Bolt Series · Human Performance Calibration · Article 15
External judgement is not the same as external truth
Coaches matter because they can see things athletes cannot.
Teachers matter for the same reason.
But another perspective is still a perspective.
It can be informed, experienced and useful.
It can also be wrong.
This matters because once an external observer is given authority, their estimate can start shaping the person being estimated.
Sport science actually studies coach-athlete disagreement
A systematic review and meta-analysis comparing training load perceived by athletes with training load planned or perceived by coaches found that overall agreement could be reasonably good, while disagreements still appeared under specific conditions.
That is enough to make an important point:
The coach’s estimate is evidence. It is not automatically the final state of reality.
The athlete may be wrong.
The coach may be wrong.
The interesting moment is the disagreement.
Why can a good coach still misjudge?
Several reasons are possible.
- The coach cannot directly feel the athlete’s internal state.
- A single performance may be unusually good or bad.
- The coach may rely too heavily on past reputation.
- The athlete may have changed faster than the coach’s mental model has updated.
- The visible outcome may hide a different internal mechanism.
- The coach may be judging one capability while the athlete is actually struggling with another.
- The training task may not transfer cleanly to competition.
None of these possibilities means coaching is unreliable in general.
They mean every human observer has a measurement position and therefore has blind spots.
Authority can freeze an outdated estimate
This becomes more dangerous in education.
A teacher decides a child is weak.
The child receives easier work.
Because the work is easier, the child gets fewer opportunities to demonstrate higher capability.
The original judgement now looks increasingly confirmed.
But perhaps the system has stopped testing the possibility that the child can do more.
This is why calibration needs repeated opportunities for correction.
The learner can sometimes correct the teacher
Imagine a teacher says:
“You do not understand this topic.”
The learner says:
“I do understand it. I ran out of time on those questions.”
Who is right?
Do not decide by authority.
Test the claim.
Give the learner a fresh problem without the same time pressure.
Ask them to explain the method.
Then add time pressure again.
Now the disagreement has produced an experiment.
The best coach wants to be correctable
A weak model of authority says:
“I am the expert, therefore my judgement is right.”
A stronger model says:
“I have expertise, therefore I should know how to test my judgement and update it when the evidence disagrees.”
That difference is enormous.
Expertise should improve calibration.
It should not make calibration unnecessary.
Bolt and Mills are useful because the relationship was not one-way
Bolt publicly acknowledged that Mills could see small technical details he could not see himself.
But Bolt also reported his own readiness, strengths, weaknesses and internal experience.
The athlete was not merely a machine receiving commands.
The coach was not merely an observer issuing verdicts.
The relationship worked because different information sources could meet around performance.
What should happen when teacher and student disagree?
Do not immediately choose a winner.
- State the two competing interpretations clearly.
- Ask what evidence supports each one.
- Identify what evidence would distinguish them.
- Run another task or observation if possible.
- Update whichever model the new evidence weakens.
This protects both sides.
The learner is protected from a mistaken external label.
The teacher is protected from being forced to accept every student self-report as true.
The evidence gets the final opportunity to speak.
Why this matters for education
A good educational system should not ask children to distrust every external judgement.
Nor should it teach them that authority is never mistaken.
It should teach something more useful:
Respect expertise. Examine evidence. Keep every model correctable.
Sometimes the student needs to update.
Sometimes the teacher does.
A calibration system becomes trustworthy not because somebody is never wrong, but because being wrong can be discovered and repaired.
eduKate Sengkang 