Cutoff vs scaling vs moderation — what is UPSC's official position?

These three are distinct. Cutoff is the minimum qualifying mark per stage/category. Scaling is subject-wise mathematical adjustment — UPSC does NOT use it. Moderation is examiner-wise standardisation — UPSC DOES use it. Conflating them is the single biggest source of cutoff myths.

The Three Concepts — Side by Side

Concept	What It Does	Used by UPSC?	Officially Disclosed?	Legal/Documentary Basis
Cutoff	Sets minimum qualifying mark per stage per category	Yes	Yes — released after each final result	UPSC Disclosure Regs 2010; CSE Rules
Scaling	Adjusts marks across optional subjects to neutralise difficulty	No	UPSC has denied this in RTIs	Multiple CIC orders; no UPSC document mentions it
Moderation	Adjusts marks within an examiner pool to neutralise strictness/leniency	Yes	Mentioned in Annual Reports	73rd Annual Report 2022-23; CSE Notification

What Each Means in Practice

Cutoff — The Ranking Line

A simple ranking line. If 14,627 are needed for Mains and you rank 14,627, you are exactly at the cutoff. Set per category. Released as a one-page PDF after final result. Cutoffs are mechanical — there is no committee deciding them; the cutoff is whatever mark the Nth-ranked candidate scored, where N is the qualifying count.

Scaling — NOT Used by UPSC

A formula like Z-score = (X − mean) / SD, applied to equalise mean and SD across optionals. State PSCs (RPSC, UPPSC, MPSC) use this. UPSC does not. This has been confirmed in:

S. Krishnan v. UPSC (Delhi HC, 2010-era)
UPSC v. Angesh Kumar (Supreme Court, 2018) — explicitly noted that UPSC "does not employ a scaling formula"
Multiple CIC orders on RTI denials
UPSC Annual Reports (which describe only moderation, never scaling)

Myth-busters: there is no "Anthropology scaling boost" or "Mathematics scaling penalty" at UPSC. Subject-wise performance differences in topper lists reflect syllabus design, GS overlap, and examiner-pool moderation within each optional — not cross-optional scaling.

Moderation — Used by UPSC

Linear transformation of an examiner's marks if their average deviates significantly from the head examiner's benchmark. So if Examiner A's average is 95/250 and Examiner B's is 115/250 on the same paper, both get pulled toward a common mean. Specifically:

Head examiner sets the standard by re-evaluating a sample of each examiner's scripts.
If an examiner's mean deviates significantly from the head examiner's expected mean, a linear transformation is applied — y = a*x + b where a and b correct for slope and intercept differences.
The result: a candidate's raw marks may go up by a few or be pulled down — but this is examiner-side correction, not subject-side.

The 73rd Annual Report (2022-23) is the most recent public reference to this methodology.

Worked Scenario — What Actually Happens to Your Marks

Suppose you write Essay (250 marks) and your script is evaluated by Examiner X, whose pool average is 110/250 vs the head examiner's expected pool average of 125/250.

Your raw mark from Examiner X: 130
Moderation upward (linear shift): your moderated mark might become ~144
This is what appears on your final marksheet.

Note: the moderation is automatic and uniform across X's bundle — every candidate evaluated by X gets the same adjustment. No script-by-script discretion.

Why This Matters to You

Don't pick an optional hoping for "scaling boost" — it doesn't exist at UPSC.
Don't fear a strict examiner — moderation neutralises most of the damage at the bundle level.
Don't second-guess the cutoff — it's mechanical, just a rank line.
Don't trust coaching infographics that show "how UPSC scales Maths to PSIR" — these are fabrications.
Do focus on raw answer quality — that's what survives both moderation and ranking.

Mentor Note

The predictability of cutoffs is what makes UPSC fair. There is no hidden multiplier — only your raw marks vs other candidates' raw marks, adjusted for examiner variance. The Supreme Court in Angesh Kumar affirmed that this opacity is a feature, not a bug — protecting the integrity of evaluation from rent-seeking RTI litigation. Focus on writing better, not gaming the system.