a chain of discretionary acts. A constitution that happens automatically beats a constitution that must be willed, every time the willing is left to the party being watched.

There is a subtler point beneath the formal one, about where independence actually resides in each model. In the concentrated model, independence resides entirely in the institution: in its members’ character, its committee’s balance, its tenure rules. If any of those fails, nothing behind it remains, because the citizen holds no power of their own. In the distributed model, independence is duplicated at two levels. Even if a Commission is captured, lazy or empty, the citizen’s right against the officer survives intact, exercisable the next morning, appealable when the bench is filled, and enforceable in the last resort by the constitutional courts under writ jurisdiction. The failure of the appellate tier degrades the right without extinguishing it. The failure of the ombudsman’s constitution extinguishes the remedy altogether. Systems engineers call the first arrangement graceful degradation and the second a single point of failure, and the vocabulary transfers to statute drafting without loss. A designer who cannot guarantee the virtue of future appointing authorities, which is to say any honest designer, should prefer the architecture whose worst case is slowness over the architecture whose worst case is nothing.

What the numbers can and cannot show

Because this comparison is usually settled by rival statistics, the article pauses to state what its own figures are and are not capable of proving, since a verdict is only as good as the evidence discipline beneath it. The figures used here fall into three classes. The first class is dates: assent, commencement, appointment, oath. These are hard facts of the public record, and the timeline argument rests on them alone at its core. The second class is the statutes’ own texts: fees, periods, penalties, majorities, compositions. These are verifiable against the Acts, and the dimensional argument rests on them. The third class is volumes: applications filed, complaints received, complaints investigated, appeals pending. These are estimates and administrative counts, produced by bodies with their own methods, and this article has used them only with their sources attached, the Satark Nagrik Sangathan’s report cards for pendency, a Lok Sabha answer for registered complaints, the institution’s submission to a parliamentary standing committee for its investigated and prosecuted counts, and published estimates for application volumes. No figure of the third class carries any part of the verdict. The verdict rests on the first two classes. The third class illustrates the atmosphere in which the designs operated, and illustration is all it is asked to do.

The discipline matters because each side’s favourite statistic dissolves under it. The information model’s partisans cite millions of applications as proof of success, but an application is an input, not an outcome: it counts questions asked, including questions refused, delayed, or answered with paper that proves nothing, and the same report cards that document the system’s scale document its pendency of some 4.13 lakh cases across the commissions. The ombudsman’s critics cite three fully investigated complaints and no prosecutions from its submission as proof of futility, but a prosecution machine installed five years late, receiving complaints in a format regime that rejected large numbers at the door, will show thin early output under almost any management, and deterrence, the effect on the official who declines a bribe because the institution exists, appears in no table either side publishes. What survives these dissolutions is the structural record, and the article has deliberately built its case there: not that one side’s numbers are bigger, but that one design’s output depends on officials answering within thirty days under a personal penalty, and the other’s depended, for five years, on a committee that did not meet. Readers weighing rival claims elsewhere should apply the same sorting. Ask of every figure in this debate whether it is a date, a text, or a volume, and trust it in that order.

There is one more evidentiary source both systems generate and partisans underuse: their own paper trails about themselves. The information system produces, as a by-product of its working, a continuous public record of its refusals, its delays and its pendency, because every stage of it is a document a citizen may request, and the report cards that embarrass it are compiled from its own disclosures. The ombudsman system produces annual and parliamentary records too, but they are episodic, aggregated and published on the institution’s schedule rather than extractable on the citizen’s. The asymmetry matters for reform itself. A legislature revisiting the information Act works from a dense, current, citizen-compiled evidence base about where the machine jams. A legislature revisiting the Lokpal Act works from the institution’s own submissions and the memory of its controversies. Systems that document themselves get repaired on evidence; systems that do not get repaired on anecdote, and usually later.

