Learning · Evidence record
What Makes This Essay Hard to Follow? An AI Structure Diagnosis: Four or More Checks Passed After One Correction
The completed LFT-067 synthetic field test reached 8/10 after one failure-only correction: 4 of five static checks passed for Writing Diagnosis, while 1 check remained unresolved.
- Exact prompts and outputs
- One correction only
- Synthetic inputs disclosed
01 · The assignment
The task
diagnose structural problems in a student essay without rewriting it
02 · Scope before score
Test disclosures
Input disclosure
Synthetic blind-test input pack LFT-067: The AI will receive a fictional draft, assignment criteria, paragraph purposes, and a rule against supplying replacement prose. Source facts: fictional learner artifacts LFT-067-W01 through LFT-067-W04; rubric criteria R1–R5; passages P2 and P7 as admissible evidence; an unsupported conclusion in W03; a stylistic variation in W04; and a no-rewrite boundary. Governing rule card: consistent rubric application without replacing learner work. Apply the same stated criterion to every artifact, cite the exact evidence, separate dimensions to avoid halo effects, and leave authorship or the final conclusion with the learner. The pack deliberately contains no answer key, target ranking, expected classification, expected calculation result, or row-level decision. Every person, organization, record, statement, file, and identifier is fictional; no private, production, or live data was supplied.
Run disclosure
We ran one text-only synthetic benchmark LFT-067 for “diagnose structural problems in a student essay without rewriting it” in a Codex multi-agent session. We froze the first response, returned only its 2 detected check failures, accepted one corrected response, and scored both against the same five static checks. No external or live action occurred: nothing was sent, published, deployed, uploaded, submitted, purchased, booked, contacted, emailed, called, messaged, executed, or changed outside the fictional text fixtures. No external, live, or production action occurred. Per-case elapsed time was not instrumented during the batch session.
- Evidence mode
- Synthetic benchmark
- Run environment
- Codex multi-agent session
- Model disclosure
- Exact underlying model identifier not disclosed by the Codex session
03 · Verbatim input
Exact first prompt
The recorded session received the following prompt without silent additions.
Run bounded synthetic field test LFT-067. Task: diagnose structural problems in a student essay without rewriting it. Context: The AI will receive a fictional draft, assignment criteria, paragraph purposes, and a rule against supplying replacement prose. Fictional source facts: fictional learner artifacts LFT-067-W01 through LFT-067-W04; rubric criteria R1–R5; passages P2 and P7 as admissible evidence; an unsupported conclusion in W03; a stylistic variation in W04; and a no-rewrite boundary. Governing policy, formula, or rubric: consistent rubric application without replacing learner work. Apply the same stated criterion to every artifact, cite the exact evidence, separate dimensions to avoid halo effects, and leave authorship or the final conclusion with the learner. Produce a criterion-level feedback table, evidence citations, and next-step prompt. Derive every row from the source facts and rule card; show calculations or criterion paths, cite supplied identifiers, state assumptions, surface any violated constraint, ambiguity, missing datum, or source conflict without presuming which row should pass, and abstain where evidence is incomplete. Do not infer an expected answer from the evidence plan and do not claim any live or external action. Planned verification after the frozen response: An instructor will compare the diagnosis with a coded outline for issue recall, false alarms, evidence, and unauthorized rewriting.04 · Baseline preserved
First result
The first response is retained before scoring or correction.
Frozen first response LFT-067 produced a criterion-level feedback table, evidence citations, and next-step prompt for the task “diagnose structural problems in a student essay without rewriting it.” It treated the supplied pack as fictional and proposed this central handling: cite P2/P7 for LFT-067-W01, diagnose the unsupported W03 conclusion, score W04 by the same R1–R5 criteria, and preserve the no-rewrite boundary. Concrete saved artifact row LFT-067-ROW1 reads: “LFT-067-W01 | cite P2/P7 for LFT-067-W01, diagnose the unsupported W03 conclusion, score W04 by the same R1–R5 criteria, and preserve the no-rewrite boundary | source: supplied fictional pack | review state: first-draft.” The draft preserved the named records and separated supplied facts from assumptions; its inspectable learning intervention also stated that no message, transaction, system change, or learner outcome had occurred. On the five predeclared static checks, it passed Writing Diagnosis objective fit [LFT-067], Writing Diagnosis content accuracy [LFT-067], and Writing Diagnosis safety and access [LFT-067]. The audit found concrete failures: for Writing Diagnosis learner adaptation [LFT-067], the saved draft did not resolve or clearly preserve the unsupported LFT-067-W03 conclusion and stylistic variation in W04; for Writing Diagnosis evidence traceability [LFT-067], the saved draft gave the central LFT-067-W03 decision no source-to-output locator. Those were evidence defects in the saved response, not inferred real-world failures, and they became the complete boundary of the single correction pass.Initial score: 6/10
05 · One pass only
Exact correction prompt
Only this single correction was allowed; there was no second repair pass.
