Building the Reference

The sequence

Produce the item to the written standard.

Everyone tastes it at once.

Then everyone builds one.

Taste blind against the reference.

And name the differences aloud.

The decisions

Whether a reference exists. Memory is not a reference — two people comparing memories cannot converge, per Module 50.

Whether to run it blind. Yes, and it is uncomfortable and it is the point.

And whether to insert a duplicate. The same sample twice, unlabeled — if it scores differently from itself, the panel is not calibrated and nothing it produced means anything.

That is a physiological finding rather than a judgment, and framing it that way is what lets people accept it.

What you are reading

The spread between builds.

And each person's consistency with themselves.

What goes wrong here

Calibration against an opinion.

And the duplicate skipped, which leaves the panel's own reliability unknown.

Verification

The duplicate.

The boundary

None specific.