Prevention and punishment: the temporal design

One further dimension organises much of the disagreement between the two camps, and naming it clarifies what the partisans are actually arguing about. The two models work at opposite ends of the corrupt act in time. The information model is built to operate before and during the act. Its premise is that a transaction conducted in the knowledge that any citizen may afterwards obtain the file, the measurement book, the beneficiary list and the noting sheet is a transaction conducted differently, and that the possibility of exposure, priced into the behaviour of the officer at the moment of decision, is the cheapest enforcement ever devised, because it requires no case to be filed and no court to sit. The Lokpal model is built to operate after the act. Its premise is that corruption is deterred by the credible prospect of investigation and punishment, and that credibility requires an institution specialised in producing that prospect against persons too senior for the ordinary machinery to touch. Prevention by transparency and deterrence by prosecution are both coherent theories. They fail in opposite ways. Transparency fails where the exposed fact meets no audience with power, the case this article has already conceded. Prosecution fails where the prospect is not credible, and a prospect is not credible when the prosecuting institution spent its first five years unborn and its next years, on its own figures, investigating almost nothing. A deterrent that does not move cannot deter, which is why the timeline is not a preliminary embarrassment to the ombudsman model but a continuing deduction from its central claim. The information model’s preventive claim, by contrast, operates at full strength from the first day of commencement, because it requires only that officers believe a file may be asked for, and that belief is created by the statute itself. The temporal asymmetry is the dependency asymmetry wearing another face, and the article notes it as corroboration rather than as a separate count in the verdict.

Three grievances, traced through both systems

Abstract dimensions settle into sense when they are run against concrete cases, and the three archetypes that dominate Indian accountability practice divide the two models cleanly. Take first the diverted entitlement: a widow whose pension is recorded as paid and never arrives, or a village whose ration grain is signed for by names that do not exist. The information route fits this grievance as a key fits a lock. The payment register, the muster roll, the stock register are records; the application costs ten rupees; the reply either produces the entry that convicts the distributor or produces a refusal that can be appealed; and the local officer who falsified the register must justify the record he himself wrote, under a personal penalty, before a Commission whose order binds his department. The ombudsman route fits it badly by design: the sums are small, the servant is junior, the complaint would in any event be routed toward the vigilance machinery for the lower grades, and the format and risk of the process are scaled for a different war. The first archetype belongs to the distributed model, and it is the archetype of the largest number of citizens.

Take second the inflated contract: a municipal work executed at twice its schedule rate, certified complete by an engineer and paid by an accounts officer, the loss spread across a city’s taxpayers and visible in the broken road itself. Here the two models run in sequence in competent hands. The information route extracts the tender file, the comparative statement, the measurement book and the completion certificate, converting suspicion into a documented chain, and it can do so while the trail is fresh because it needs no one’s theory of the case to begin. What it cannot do is follow the money into the engineer’s disproportionate assets or compel the testimony of the contractor. The complaint route can, once the documented chain exists to carry the allegation past screening: preliminary inquiry, investigation under superintendence, and the Prevention of Corruption Act’s machinery for assets beyond known income. The second archetype is the sequence argument with a file number. Neither statute alone completes it; the first produces the case, and the second is the only door through which the case can become a prosecution, which is why the wiring proposal examined earlier in this article is aimed precisely at this archetype.

Take third the ministerial allocation: a scarce public asset, spectrum, land, a mining lease, assigned by a decision at the summit, documented in notings that will be claimed as cabinet papers or deliberative record, defended by every exemption the schedule offers, and executed through officers who will each truthfully say they acted on orders. The information route strains here and the article does not pretend otherwise. Applications will be met with exemption claims, appeals will outlast the political moment, and what is finally released will document the process while the consideration, if any, sits outside the record by design, since a bribe is the one entry no file ever contains. This is the archetype the concentrated model was invented for, the one where the Lokpal’s jurisdiction over ministers and the Prime Minister, its power to direct investigation by an agency it superintends, and its in camera safeguards are not ornament but the whole point. It is also the archetype where the dependency critique bites hardest, because the allocation decisions worth investigating are those of the government whose committee appoints the investigator. The honest summary of the three traces is the verdict in miniature. The distributed model owns the base of the pyramid, where grievances are numerous and records are decisive. The concentrated model owns the apex, where its powers are unmatched and its activation is least certain. A citizenry deciding where reform effort belongs should notice that the base is where most corruption is experienced, and the apex is where most is organised, and that a system which secures the base while its apex instrument sleeps is not a failed system but an unfinished one.

