A hypothetical café charges $2 to skip the queue. That's two mechanisms, not one undifferentiated feeling — and whether they're connected to each other is a separate claim from whether either is present at all. This week closes the toolkit — the audited system's own control mechanism, classified for the first time, and the relationships between mechanisms, read only where the evidence actually supports reading them.
A hypothetical café — not a system anyone in this course has actually observed — runs a standard queue and, alongside it, a $2 “priority pickup” option. Two mechanisms are visible immediately: priority (someone is served sooner) and value (money is exchanged for that). Week 9 already taught you to tell those two apart rather than filing the feature under whichever label is convenient. What Week 9 never asked is whether the two are connected — whether the $2 is what actually grants the faster service, or whether they merely happen to sit in the same system without one implementing the other. That’s a third, separate claim, and this week teaches how to make it honestly.
This week’s claim, plainly#
A real system rarely runs one mechanism at a time — mechanisms may depend on each other, interact, or remain independent, and saying which is true for a specific pair is a claim that needs its own evidence tier, exactly like any other claim this course has asked you to make honestly.
This is the last lecture of the course, and it closes two gaps rather than opening a new mechanism. Value remains this week’s required tag — the schema needs one, and Week 9 is where this course first taught you to test whether two mechanisms overlap, so that’s the precedent this week generalises. This is not a second value lesson.
Where the toolkit still has a hole#
By the end of Week 11, you hold nine mechanisms — five carried forward and re-tiered from your Queue Case Study, four built from scratch — each with a status and an evidence tier, on the same system revisited across Weeks 9, 11, and 12. Two things are still missing, and neither is “add another mechanism”:
- Control has never been classified in this system at all. Your Week 10 experiment tested control somewhere — in a simulation, a prototype, or on yourself — but that is not the same claim as “this audited system contains a control mechanism.”
- No claim has ever been made about how two mechanisms in this system relate to each other. Every finding so far has been about one mechanism in isolation.
The tenth row: control, classified from this system’s own evidence#
Control gets exactly the same two-axis treatment as every other mechanism, built fresh from what your audited system actually shows:
- Mechanism status — present / absent / ambiguous / cannot determine. Is there a genuine, no-penalty choice to continue or leave visible in this system — a way out, a pause, an opt-out — or is there evidence that there isn’t one?
- Evidence tier — directly evidenced / plausible interpretation / undetermined, for whatever specific claim you make about it.
Your Week 10 experiment can provide additional evidence about a control-related claim, but it does not by itself establish that the audited system contains a control mechanism. The audited system must still be classified from its own accessible evidence. If your experiment tested a simulation, a prototype, or a different context entirely, that result may be worth citing as additional evidence where it’s genuinely relevant to this specific system — and just as often, it isn’t relevant at all, and the honest thing to write is that the control row is classified from the system’s own evidence alone, with the experiment noted separately as not transferable here. A system with no visible way to leave, pause, or opt out is legitimately absent, directly evidenced by that absence, regardless of what your own experiment found somewhere else.
The relationship claim: a third question, not a merge of the first two#
Two mechanisms can each be individually well-evidenced as present without that telling you anything about whether they’re connected. The connection itself is a separate claim, answered with one of three outcomes:
- Interacting — evidence indicates that one mechanism’s operation depends on, implements, modifies, or is linked to the other.
- Independent — available evidence supports that the mechanisms operate separately in this system.
- Cannot determine — available evidence is insufficient to establish whether they interact or operate independently.
Two mechanisms merely coexisting in the same system does not establish that they are independent. Independence is itself a claim that needs evidence — a system can have both value and priority present, with no evidence either way about whether they’re connected, in which case “cannot determine” is the honest relationship finding, not “independent” by default.
Every relationship claim gets its own evidence tier — directly evidenced, plausible interpretation, or undetermined — separate from the tiers already assigned to the two mechanisms individually. And one boundary matters as much as the tier itself: a relationship claim is about how two mechanisms operate in this particular system, not about whether those mechanisms generally interact across waiting systems. Finding that value implements priority at one café says nothing about whether it does anywhere else — the claim doesn’t survive being carried out of the one system it was made about.
You are not required to consider every possible pair. With ten mechanisms there are up to forty-five; testing all of them would manufacture work, not insight. Identify only the pairs where there’s a plausible, evidence-grounded reason to think a connection might exist, and leave the rest alone. A short relationship log, or an empty one, is a complete and correct result for a system whose mechanisms genuinely don’t show any evidenced connection to each other.
The café, worked in full#
- Priority: present. A stated faster lane exists.
- Value: present. A stated $2 price exists.
- Relationship claim: “the payment buys the priority.” If the system’s own published terms explicitly state that the $2 option grants faster service, this is interacting, tiered directly evidenced — there’s no daylight between the two features to interpret, the terms say it outright.
- Claim: “standard-queue customers therefore feel unfairly treated.” This is undetermined. Nothing about a stated price-for-priority relationship tells you how anyone waiting in the standard queue actually felt about it — that would require asking them, which this kind of observation doesn’t permit.
What this isn’t#
A write-up that says “this system has value, priority, information, and control, and together they make the wait feel modern and controlled, which is why customers keep coming back” fails on every count this course has spent twelve weeks building against: four mechanisms bundled into one undifferentiated verdict, a claim about how the wait feels with no evidence for anyone’s experience, a claim about behaviour (repeat custom) with no evidence for it, and not one relationship among the four actually stated or evidenced — “together” is doing all the work a specific, tiered relationship claim should be doing instead. It’s the same move Maister’s own practitioner propositions risk when read uncritically: several genuinely different mechanisms folded into one intuitive “this is why waiting here feels fine” story. Useful as a prompt for where to look, as this course has said of Maister since Week 2 — never a substitute for naming which specific mechanisms are doing what, and how strongly you can actually show it.
Why this matters#
Waiting is an experience, not merely a duration — the claim Week 1 made testable by splitting what a clock shows from what only the waiting person’s own judgement could report. Twelve weeks and ten mechanisms later, that same split holds at the scale of a whole system: what’s directly observable — a rule, a price, a published term, a feature’s presence — and what’s only ever a plausible or undetermined claim about anyone’s actual experience of it. The Waiting Audit does not prove that thesis; nothing a single system observed a few times over one course could prove it. What twelve weeks buys you is the discipline to say, for any claim about this one system, exactly which of those two kinds of thing you’re actually saying — and to say “cannot determine” or “undetermined” as a complete, honest answer wherever that’s what the evidence supports, rather than rounding up to sound more finished.
What you’re doing this week#
The session has you build the tenth row — control, from your audited system’s own evidence — and a short, selective log of any mechanism relationships your evidence actually supports, closing out the material the Waiting Audit asks you to submit in full.
Teaching team
- Marisol Quaye — convenor