The case against her was that she moved to a compromised computer. The Commission’s own annexure does not list that computer

Seven years’ debarment for unfair means, upheld by nobody who compared two documents. The Delhi High Court says a court must test the allegation candidate by candidate — and this one collapses on a MAC address.

“We are painfully aware that we are treading, here, on quicksand.”

That is the first line of a judgment delivered on 24 September 2026 by a Bench of Justice C. Hari Shankar and Justice Vinod Kumar, and it is an accurate description of the ground. A court asked to review a finding of examination malpractice is being asked to second-guess an expert committee on a matter where the public interest in the purity of the process is overwhelming, and where judicial restraint is not merely conventional but right. The Bench says as much, twice, before deciding against the authorities anyway.

What happened to the candidate

Himanshi appeared in a Combined all-India Examination conducted by the Staff Selection Commission. Following the report of a Committee constituted for the purpose, her candidature was cancelled and she was debarred from participating in any SSC selection for seven years, on the ground that she had resorted to unfair means.

She went to the Central Administrative Tribunal, which allowed her Original Application and set aside the action. The SSC came to the High Court in challenge. Mr. Syed Abdul Haseeb, CGSC, appeared for the Commission; Mr. K.C. Mittal with Mr. Yugansh Mittal, Mr. Pulkit Jain and Mr. Keshav Poonia for the respondent.

The allegation, made against all the debarred candidates, was specific and technical. They had been shifted, during the Tier-I examination, from the terminals allotted to them to other terminals which were compromised; and having attempted the rest of the paper on those compromised terminals, they performed outstandingly.

Two routes, and why the choice matters

Before reaching the facts the Bench set out a framework that is the most portable part of the judgment.

Where candidates at a particular centre are suspected of malpractice, the authorities have two options. The first is to subject every candidate at that centre to a fresh examination — the route usually taken where the use of unfair means is widespread and it is not reasonably possible to identify everyone involved. A court reviewing that decision proceeds differently, keeping in mind throughout the necessity of maintaining purity in the examination process.

The second option, which the SSC took here, is to cancel the candidature of and penalise select candidates identified as the ones who alone employed unfair means. And in that case, the Bench held, the onus on the authorities and the task of the Court both become more onerous. A “one size fits all” approach would be totally contrary to law. The authorities have to establish, in respect of each allegedly delinquent candidate, that she or he employed unfair means — and the Court has to examine, in the case of each candidate, whether the Committee’s view and the consequent action are justified.

That framework does not weaken the authorities’ hand where cheating is systemic; it says that having chosen to punish individuals rather than annul the examination, they must carry an individual case against each one.

The Bench also noted, from State of U.P. v. Baishakhi Bhattacharyya, the settled position on the prior question: where it is possible to identify the candidates who indulged in unfair means, those candidates alone should be debarred and the examination as a whole should not be nullified; where the numbers are very large, or large-scale copying is possible although positive material exists only against a select few, cancelling the whole examination is justified. Here that question did not arise — the SSC’s own case was that it had identified the candidates.

The Committee’s indicia

The Committee’s report set out the factors on which it worked, and the Bench reproduced them. At one centre, it observed that the terminals of almost all candidates had been shifted before closure of the entry gate; that there seemed to be some fixed compromised terminals in every lab; that most suspected candidates had been shifted only to those terminals; that almost none of the shifted candidates had complained about the old terminal not working; that candidates had instead been told by invigilators to shift; that in one or two cases the earlier terminal of a shifted candidate was later used by another candidate, indicating no fault in it and that the shifting was with ill motives; that most suspected candidates were the first to report at the frisking point and then in the lab; and that in some cases CCTV footage appeared to have been interfered with.

Those are reasonable indicia. The Bench’s point is what follows from the way they were recorded: the adverse observations covered all the candidates whose performance was under scrutiny and were therefore generalised in nature, with the annexures to the report carrying the individual detail. So the question in Himanshi’s case was whether the report, read with its annexures, made out a case of unfair means by her.

The discrepancy

It did not, and the reason is documentary.