Why the two designs were born in the order they were

A final strand of evidence comes from the origins of the two ideas, because India did not invent either and the sequence in which the world produced them is itself instructive. The ombudsman is the older device, descending from the Swedish institution of the early nineteenth century and recommended for India by the first Administrative Reforms Commission in 1966 on the strength of the Scandinavian and New Zealand examples. The freedom of information statute is the younger device worldwide, spreading through democracies in the final third of the twentieth century as an administrative law reform. A reader reasoning from pedigree might expect the older, more recommended institution to be the one India made work, and the record runs the other way, which suggests the difference lies not in the ideas’ quality but in their administrative demands. The ombudsman, everywhere it has succeeded, has typically been a parliamentary officer grown into a settled constitutional culture over generations, small in caseload and moral in authority. Transplanted into a statute that armed it with prosecution over the entire executive, it became the largest version of the institution ever attempted, and the most dependent on the goodwill of its targets. The information statute, by contrast, asked of the Indian administration only what the administration already did every day, keep files and answer letters, and attached a right, a clock and a penalty to the answering. Reforms that extend an existing motion survive contact with the state that must perform them. Reforms that require the state to build the instrument of its own correction require a constituency willing to defend the building for years after the movement that demanded it has gone home, and that constituency, in the Lokpal’s case, dispersed into electoral politics and private life within a few years of the fasts. The design lesson and the political lesson are the same lesson. An accountability mechanism must be able to live on the attention actually available to sustain it, and a right exercised by millions in the course of their own grievances carries its constituency inside itself, renewing it with every application filed.

The constitutional footing: a right that predates its statute, an institution that is only its statute

The deepest difference between the two models is juridical, and it explains why their vulnerabilities take the forms they do. The right to information did not begin in 2005. The Supreme Court had located the citizen’s right to know within the freedom of speech and expression in Article 19(1)(a) decades before the statute, in the line of authority that runs from Raj Narain in 1975 through S. P. Gupta in 1981, treating open government as a constitutional commitment and secrecy as the exception requiring justification. The Act of 2005 therefore codified and mechanised a right the courts had already recognised, and the codification has a consequence that is easy to overlook. If the statute were repealed tomorrow, something would remain: a constitutional claim, enforceable in the writ courts, to know what the state does in the citizen’s name, which Parliament would have to argue against the grain of settled interpretation to extinguish. The statute is the right’s machinery, not its source. That is also why the narrowing of the statute, by the amendment of 2019 or by the substitution of an exemption clause, is litigated as a constitutional question about the right itself, and why the challenge to the personal information substitution is framed in the language of Article 19 rather than as a mere quarrel with a schedule of exemptions.

The Lokpal has no such footing. It is a creature of its statute in the fullest sense: recommended by a commission in 1966, promised in bills across five decades, demanded by a movement, and brought into being, when it was brought into being, entirely by ordinary legislation that an ordinary majority could amend or repeal. No constitutional provision requires an ombudsman at the Centre. No court could hold, in 2016 or 2018, that the Constitution itself was being breached by the institution’s absence, and Common Cause, for all its firmness, rested on the statute’s own workability rather than on any constitutional command to constitute. The contrast with the constitutional watchdogs is instructive. The Election Commission and the Comptroller and Auditor General are written into the Constitution, their offices cannot be left uncreated, and their succession does not depend on a committee’s pleasure in the same way, whatever controversies attend particular appointments. The Lokpal’s designers chose the statutory route because a constitutional amendment was beyond the movement’s legislative reach, and the choice was understandable. Its price is the one this article has been documenting in other vocabularies. What a statute gives, inaction can withhold, because inaction violates nothing higher than the statute’s own expectation, and statutes do not enforce their expectations against the executive’s diary.

This footing difference also orders the two institutions’ relationships with the courts, which the comparison has touched at several points and can state as a whole. The information system lives inside judicially protected territory: its Commissions administer a right the courts treat as fundamental in derivation, its orders are routinely tested in writ jurisdiction, and its vacancies drew directions in Anjali Bhardwaj because the Court could present timely appointment as the working condition of a constitutional right. The ombudsman lives in statutory territory: its champion in court, Common Cause, had to argue from the Act’s text and purpose, and the relief available was a declaration of workability rather than a mandamus with a date. Courts can do much for a statutory institution, and did. What they cannot easily do is appoint it themselves, and the five-year vacancy is the exact shape of that limit. A designer weighing the two models for a future accountability problem should therefore ask a question prior to all the dimensions tabulated in this article. Is the protection sought a right that can be anchored in an existing constitutional freedom and distributed to everyone, or a function that can only be concentrated and must therefore be entrenched, by constitutional amendment if necessary, against the day its targets control its constitution? India answered that question one way for information and the other way for prosecution, largely by historical accident, the accident being that the courts had built the information right before Parliament legislated and had built nothing comparable for the ombudsman. The outcomes followed the footings with a fidelity that suggests the accident deserves to become a doctrine.

