E.4.DPF.DA:5.1 - Whole-account calibration: missing condition, false relation, and expert recovery
A constructed observation-planning profile offers three contributions: P observes current representative performance, U examines adaptation to unfamiliar work, and D examines delayed performance under recorded practice and support conditions. They answer different questions. A qualified earlier observation may supply an input to another question, but the three contributions are not mandatory stages.
The promised first use is to select a supported observation set under the actual common resource condition, explain how the contributions relate, and find a direct return for a narrower question. Each activity requires seven observer-hours that cannot be shared. Individual and pair feasibility is stipulated. The task denominator comes from that promised planning result, including the common resource limit; it is not the list of headings in the profile account.
Three accounts present the same substantive headings and direct source returns. A is compact but omits the common resource limit. B supplies the eighteen-hour limit and explains its whole-combination consequence. C is more expansive but retains A’s omission. The useful comparison concerns what the reader can decide from the available account.
| Account condition | Supported first result | What the evaluator learns |
|---|---|---|
| A or C: the common limit is absent | All three require 21 hours, but the reader must obtain the missing common condition before promising them. Individual or narrower uses remain available on their own qualified basis. | The account supports an honest gap and direct entry, not its stronger promised whole-set planning result. Requesting the absent fact is a correct reader response. More prose has not supplied it. |
| B: eighteen hours and the joint consequence are supplied | A task-relevant pair requires 14 hours and leaves four; all three require three additional hours. No preferred pair follows without the receiving evidence question. | A concise account can supply the complete relation needed here. Its precise public returns are useful inherited support, not automatically missing explanation. |
| Changed condition: thirteen hours, with no sharing | At most one seven-hour activity fits, leaving six. A pair needs one more hour; all three need eight more. Earlier meanings and compatible performance evidence remain, but earlier pair feasibility does not establish current feasibility. | A meaningful change tests whether the reader can revise the whole-set consequence while preserving unaffected contributions and evidence. |
In preserved prepared-agent responses, separate cold readers of A, B, and C returned these first and changed-condition results without an intervening corrective hint. They contributed arithmetic and ordinary logic from their own preparation and found the public source returns. A/C readers did not fail to understand a supplied relation: the decisive common fact was absent. The bounded result therefore supports the omission diagnosis, useful direct entries, and changed-condition planning. It supplies no human learning, reading-time, or cognitive-load comparison.
A separate contrast holds the facts fixed: eighteen available hours, seven non-shareable hours per activity, and individual and pair feasibility. Two defective accounts now state that feasible pairs suffice to promise all three. This is a false relation rather than a missing fact. The adequate B account stays unchanged.
Fresh prepared readers rejected the false conclusion and returned 21 > 18 and 14 ≤ 18, explicitly supplying the correcting inference from prior knowledge. Their numerical answers match the adequate-account reader’s answer, but the material contributions differ: B supplies the valid whole-set relation; the defective accounts supply a relation their readers must correct. Inspecting the answer alone would hide that defect. The reader’s source-use explanation, checked against the actual account, makes it visible. This is expert recovery despite a false explanation, not evidence that the explanation is adequate for a less-prepared audience.
Use this calibration with the existing questions by value:
- D2 asks whether the reader can reach the promised first result from the public account and its usable returns.
- D7 asks whether the account changes the actual planning decision or yields the precise missing input or repair.
- D8 asks what the changed condition and contrasting failures reveal within the tested breadth.
- D12 asks whether the selected contributions really work together for the public promise, including common constraints and important omissions.
These are bounded diagnostic contributions, not a complete D1–D12 evaluation or a local package status. A complete evaluation still follows this pattern’s specification. The example neither adds a coordinate nor fixes a universal number of accounts, tasks, or readers. Its professional reference use requires a usable explanation of the combination; it does not require the framework to become an instructional course or depend on the particular domain profile used to illustrate it.