When Evaluators Watch Cricket: Why Manual Exam Marking Cannot Self-Regulate

The October 2024 Video That Surfaced an Uncomfortable Truth

In October 2024, students submitted videos to the Deccan Herald showing teachers at an evaluation centre at SJR Pre University College in Bengaluru using mobile phones while marking undergraduate answer scripts for Bengaluru City University (BCU). Some were making calls. Others were watching a cricket match.

Karnataka's Higher Education Minister Dr M C Sudhakar sought an explanation from the university. The BCU registrar's initial response was that the teachers were handling emergency calls and that this did not constitute misconduct warranting immediate action.

The story did not make national headlines. It was a one-day local news item. But it crystallised a problem that examination administrators in India have largely treated as an operational nuisance rather than a structural failure: manual evaluation at scale cannot reliably enforce the basic conduct standards it sets for itself.

The Arithmetic That Should Not Be Possible

Before examining conduct, consider a more fundamental category of failure: arithmetic.

In Bengaluru, a Bangalore University student received 89 marks out of 70 in a semester examination. The maximum score for the paper was 70. The student's mark certificate showed 89. This is not a marginal rounding difference or a data entry transposition of adjacent digits. It is an impossible output — a score that cannot occur in a correctly functioning evaluation process.

A separate Karnataka 2nd PUC case in April 2026 produced a mark certificate showing 111 marks out of 100 in Hindi. The Karnataka Pre-University Examination Board faced sharp criticism following the error. Reports at the time raised the question of whether the evaluation system or the marks-entry software had failed to flag a number that exceeded the paper's maximum.

Both errors should have been caught automatically before they appeared on any mark certificate. A correctly designed system would reject any entered mark that exceeds the maximum allocated for that question or paper. Manual data entry combined with absent validation logic allows impossible numbers to propagate from evaluator to data-entry operator to certificate with no intervention.

Revaluation Swings That Should Not Happen

Beyond arithmetic errors, there is a second category of manual evaluation failure: scores that change dramatically on revaluation in ways that reveal systematic under-marking or over-marking in the original cycle.

Bangalore University blacklisted more than 50 evaluators in January 2025 after a verification exercise for the 2022-23 batch mark cards revealed that students who had failed in the main examination had been marked as passed after revaluation, only to fail again when their answers were checked a second time. The registrar (evaluation) confirmed the blacklisting and noted that the evaluators were barred from two subsequent semester evaluations. That was the announced consequence.

What was not announced was any mechanism that would prevent the same pattern from recurring with a different set of evaluators in the following cycle. Blacklisting addresses a named set of past actors. It does not change the conditions that made the errors possible.

The Karnataka II PUC board data shows a broader pattern. In one reported year, more than 2,100 students saw mark changes after revaluation. In a separate year, the largest single swing recorded was a student who scored 40 marks in the main examination and received 99 after revaluation — a 59-mark difference on the same paper, evaluated by the same board, within the same academic year.

A 59-mark difference is not attributable to interpretive disagreement between two reasonable examiners. It is evidence that one of the two marking attempts was substantially incorrect.

How the Self-Regulation Cycle Works — and Why It Fails

Indian universities have a standard response sequence to evaluation errors:

This response treats each incident as attributable to individual conduct. It does not examine whether the evaluation environment itself — physical isolation, no verification layer, no automated totaling, no structured mark entry with range validation — creates conditions in which errors are systematically probable rather than randomly distributed.

The phone use at BCU illustrates the environment. Evaluators sit in a physical hall with physical answer scripts. There is no technical mechanism preventing them from accessing a phone. The only enforcement is the presence of an observer and professional norms. When those norms are not enforced — and the BCU registrar's initial response suggested they were not — there is no fallback.

Compare this with the technical environment of an on-screen marking system. The evaluator interface is a locked browser session. External applications cannot be opened while the marking portal is active. Each mark entered is validated against the maximum permitted for that question before the evaluator can proceed to the next response. The total is calculated by the system, not transcribed by a data-entry operator. Activity logs record timestamps, mark entry sequences, and session duration for every evaluator on every script. Any deviation from expected patterns — an evaluator who marks 200 scripts in an hour at a rate that is statistically impossible given the script lengths — is flagged for review by the system itself.

None of these controls exist in a physical evaluation hall. No blacklisting policy can retroactively supply them.

The Scale Problem in Manual Evaluation

It is worth quantifying what India's university system asks of manual evaluation each semester.

A university affiliated with 200 colleges, each running three-year undergraduate programmes across four disciplines, may process 50,000 to 1,50,000 answer scripts per semester examination cycle. Each script requires a first evaluator and, for disputed papers, a second evaluator and a moderator. The evaluation window is typically four to six weeks. The evaluators are practising faculty members who also have teaching and administrative responsibilities.

Under these conditions, the conditions that produced the BCU cricket incident are not exceptional. They are probable. An evaluator who is allocated 400 scripts to mark over a two-week evaluation camp, sitting in a physical hall for five to seven hours each day, with no digital supervision, in conditions that require sustained concentration on handwritten text, is operating in an environment that creates the conditions for distraction, fatigue-related error, and conduct lapses.

Blacklisting 50 evaluators is a disciplinary response to 50 cases. The underlying evaluation environment is unchanged. The next cycle will produce new cases with different names.

What Changes with Structured Digital Evaluation

The shift to on-screen marking does not eliminate human judgment in evaluation. It eliminates the conditions that allow human errors and conduct failures to propagate undetected.

Structured mark entry means evaluators enter marks question by question, not as a single total. The system validates each entry against the allocated maximum. The total is computed automatically. A student cannot receive 89 marks out of 70 in a system where no mark entry can exceed 70 and the total is summed by the software.

Audit trails mean every marking action is timestamped and logged. If an evaluator marks 50 scripts in a time window that the system flags as improbably fast, a reviewer can inspect those scripts. The BCU evaluators watching cricket would have produced a session log showing zero marking activity during the period their phones showed live match data — an immediately detectable anomaly.

Double-marking workflows mean a second evaluator reviews scripts where the first evaluator's marks fall outside expected distributions. This is not a punitive measure directed at named individuals; it is a standard quality checkpoint applied uniformly.

The question is not whether these controls are possible. They are deployed in examination systems across India and internationally today. The question is whether universities choose to implement them before the next blacklisting announcement, or after.

The Pattern Will Continue Without Structural Change

Universities that respond to evaluation failures with blacklisting are treating a system problem as an individual behaviour problem. The evaluators who are blacklisted are replaced by evaluators who face the same environment, the same workload, the same absence of automated controls, and the same low probability of detection in the short term.

The student who scored 40 marks and received 99 after revaluation lost examination time, potentially lost a semester, and experienced a trust failure with their institution that blacklisting one evaluator did not repair.

The pattern documented across Karnataka, Punjab, UP, and West Bengal over successive years is consistent: manual evaluation errors occur, students or journalists surface them, institutions announce disciplinary action, and the structural environment remains unchanged.

Universities that have implemented on-screen marking with structured mark entry and activity logging report substantially lower revaluation request volumes in subsequent cycles — not because their evaluation is markedly better in every respect, but because the error categories that drive revaluation (arithmetic mistakes, missed question sections, out-of-range marks) are systematically eliminated by the platform design.

That is not a claim about evaluator quality. It is a claim about system design.

Related Reading