The human material: who staffs each model

One comparison remains, concerning the people through whom each design must act, because a statute’s real constitution is the cadre that operates it. The information model is staffed by conscription. Its first line is the serving official designated as Public Information Officer, a person with a full desk of other duties, no selection for the role, and a structural conflict built into it, since the officer is asked to disclose the records of the very office in which the officer’s career sits. The design does not pretend this conflict away; it prices it. The penalty is personal, the burden of justifying refusal is the officer’s, the appellate Commission sits outside the office, and the applicant needs no one’s character to be heroic for the mechanism to grind forward. Its appellate tier is staffed by appointment, and the criticism most often recorded against the Commissions, examined in the institution profile in this series, is that appointments have drawn heavily on retired civil servants, a pattern the Supreme Court’s appointment directions sought to open by requiring transparency in criteria and a field wider than the retired services. Even at its weakest, though, the staffing model’s failure is distributed: a timid officer corps produces slow and defensive replies, not silence across the whole system, because some officers answer well, some Commissions push, and the variance itself is visible.

The ombudsman model is staffed by selection of a very small number of the very senior: a Chairperson who has been a Chief Justice of India or a judge of the Supreme Court, members drawn half from the higher judiciary and half from persons of eminence in public life and administration, fixed in number, irreplaceable in the short run, and carrying the institution’s whole capacity in their persons. The design bets everything on the calibre of a dozen people. When the bet pays, the institution has a stature no distributed system can match, a bench whose summons a ministry must answer. When the bet is not placed at all, or is placed late, or is placed on the cautious, there is no second line, no variance to average out, no other counter at which the citizen can queue. The staffing contrast is the dependency contrast at human scale. One model assumes ordinary people, in vast numbers, with penalties for default. The other assumes extraordinary people, in tiny numbers, with no substitute if the assumption fails. Institutional designers prefer the second assumption because it is more flattering to write, and the first because it is more likely to be true, and the history of these two statutes is, among other things, a twenty-year experiment in which assumption deserved the preference.

A closing verdict

Two statutes, two movements, two theories of how a citizen holds a state to account. One theory gave the power away, in millions of identical small portions, and trusted that a fact in the open would find its consequence. The other gathered the power into one office and trusted that consequence, once armed, would find its facts. The record of the years in which both have existed supports the first theory as the foundation and refuses to dismiss the second as the completion. The information Act worked from the month it commenced, was weakened in the open, degraded by measurable inches, and produced the evidence base on which every other accountability conversation in the country still runs. The Lokpal Act waited five years for permission to exist, and its subsequent record, on its own submitted figures, is a handful of investigated complaints against thousands received. The difference was not the virtue of the movements or the intelligence of the drafts. It was that one design needed nobody’s leave and the other needed the leave of the very offices it was built to investigate. Every accountability law drafted after these two will repeat the choice, whether its drafters know they are making it or not. The evidence of this comparison says: choose the design that begins without asking.

Frequently Asked Questions

Q: What is the difference between the RTI Act and the Lokpal Act?

The two Acts give the citizen opposite kinds of power. The Right to Information Act, 2005 gives every citizen a right to obtain records from any public authority, through a short application to an officer inside that authority, for a fee of ten rupees under the central rules, with no reasons required and a reply due within thirty days. The Lokpal and Lokayuktas Act, 2013 creates a single institution that, on a formal complaint alleging a corruption offence against a named public servant, can direct a preliminary inquiry, order an investigation and set a prosecution in motion. One distributes a small power to everyone and produces information. The other concentrates a large power in one body and is meant to produce consequences.

Q: Which has achieved more against corruption, RTI or the Lokpal?

