India’s grievance portal is a real achievement — and a live map of where the state hesitates. The next reform is to turn AI on the absent decision, and make decision velocity count the way integrity does.

Something quietly remarkable happened in Shillong this week. Officials from across the country gathered for the national conference on NextGen Administrative and e-Governance Reforms, and the centrepiece was not a highway or a dam but a complaints portal. CPGRAMS, the government’s grievance platform, now receives nearly twenty-five lakh grievances a year — up from about two lakh in 2014 — and a chatbot speaking twenty-two languages walks citizens through the process.

Give the achievement its due. A state that builds a counter where any citizen can walk up — in her own language, at midnight, without a tout — and lodge a complaint that lands on a named officer’s dashboard is a state taking its citizens seriously. For most of our administrative history, the aggrieved Indian had two options: the dak register or despair. Today a bot greets him in his mother tongue. The officials who built this have earned their applause.

But the number deserves a harder look. Why has the flow of grievances multiplied twelve-fold in a digital decade? Part of the answer is benign: awareness. Every grievance redressed recruits ten new complainants. Rising numbers partly measure rising trust — much as rising crime registration often signals better policing, not worse streets.

Part of the answer, though, is structural, and this is what thirty-three years inside the machine taught me. Most of what citizens complain about is not corruption. It is not incompetence either. It is the pending decision: the pension not sanctioned, the mutation not recorded, the permission neither granted nor refused. The government’s own data says so. DARPG’s analysis of five lakh recurring complaints (2022–25) found the biggest clusters were stopped PM-KISAN payments, stuck tax refunds and unresolved bank-fraud cases. Entitlements undelivered, decisions unmade. Not bribery — delay. And decisions stall not because our officers are timid by temperament but because of the conditions under which they decide. Our accountability architecture scrutinises action with ferocity — every signature can summon an audit para, a vigilance query, a court notice years later — while inaction stays invisible to it. I never saw an inquiry opened into a file that sat still. Under those conditions, the rational course for the honest officer, above all, is not to decide. In my book The Fear Tax, I call this Rational Abdication, and its cost to the citizen the Fear Tax. Twenty-five lakh grievances a year is, in large measure, that tax being collected.

The rejoinder writes itself: CPGRAMS disposes grievances in thirteen days on average. But disposal is not resolution. Five lakh citizens returning with the same complaint means files closed and decisions untaken — speed at the counter, stillness in the office behind it.

The point bears insisting on: this is a defect of design, not of character. Haryana gave me the proof. When we were losing crores to cyber fraud, we changed exactly one condition — bank officers sat physically beside the police staff answering the 1930 helpline, with authority to freeze a stolen transfer while the victim was still on the line. Nobody was exhorted. Nobody was packed off to motivational training. The same people, under new conditions, behaved differently — and recovery of defrauded money rose from seven per cent to thirty-six per cent within a year. Change the decision conditions and behaviour changes overnight. Sermons change nothing.

Which is why the AI being celebrated in Shillong is a bigger opportunity than its builders may realise. Until now, no instrument of the state could see an absent decision. The auditor needs a transaction. The court needs an order. Vigilance needs a signature. A file that never moved escaped all three. But AI reads patterns, not signatures. The same systems that route twenty-five lakh grievances can be turned around to ask questions no auditor has ever asked: which offices generate the most grievances per sanctioned post; where the median file overshoots its own citizen’s charter; which docket shows the stillness of Rational Abdication. CPGRAMS is already a live map of where the state hesitates. Let AI audit inaction with the rigour we have always reserved for auditing action.

In The Fear Tax, I give this instrument a name: Stage-Tracking. Every file passes through a handful of identifiable stages — received, examined, consulted, decided, dispatched — and each stage carries a timestamp. Reviewed weekly, those timestamps become a mirror held up to the office — and what is measured weekly gets managed. This needs no new software and no new posts. AI simply lets the mirror cover the whole government at once.

The mirror alone, however, will frighten rather than cure, unless the state also changes what happens to the officer who does decide. That is the companion protocol: Safe Harbour. It is a written assurance that a decision taken to the agreed standard — reasons recorded, decider named — will not be reopened by vigilance, audit or disciplinary proceedings on outcome alone. Mala fide, alleged and prima facie established, is the only key that reopens the file. The state has gestured at this before — Section 17A of the Prevention of Corruption Act has, since 2018, required prior approval before an officer’s official decision can even be inquired into. But 17A guards one gate. The audit para, the departmental proceeding and the court’s stricture walk through the others. Safe Harbour must bar them all, in writing, or it protects nothing.

Track the stalled file, shelter the honest decision, and decision velocity becomes as consequential as integrity. Publish, office by office, the median time-to-decision. Weigh an officer’s record of deciding — defensible mistakes included — in his favour at empanelment and promotion, as we weigh his integrity record. The officer who decides ten cases and errs once must stand taller than the officer who decides nothing and errs never. Until that inversion happens, every new platform will lighten the citizen’s burden of complaining while leaving his reason to complain untouched.

Shillong showed a state willing to listen at scale. The better reform is to make listening unnecessary — a state that decides before the citizen needs to complain. The technology is finally equal to the task. The question is whether the accountability system will let it look. / DAILY WORLD /

(The writer, a retired IPS officer, was Director General of Police, Haryana. His forthcoming book The Fear Tax: Decision Velocity for Viksit Bharat examines administrative inaction.)