Audience Knowledge-State Walkthrough¶
Evaluation walkthrough — instantiates Event–Narration Order Decoupling
Steps a stand-in reader through the presentation one beat at a time, recording what they now know, infer, and misunderstand, to catch where reveal order misleads.
An Audience Knowledge-State Walkthrough treats comprehension as a running variable rather than an end-state. A reviewer plays the part of a first-time reader and moves through the narration in exactly the delivered order, pausing after every unit to write down three things: what this reader now knows for certain, what they will infer (rightly or wrongly) from the order and framing, and what they still misunderstand or lack. The record is a knowledge-state trajectory — a step-by-step ledger of a mind being built by the telling. Because the walkthrough follows the presentation order, not the event order, it is the mechanism that catches the moment a reveal plants a false belief: the beat where the reader "learns" that A caused B simply because A was shown first. It is a test, not a design tool; its output is a list of misreads to fix, each tied to the step that caused it.
Example¶
A product team ships an onboarding flow for a new analytics dashboard. The flow opens with a flashy "here's your finished report" screen, then teaches data-connection, then filtering, then sharing. Before release, a reviewer runs a knowledge-state walkthrough as a brand-new user. After screen one she notes: knows the tool makes reports; infers the report is already populated with her data (it is not — it is sample data); lacks any idea she must connect a source first. That single inferred-but-false belief predicts the exact support ticket the team later dreads — "where did my numbers go?" She continues beat by beat, and by the connection step records that the earlier false inference is never corrected, only compounded. The walkthrough's deliverable is not a redesign but a defect list: step 1 induces a false "data is live" belief; add a "sample data" label or reorder so connection precedes the preview. The team fixes two screens and the phantom ticket disappears from the beta cohort.
How it works¶
The walkthrough's discipline is that the reviewer may only use knowledge the delivered order has already supplied — no peeking at the full chronology, no author's-eye omniscience. Each step produces a small three-column entry (know / infer / still-missing), and the reviewer explicitly flags any inference that is plausible but wrong given the true event order. Every such flagged misread is checked against the unit's stated purpose: did this beat achieve what it was placed to achieve, or did it also smuggle in a distortion? Steps that pass silently; steps that mislead become defects with a proposed fix (relabel, reorder, add an orientation cue, or supply the withheld context sooner). Running the same script with two or three reviewers of different backgrounds surfaces misreads that any single stand-in would rationalize away.
Tuning parameters¶
- Reader profile — how naïve or expert the stand-in is. A novice profile catches orientation failures; an expert profile catches subtler false-causality inferences. The trade-off: the wrong profile flags the wrong defects.
- Granularity — whether a "step" is a paragraph, a scene, a screen, or a whole section. Finer steps catch more but cost more time and generate noise on beats that never mattered.
- Panel size — one reviewer is fast and cheap; several expose idiosyncratic misreads but must be reconciled.
- Inference strictness — how aggressively the reviewer records possible wrong inferences versus only likely ones. Loose settings over-report; tight settings miss the rare-but-costly misread.
- Stop condition — walk the whole account, or stop at the first objective-defeating misread and fix before continuing.
When it helps, and when it misleads¶
Its strength is that it converts a vague worry — "will people get confused?" — into located, reproducible defects tied to specific beats, and it is almost the only mechanism that detects reveal-order-causality illusion from the reader's side rather than by inspecting the text. It is especially good at catching the false belief that is never explicitly stated but is unavoidably inferred.
Its central failure mode is the curse of knowledge:[n1] the reviewer already knows the true chronology and must actively suppress it, and any lapse turns the walkthrough into a rubber stamp that "confirms" clarity the real audience will not have. It also generalizes poorly from a single stand-in — one reviewer's misread may be idiosyncratic, and one reviewer's non-misread may be luck. The guarding discipline is to recruit reviewers who genuinely lack the chronology, to script the exact stopping points in advance so the reviewer cannot skim ahead, and to treat a clean walkthrough from a single reader as suggestive, not proof.
How it implements the components¶
reader_or_user_knowledge_state_model— the walkthrough is this model, made concrete as a per-step ledger of what the audience knows, infers, and misses.disclosure_objective— each step's misreads are judged against what that beat was placed to accomplish, so the objective is the pass/fail rubric.fidelity_audit_checkpoint— the run is a fidelity test executed from the audience's vantage: does the delivered order preserve a recoverable, undistorted understanding?
It does not maintain the two order maps or the cell-level crosswalk (event_chronology_map, narration_sequence_map, event_to_discourse_mapping) — that is the Fabula–Syuzhet Matrix; and it does not test fidelity by comparing two drafts (revision_crosswalk_control) — that is the Timeline Disclosure Diff.
Related¶
- Instantiates: Event–Narration Order Decoupling — supplies the reader-side fidelity test the dual-order design is validated against.
- Consumes: Reveal-Order Storyboard — the designed narration sequence the walkthrough steps a reader through.
- Sibling mechanisms: Dual Timeline Board · Fabula–Syuzhet Matrix · Fact–Argument Crosswalk · Flashback/Foreshadowing Plan · Interactive Timeline Toggle · Investigation-Report Chronology Appendix · Reveal-Order Storyboard · Temporal Marker Style Guide · Timeline Disclosure Diff
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Steps a stand-in reader through the presentation one beat at a time, recording what they now know, infer, and misunderstand, to catch where reveal order misleads, making its operative form a bounded evaluation of existing evidence or work that produces a finding or disposition.
Independent corroboration: The frozen evidence defines Audience Knowledge-State Walkthrough as 'Steps a stand-in reader through the presentation one beat at a time, recording what they now know, infer, and misunderstand, to catch where reveal order misleads', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Human-Computer Interaction
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Specialized
Rationale: Usability walkthroughs model a first-time user's knowledge and inference step by step through a presented sequence.
Related originating lineages:
- Cognitive Science — Mental-model and curse-of-knowledge research explain evolving reader state.
- Communication & Media Studies — Narrative ordering and audience reception shape what is inferred at each beat.
Review resolution: Both reviewers identify HCI cognitive walkthroughs as primary. Cognitive models of knowledge and communication sequencing materially shape the narration-focused extension; its stepwise knowledge ledger is an Encyclopedia synthesis with specialized evaluation use.
Attribution caveat: The knowledge-state ledger adapts HCI cognitive walkthroughs to reveal order and narration.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; medium confidence.
Notes¶
[n1] The curse of knowledge is the cognitive bias by which someone who knows something struggles to model a mind that does not — an expert cannot easily recover the confusion of a novice. It is the standing threat to any walkthrough run by a person who already knows how the story ends, which is why the mechanism leans on genuinely uninformed stand-ins. ↩