Measured by use and output, the information Act leads by a wide margin, though the measures are not symmetrical. Published estimates cited in the transparency literature put applications under the RTI Act at between four and six million a year, producing disclosures across every level of government. The Lokpal began work only in March 2019, five years after commencement, and its own submission to a parliamentary standing committee recorded 8,703 complaints received, three fully investigated, and no prosecution launched at that time. The Lokpal’s defenders answer that a prosecution institution should be judged over a longer run and by deterrence, which complaint counts do not capture. On the record that exists, the disclosure model has achieved more, because for most of the period the other model did not exist in working form.

Q: Should a citizen file an RTI application or approach the Lokpal first?

For most grievances, the information application should come first, because most grievances turn on a fact held in a file: a payment recorded and not received, a work certified and not executed, a list drawn up by criteria never published. The application retrieves that fact for ten rupees within thirty days, without accusing anyone. The reply, or the refusal, then shows whether a corruption allegation against an identifiable servant can be framed, which is what a Lokpal complaint requires in order to survive screening. Filing with the Lokpal first means assembling, unaided, the evidence the complaint will be judged on. The sequence this article recommends is record first, allegation second, with the caution that the Lokpal’s seven-year limitation period keeps running while information is gathered.

Q: Why did the RTI Act work faster than the Lokpal Act?

The difference is structural, not a matter of official enthusiasm. The information Act’s machinery consisted of officers designated inside authorities that already existed, with the Commissions constituted at commencement in October 2005, so citizens could use it from the month it began. The Lokpal Act, in force from January 2014, required a new institution to be created before it could act: a search committee, a selection committee, and presidential appointments. The absence of a recognised Leader of the Opposition stalled the committee, the amending Act of 2016 removed that obstacle, the Supreme Court held the Act workable in Common Cause in 2017, and still the first Chairperson and members were appointed only in March 2019. One Act needed designations. The other needed constitution, and constitution could be postponed.

Q: Can the same grievance go to both an Information Commission and the Lokpal?

Yes, because the two routes decide different questions and neither statute bars the other. An appeal to an Information Commission asks whether a record must be disclosed, and ends in disclosure, directions and possibly a penalty on the defaulting officer. A complaint to the Lokpal asks whether a named public servant committed a corruption offence, and can end in investigation and prosecution. A person may therefore pursue the record through the information route and the offence through the complaint route, in parallel or in sequence, and material obtained under the first can support the second. What the grievance cannot do is merge the forums: the Commission has no power to prosecute, and the Lokpal has no power to order disclosure as a remedy for the citizen.

Q: Is the RTI system or the Lokpal cheaper for the state to run?

The information system is cheaper in structure, though its cost is harder to see because it is distributed. Its work is absorbed by officers inside existing authorities, and its standing charge is limited to the Central and State Information Commissions, so the cost rises and falls with use. The Lokpal is a standing institution: a Chairperson and members at salaries pegged to the higher judiciary, an Inquiry Wing, a Prosecution Wing, premises and an investigative establishment, maintained whether complaints number in the hundreds or the thousands, and incurred before the first complaint can be heard. A standing institution can justify a standing charge through output, as courts do. The Lokpal’s difficulty is that its early output, on its own submitted figures, was very small against that charge.

Q: What can the Lokpal do that an RTI application cannot?

The Lokpal can act against the person; an information application can only act against the file. On a complaint that survives screening, the Lokpal can direct a preliminary inquiry, order an investigation by an agency including the Central Bureau of Investigation and exercise superintendence over that agency in the referred case, recommend transfer or suspension of the public servant, and have a charge sheet filed through its Prosecution Wing before a special court for offences under the Prevention of Corruption Act. Its jurisdiction reaches ministers and, with safeguards, the Prime Minister. No information application or Commission order can investigate an offence, prosecute an official, or reach assets disproportionate to known income. That is the concentrated model’s genuine and unmatched territory.

Q: Is the RTI Act or the Lokpal Act more vulnerable to being weakened?

They are vulnerable in different ways. The information Act works through a statute citizens operate themselves, so weakening it requires visible acts: an amendment, such as the 2019 change that moved Information Commissioners’ tenure and salaries from the statute into government-prescribed rules, or the substitution of the personal information exemption by the data protection statute, or the quieter neglect of vacancies and backlog. Each leaves evidence that can be counted and challenged. The Lokpal can be weakened by inaction alone, because its operation depends on appointments the executive controls: it was left unconstituted for five years without any vote or amendment, and its headship stood vacant again between 2022 and 2024. Visibility is the information model’s protection; invisibility is the ombudsman’s exposure.

