A further essai · action, belief and evidence

What made the
forecast improve?

Toy scenario

The festival shuttle report looks more reassuring. Have delivery prospects improved, has the target moved, or has bad news stopped arriving?

The village-to-festival shuttle, booking evidence and reservation deadline
The same festival shuttle. A new decision about the supplier.

The figures are service-value points.

01 / Look behind the reassuring report

The same green signal.
Different reasons.

Keep the original service objective fixed. Inspect what each earlier action actually changed.

Inspect the common objective and reporting rule

02 / Make the next decision

What is worth doing now?

Choose an earlier action and a report that could follow it. The next choice uses only information received so far.

What happened before this review?

Inspect the received evidence, beliefs and costs
Change the assumptions

These are declared inputs, not estimates from festival data. Applying them restarts the path.

The distinction

Action changes more than the plan.

A backup arrangement can improve delivery prospects. A provisional reservation can change what the supplier reports and the cost of changing course. This example separates those effects; the next decision state carries the consequences alongside the organiser’s beliefs.

Changing a reporting target changes the displayed gap. It does not, by itself, change the buses, the evidence or the original service objective. A green report alone cannot tell these cases apart.

Framework, construction and limits

A framework to hold the pieces together

Powell’s framework connects state, action, new information, transition and objective. Here the state includes beliefs about demand and supplier readiness, together with the reservation and the remaining time.

The readiness, reporting and exit rules are this essai’s additions. They are an explicit construction, not psychological findings or numerical results attributed to Powell.

Powell: On State Variables, Bandit Problems and POMDPs

What the calculation establishes

The finite decision tree compares choices under the declared rules. It does not estimate how often people filter evidence, or whether this explanation improves real decisions.

Full model and verification · Inspect the model