Week 11 Session: Calibrating the evidence

Building a nine-mechanism table from your Queue Case Study system — re-tiering the five mechanisms you already classified, and classifying four more from outside for the first time.

The lecture named two separate jobs: add an evidence tier to five findings you already have, and build both axes from scratch for four you’ve never looked at from outside before. This session is where both actually get done, on your own Case Study system, producing the table the Waiting Audit will ask you to extend to all ten mechanisms.

Before the session#

Read the Week 11 lecture first. Bring your Queue Case Study submission — you’ll need your original five findings in front of you, not reconstructed from memory.

Bridge operation 1: re-tier what you already have#

For each of interface, environment, queue structure, priority, and value, find the claim you made about what that mechanism does or how it affects the wait, and ask: how strongly does the evidence I actually presented support that specific claim? Assign directly evidenced, plausible interpretation, or undetermined — leaving the original present / absent / ambiguous / cannot determine finding untouched. A confident status finding can still carry a claim tiered no higher than plausible interpretation; that’s a correct outcome, not a downgrade of your earlier work.

Work through the priority example from the lecture again if it helps: “policy states no priority criterion” is absent and directly evidenced; “no priority mechanism could be identified” is cannot determine and undetermined — the second isn’t a weaker version of the first, it’s a different, equally legitimate finding about a different kind of evidence.

Bridge operation 2: build the four from scratch#

Revisit the same system using your existing Case Study evidence and, where safely and accessibly possible, additional non-intrusive public documentation — you do not need to visit the system again in person. For each of time, attention, expectation, and information:

1. Note what’s directly observable: a published or timed duration; a feature that could occupy attention; a stated estimate or promise; a display or announcement — or the absence of any of these.
2. Assign a mechanism status from that observation alone.
3. State the specific claim you might want to make about effect on experience, and tier it separately — directly evidenced, plausible interpretation, or undetermined — justifying why it doesn’t earn a higher tier than it does.

Use expectation as your model: a stated estimate is present and directly evidenced as a feature; any specific visitor’s actual expectation is cannot determine; a claim that the feature shaped it is plausible interpretation at best, never directly evidenced by the feature’s presence alone.

What you produce#

One table, nine rows:

Five rows carried forward and re-tiered; four rows built from both axes for the first time. Every row needs a one-line justification for the status and a separate one-line justification for the tier — two justifications, not one.

Where control fits#

Control is not a row in this table. State plainly, in a line beneath it, whether your Week 10 experiment is relevant to this same system, and if so, what it tested and how strongly its own design supports that specific claim. That statement travels into the Waiting Audit on its own, exactly as strongly as your experiment actually supports it — it is not folded into the nine-mechanism table above.

What’s out of bounds#

No interviews with staff or customers. No surveys of visitors. No identifying, naming, or photographing any individual. No intrusive observation of any kind. No medical, emergency, clinical, or triage setting. These are the same rules your Queue Case Study already worked under — this session doesn’t relax or extend them.

What this can and can’t show#

A well-built table separates what you can point to from how far you can trust it — it doesn’t turn either into proof. “Cannot determine” and “undetermined” are correct, complete answers wherever your evidence actually stops there; a table with several of them, honestly justified, is a better piece of work than one that pushed every row to “directly evidenced” to look more finished.

Afterwards#

The Waiting Audit, due at the end of Week 12, asks for all ten mechanisms on this same system. You arrive with nine already built here — five re-tiered, four classified from scratch — and the tenth, control, already sitting separately from your Week 10 experiment, bounded exactly as strongly as that experiment supports. The Audit’s job is the whole-system synthesis across all ten, not building any of them from nothing.