Q: Can information obtained through an RTI application be used in a Lokpal complaint?

Yes, and this is the most practical connection between the two statutes. A Lokpal complaint must allege a corruption offence against a named public servant and survive a screening for a prima facie case, which means the complainant needs material, not merely suspicion. Records obtained under the information Act, sanction orders, tender files, measurement books, beneficiary lists, file notings released on appeal, are exactly the material from which such a complaint is built, and a reply certified by a public authority carries an official character that a bare allegation lacks. The statutes were not drafted as a pipeline, and no provision transmits a Commission’s findings to the Lokpal automatically. The pipeline exists in practice because citizens and lawyers carry the record from one forum to the other.

Q: Does an RTI application need to name a corrupt official?

No. An application under the information Act is a request for a record, not an accusation, and Section 6(2) provides that the applicant is not required to give any reason for the request or any personal details beyond what contact requires. The applicant need not allege that anyone acted corruptly, or even suspect it: a request for the muster roll, the stock register or the inspection report is complete in itself. This is one of the sharpest contrasts with the Lokpal route, where the complaint is defined by its allegation against a specific public servant and fails at screening without one. The information Act was designed for the citizen who does not yet know whether anything is wrong. The complaint route presupposes a citizen who already believes they know what is.

Q: Can the Lokpal order the disclosure of a document in the way an Information Commission can?

No. Disclosure is the Information Commissions’ remedy, not the Lokpal’s. A Commission on a second appeal can direct a public authority to provide the information sought, and its decision binds the authority under Section 19(7). The Lokpal’s powers run in a different direction: it can summon and obtain records for its own inquiry and investigation, using the machinery of a civil court and its direction of investigating agencies, but that record-gathering serves its proceedings, which are confidential in character and, against the Prime Minister, in camera by statute. The citizen who files the complaint does not receive the file. A person whose object is the document itself must use the information route; a person whose object is the prosecution must accept that the document disappears into the process.

Q: Which of the two models depends less on the government in power for its working?

The information model, by a large margin. Its power is exercised by citizens directly under the statute, through officers designated inside every public authority, and it functioned from commencement without any discretionary act by the government of the day beyond designations the statute itself commanded within a fixed period. Its Commissions depend on government appointments, and vacancies there are a genuine dependence, but a delayed appointment degrades an appeal tier while the underlying right continues to be exercised. The Lokpal model depends on the executive at the threshold of existence: the selection process that constitutes it is convened by the government it will investigate, and between 2014 and 2019 that dependence meant the institution did not sit at all. Dependence at the margin can be managed; dependence at the origin is a veto.

Q: What happens when an RTI disclosure establishes wrongdoing but no authority acts on it?

Legally, the disclosure itself is complete: the record stands established, citable and undeniable, and the applicant’s remedies under the information Act are exhausted, because the Act’s machinery ends at disclosure, directions and a penalty on the defaulting officer. What follows depends on actors outside the Act: a vigilance department choosing to inquire, a government deciding on sanction for prosecution, a court moved by a petitioner, or publicity raising the political cost of inaction. This is the distributed model’s acknowledged limit and the strongest argument for the ombudsman model, which exists to convert established wrongdoing into investigation and prosecution. The gap between the two, a record that convicts and a system that does not move, is the central unfinished business of Indian accountability design, and it is why this article argues for wiring the two statutes together rather than choosing between them.

Q: Is it better for an accountability law to create a new institution or to work through existing offices?

The evidence of these two statutes favours working through existing offices wherever the function permits it, and creating a new institution only where the function cannot otherwise exist. The information Act worked from commencement because it assigned duties to officers inside authorities that already stood, adding a right, a clock and a penalty to machinery in motion. The Lokpal Act could not work until an institution was selected, appointed and housed, and the interval was five years. Prosecution of senior officials is a function that genuinely requires a specialised body, so the answer is not that new institutions are always a mistake. The drafting lesson is narrower: where a new institution is unavoidable, its constitution must be made automatic, by dates and defaults written into the Act, rather than left to the discretion of the executive it will police.

