← The JournalThe Science of Judgment

Label the evidence or the tone will upgrade it for you

Verified fact, reported fact, estimate, assumption, prediction, preference — confident tone collapses them into one pile. Examination starts by forcing the labels back apart.

High-stakes decisions fail less often from a lack of claims than from a collapse of claim types.

Someone says: The client will leave unless we agree this week.

The room hears a fact. What arrived was a stack:

  • Client stated a one-week deadline — reported fact (if they said it)
  • The deadline cannot move — assumption
  • The client will leave — prediction
  • Losing them damages revenue materially — estimate
  • The team wants immediate agreement — preference
  • Legal needs ten days — constraint, if verified

Confident tone upgraded the stack into a single object: emergency. Examination starts by forcing the labels back apart. Soft answers thrive in the collapsed pile. So does agreeable AI — fluency is a professional upgrade service for weak claims.

The label set

Use more labels than you want. Friction here is the point.

Load-bearing assumptions

Not every assumption is equal. Find the three to five whose failure would change the choice.

For each:

  1. Why does it matter?
  2. What supports it?
  3. What contradicts it?
  4. How independent is the evidence?
  5. How current is it?
  6. Can it be tested?
  7. What would falsify it?
  8. What happens if it is wrong?

Matthews's version of this discipline is continuous: expose the belief, test its factual basis, rebuild from stronger premises. Trusted sources stay questionable.

Same assumption. Two treatments.

Collapsed
"We know the market wants this." Tone finishes the sentence. Alternatives shrink. Soft progress language arrives.
Labeled
Class

Assumption dressed as verified fact

Support

Which interviews, which dates, which disconfirming attempts

Falsifier

What would force us to drop the assumption this week

Soft answers are unlabeled stacks

Feedback has been positive often means: a few people said encouraging words (reported), we interpreted them as demand (interpretation), we assume that generalizes (assumption), we predict purchase (prediction), and we prefer that story (preference). One soft sentence. Five unpaid labels.

Soft answers to load-bearing questions is what happens when the stack funds irreversible spend. Labeling is how you stop the funding without inventing false precision.

What the audit does with labels

It does not demand omniscience. Invented verified stamps are another soft answer.

It demands that what you are treating as settled is named as settled for a reason — or renamed until the reason is honest. That is why the close separates blind spots and decision risk: many facts were preferences wearing certainty.

Practice for tonight

Take one sentence carrying your open decision. Force every clause into a label. Circle the ones that are assumptions or predictions. Ask which of those are load-bearing. Ask what would falsify each.

If you cannot do that without getting angry, you have found the soft answer that was doing emotional work.

Then audit the decision — not for a nicer paragraph, for a labeled stack the world cannot quietly rewrite when the outcome arrives.

Related: Confidence is not evidence; [A good outcome is not a good decision](/blog/good-outcome-is-not-good-decision); Soft answers to load-bearing questions.

Questions worth asking

What is the difference between a reported fact and a verified fact?
A reported fact is something someone said or wrote. A verified fact is something you have confirmed by an independent check. Tone does not convert reports into verification.
Why separate assumptions from predictions?
Assumptions are treated as true for the decision now. Predictions are claims about the future. Mixing them hides what would falsify the choice.
How does this relate to a decision audit?
The audit's evidence pass is labeling before optimizing. Soft answers and agreeable completion both thrive when labels are collapsed.

Thank you for reading. If this sharpened how you think about a decision you are facing, the instrument is one step away.

  • evidence
  • assumptions
  • epistemology
  • decision quality
The instrument

An essay sharpens how you think. An audit sharpens a decision you are actually facing.

One essay, one email. Not a newsletter unless you want one below.

No hype. No frequency promises. One quiet list for people who care about how they decide.

A response, a correction, a reframe from your own field — considered, not a comment thread.