Small Group Tutorials

Here to help students catch up, keep up, and move ahead. Book a consultation here.

Bolt 35 — Calibration Is Not Human Worth

Bolt Series · Human Performance Calibration · Article 35

There is a line Bolt must never cross

This series has argued for better measurement.

Better self-estimation.

Better external observation.

Better prediction.

Better updating from evidence.

All of that is useful only if we keep one boundary absolutely clear:

A measurement of performance is not a measurement of human worth.

A person can run slowly.

Score poorly.

Be unprepared.

Misjudge their own ability.

Fail repeatedly at a particular skill.

All of those can be true statements about performance.

None of them logically becomes:

“Therefore this person is worth less.”

Educational measurement already gives us part of this safeguard

Modern validity theory does not treat a test score as an unlimited statement about a person.

Validity concerns whether evidence and theory support particular interpretations and uses of scores.

Educational Measurement, Fifth Edition (2025), emphasises that more ambitious claims require more validity evidence and that inferences from an observed assessment to wider real-world performance have to be justified.

This gives us an important discipline.

If a Mathematics paper supports a claim about performance on sampled Mathematics tasks, we should not silently expand the claim into intelligence, character, future potential or human value unless there is separate evidence for the additional inference.

And no school test is designed to measure a person’s dignity.

The category error happens very easily

A child receives 42%.

The factual statement is:

“I received 42% on this assessment.”

Then the sentence changes:

“I am a 42% student.”

Then perhaps:

“I’m stupid.”

Then:

“I’m a failure.”

Notice how far the final statement has travelled from the original observation.

The score has moved from a performance result to a total identity claim.

Nothing in the number justified that expansion.

Grades can become tied to self-worth

Psychological research gives us a reason to take this boundary seriously.

Research on contingent self-worth examines what happens when people base their sense of worth heavily on success or failure in particular domains.

One study of university students found that daily self-esteem rose on days students received good grades and fell on days they received poor grades, with stronger effects among students whose self-worth depended more heavily on academic competence.

More recent research continues to examine links between academic contingent self-worth, unstable self-esteem and mental-health risks, while also showing that the construct is more nuanced than one simple harmful dimension.

The lesson for Bolt is not that students should stop caring about grades.

It is that grades should inform learning without being allowed to carry a psychological job they were never designed to perform.

A grade can tell you something about a performance. It should not have to tell you whether you deserve to feel valuable.

Accurate self-knowledge can include weakness without becoming self-rejection

Suppose a learner says:

“My current algebra transfer is weak.”

That can be an accurate statement.

It is useful because it is specific.

It identifies something trainable.

Now compare:

“I am weak.”

The mechanism has disappeared.

The whole person has swallowed the local performance problem.

Calibration should move in the opposite direction.

More resolution.

Narrower claims.

Clearer uncertainty.

More correctable mechanisms.

This boundary protects high performers too

It may seem that only low-scoring students need protection from identity collapse.

High performers can face the same problem in reverse.

A child becomes:

“the smart one.”

Excellent scores become part of the identity.

Then one serious failure arrives.

If worth has been tied to visible ability, the failure threatens much more than one performance.

Classic research on praise for intelligence found that ability-focused praise could make children more performance-oriented and more vulnerable after failure than praise focused on effort.

The broader point is not that intelligence should never be discussed.

It is that a child should be able to discover a real weakness without experiencing it as a collapse of the self.

Bolt needs two ledgers

One ledger is about performance.

  • What can I do?
  • What can I not yet do?
  • How reliable is the performance?
  • How accurate is my prediction?
  • What changes under pressure?
  • What should I train next?

The other ledger is not scored.

It contains the fact that the learner is a human being whose value is not reducible to educational output.

Do not merge those ledgers.

The first needs relentless accuracy.

The second is not waiting for a test result.

This makes honest feedback safer, not softer

Some adults avoid accurate feedback because they fear damaging confidence.

But if performance and worth are kept separate, we can be much more truthful.

We can say:

“Your current understanding is not sufficient for this task.”

without meaning:

“You are insufficient.”

We can say:

“You have overestimated your readiness.”

without meaning:

“You are arrogant.”

We can say:

“You are performing far above what you predict.”

without needing to manufacture praise detached from evidence.

Specific truth becomes less threatening when it is not forced to carry a verdict on the whole person.

Why this matters for education

Schools need measurements.

Teachers need judgements.

Students need feedback.

And learners eventually need accurate self-knowledge.

None of that requires pretending everyone performs equally.

It requires something more disciplined:

Measure capability honestly. Keep the claim bounded to what the evidence supports. Never convert the measurement into a numerical description of the human being who produced it.

A learner may need to become better at Mathematics.

Better at writing.

Better at judgement.

Better at knowing themselves.

But human worth is not the hidden denominator waiting underneath the next score.

Bolt must never pretend that it is.

Evidence and further reading