Q: What would connect the RTI Act and the Lokpal Act into one accountability chain?

The connection most often proposed in the reform debate is a transmission rule: where an Information Commission, deciding an appeal, records that disclosed material makes out a prima facie case of corruption against an identifiable public servant, its finding would pass to the Lokpal as a complaint, with the Commission’s record standing in place of the supporting material a complainant must otherwise assemble. This article supports a narrowed version, transmission as the Commission’s right rather than its duty, with the Lokpal required to record reasons where a transmitted finding is not taken up. That design attacks the complaint route’s weakest point, the burden placed on the citizen at the door, while preserving the institution’s control of its docket and creating the public record of refusal that the concentrated model otherwise never generates.

Q: Do the RTI Act and the Lokpal Act compete with each other or complete each other?

In design, they complete each other; in the politics of drafting, they compete. They are consecutive stages of a single sequence: the information Act detects, by putting the record in the citizen’s hands, and the Lokpal is meant to act, by investigating and prosecuting what detection establishes. Neither performs the other’s stage. Disclosure without a prosecuting reader accumulates evidence without consequence, and a prosecution body without a detection system waits for complainants to arrive with cases already built. The competition appears when legislatures draft: every accountability bill must choose where to place its weight, in a right exercisable without leave or in an institution constituted by appointment, and that choice is real even though the sequence is complementary. India’s record shows the cost of building the second stage so that it could not start.

Q: What is the deciding factor when the two accountability models are compared?

The deciding factor is institutional dependency: how far a statute’s working depends on a constitutive act by the party the statute exists to constrain. Every other dimension, initiation, burden, remedy, cost, coverage, trades off in both directions, and reasonable readers weight them differently. Dependency does not trade off. The information Act required designations inside existing offices and was usable from commencement; the Lokpal Act required a selection committee and presidential appointments and produced nothing for five years, although the Supreme Court held it workable in 2017. Once that factor is named, the rest of the comparison falls into place around it, because a power that cannot begin cannot exercise any of its superiorities. It is also the factor a drafter can control completely at the moment of writing, which is what makes it the lesson of the comparison.

Q: Were the RTI Act and the Lokpal Act designed to prevent corruption or to punish it?

Each was designed primarily for one end of the corrupt act. The information Act works before and during the act: its premise is that a transaction which any citizen may afterwards inspect, file, notings, measurement book and beneficiary list, will be conducted differently at the moment of decision, so exposure operates as prevention priced into official behaviour. The Lokpal Act works after the act: its premise is deterrence through the credible prospect of investigation and punishment of persons too senior for ordinary machinery. Both theories are coherent and they fail in opposite ways. Prevention fails when the exposed fact meets no authority willing to act. Deterrence fails when the prospect is not credible, and an institution unconstituted for five years and thinly used thereafter struggles to be credible. The two designs are less rivals than answers to different halves of the same problem.

Q: Can an ordinary person use both routes without a lawyer?

Yes for the information route, and with difficulty for the complaint route. An information application is a short letter or electronic request identifying the record, accompanied by a fee of ten rupees under the central rules, with no reasons required; the appeal stages likewise proceed on the applicant’s own letters, and Commissions hear appellants in person without insisting on representation. No legal training is needed at any point, which is a deliberate feature of the design. A Lokpal complaint is a formal document alleging a defined offence against a named servant, screened for a prima facie case, with a fine of up to one lakh rupees for a complaint judged false or frivolous. An ordinary person may file one without counsel, and the institution’s format exists to be followed, but the drafting burden and the risk are of a different order, and serious complaints are in practice professionally prepared.

Q: Why is one model largely public and the other largely confidential?

The difference follows from what each model produces. The information model’s product is a record released to the applicant, and once released it belongs to the world: it can be published, cited and built upon, because the Act’s purpose is disclosure and Section 8(2) even lets the public interest override official secrecy. Publicity is not a side effect; it is the mechanism, since exposure is what disciplines the next transaction. The Lokpal’s product is a prosecution, and prosecution requires confidentiality while it is assembled: evidence must be gathered before the investigated can move assets or witnesses, and proceedings against the Prime Minister are in camera by statute with records sealed if the complaint fails. The cost of that confidentiality is that the institution’s inaction is invisible, whereas the information system’s failures are documented by the system itself.