The SSC’s own affidavit in rejoinder, filed in these proceedings, identified the IP and MAC addresses of Himanshi’s original terminal and of the terminal to which she was shifted. Paragraph 23(d) set out the seat chronology in detail: she was seated at the node with MAC No. 7845C4368CC2 from 13:04:00 to 13:24:53, during which she visited only fourteen questions and gave a final response to seven; she was then seated at the node with MAC No. D4BED9CCAA3B from 13:28:49 to 13:49:02, during which the remaining eighty-four questions were attempted — which the Commission described as a behavioural shift inconsistent with a human test pattern. Paragraph 26 repeated the point: ninety-one final responses in twenty minutes and thirteen seconds, seventy-four of them attempted on first visit, followed by sixteen minutes and fifty-four seconds in which nothing was attempted.

Annexure 5 to the Committee’s report is the list of IP and MAC addresses used more than once at a given venue — the compromised terminals. For the centre in question it lists, among others, the MAC addresses 1803734B8119, D4BED9DB5CFC, 4437E6C27DA1 and FC4DD432273C.

The terminal the Commission says Himanshi moved to is not on that list. As the Bench recorded, the annexures note the MAC and IP addresses of the systems to which candidates had been shifted and which were allegedly compromised; and so far as Himanshi is concerned, no such MAC or IP address has been identified.

The Court called it a glaring discrepancy reflecting near-total non-application of mind, and said that this single factor, by itself, demolishes the allegation that she employed unfair means — apart from various other reasons discrediting it.

The chain, in other words, had a missing link that nobody checked: the allegation was that she performed impossibly well because the machine she moved to was rigged, and the Commission’s own evidence of which machines were rigged does not include it. The statistical pattern in the logs, however striking, proves nothing about unfair means once the mechanism alleged for it disappears.

A criticism of the Tribunal, and a limit on the judgment

The Bench was not uncritical of the order it upheld. The manner in which the Tribunal dealt with the issue, it said, cannot be regarded as satisfactory: there was no detailed examination of the law or the facts even as they applied to the respondents, and the Tribunal had clubbed the cases of thirty-four applicants in individual Original Applications into one judgment that examined the facts only in Himanshi’s case — despite the facts, and in some cases the allegations, being different, differences the judgment failed to notice.

Having said that, the High Court drew exactly the same boundary around itself, and did so deliberately. Its judgment is restricted to Himanshi, the applicant in OA 2953/2022. It is not commenting on the Tribunal’s order so far as it concerns the other applicants, and it should not be regarded as a final pronouncement on the allegation of unfair means against any other candidate whose candidature was cancelled. By its first order of 6 October 2025 the Court had confined the writ petition to Himanshi’s case, with liberty to the SSC to file individual petitions against the others; those petitions have been filed, are listed on other dates, and will be examined individually.

That is the framework applied to the Court’s own work. Having held that individualised allegations require individualised adjudication, it would have been incoherent to decide thirty-five cases on one candidate’s facts — which is precisely the error it identified in the Tribunal.

The penalty question, left open

One exchange is worth recording. Mr. Haseeb, with what the Bench called characteristic forthrightness, acknowledged that debarring Himanshi from any further SSC examination for seven years might be too extreme a penalty.

The Court took the concession but declined to build on it. Had it concluded that the Committee’s finding of unfair means did not merit interference, it observed, it would be arguable whether it had jurisdiction to reduce the penalty below the statutorily envisaged punishment of cancellation of candidature with seven years’ debarment — the validity of that punishment as such not being under challenge.

So the case was not decided on proportionality, and the question whether a court can moderate a statutory penalty of this kind remains open. It was decided on the finding itself: the decision to cancel her candidature and debar her for seven years was not justified on facts or in law.

The order

The writ petition was dismissed with no order as to costs, and resultant orders in Himanshi’s case, in accordance with law, were directed to be passed within four weeks.

What the judgment asks of examination authorities is not lenience. It is that when they choose to name individuals rather than annul a paper — and that choice protects the honest majority, so it is usually the right one — the file against each name must actually hold together. Here the two documents that mattered were both the Commission’s own, and they contradicted each other on the single fact the case turned on. Nobody laid them side by side until the High Court did.