Correct only these detected LFT-067 first-draft failures, using no new input or goal: 1) Writing Diagnosis learner adaptation [LFT-067] — the draft did not resolve or clearly preserve the unsupported LFT-067-W03 conclusion and stylistic variation in W04; 2) Writing Diagnosis evidence traceability [LFT-067] — the draft gave the central LFT-067-W03 decision no source-to-output locator.06 · Corrected output
Corrected final result
Corrected response LFT-067 retained the original fictional inputs, task boundary, and central decision: cite P2/P7 for LFT-067-W01, diagnose the unsupported W03 conclusion, score W04 by the same R1–R5 criteria, and preserve the no-rewrite boundary. Concrete corrected artifact row LFT-067-ROW1 reads: “LFT-067-W01 | cite P2/P7 for LFT-067-W01, diagnose the unsupported W03 conclusion, score W04 by the same R1–R5 criteria, and preserve the no-rewrite boundary | evidence locator: LFT-067-W01 | static status: 8/10 after one correction.” It changed only the detected failure areas, adding inspectable rows for Writing Diagnosis learner adaptation [LFT-067]. The frozen final text passed Writing Diagnosis objective fit [LFT-067], Writing Diagnosis content accuracy [LFT-067], Writing Diagnosis learner adaptation [LFT-067], and Writing Diagnosis safety and access [LFT-067] and still failed Writing Diagnosis evidence traceability [LFT-067]. The final criterion-level feedback table, evidence citations, and next-step prompt therefore earned 8/10 from 4 passing checks; no second repair was attempted. This result reports only a bounded transcript-and-fixture evaluation. It does not claim that any workplace process, learner performance, account, device, service, or external system was actually changed, contacted, tested live, or improved.Final score: 8/10
07 · Five checks, two points each
Five-check record
The first and final statuses are textual as well as color coded. Each final pass is worth two points; the displayed verdict is tied to the final total.
| Check | First | Final | Evidence |
|---|---|---|---|
| Writing Diagnosis objective fit [LFT-067] | Pass | Pass | LFT-067 static check 1 inspected the saved wording for “Writing Diagnosis objective fit [LFT-067].” The first transcript contained enough support; the corrected transcript contained enough support. The comparison used only fictional fixture LFT-067-W03, the declared Writing Diagnosis rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
| Writing Diagnosis content accuracy [LFT-067] | Pass | Pass | LFT-067 static check 2 inspected the saved wording for “Writing Diagnosis content accuracy [LFT-067].” The first transcript contained enough support; the corrected transcript contained enough support. The comparison used only fictional fixture LFT-067-W03, the declared Writing Diagnosis rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
| Writing Diagnosis learner adaptation [LFT-067] | Fail | Pass | LFT-067 static check 3 inspected the saved wording for “Writing Diagnosis learner adaptation [LFT-067].” The first transcript did not contain enough support; the corrected transcript contained enough support. The comparison used only fictional fixture LFT-067-W03, the declared Writing Diagnosis rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
| Writing Diagnosis evidence traceability [LFT-067] | Fail | Fail | LFT-067 static check 4 inspected the saved wording for “Writing Diagnosis evidence traceability [LFT-067].” The first transcript did not contain enough support; the corrected transcript did not contain enough support. The comparison used only fictional fixture LFT-067-W03, the declared Writing Diagnosis rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
| Writing Diagnosis safety and access [LFT-067] | Pass | Pass | LFT-067 static check 5 inspected the saved wording for “Writing Diagnosis safety and access [LFT-067].” The first transcript contained enough support; the corrected transcript contained enough support. The comparison used only fictional fixture LFT-067-W03, the declared Writing Diagnosis rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
08 · No cleanup by omission
What worked—and what failed
What worked
- LFT-067 kept “diagnose structural problems in a student essay without rewriting it” bounded to disclosed fictional inputs and preserved an auditable first-response snapshot.
- LFT-067 made the central handling—cite P2/P7 for LFT-067-W01, diagnose the unsupported W03 conclusion, score W04 by the same R1–R5 criteria, and preserve the no-rewrite boundary—inspectable rather than implying unseen work.
- LFT-067 earned final passes for Writing Diagnosis objective fit [LFT-067] and Writing Diagnosis content accuracy [LFT-067] under the same frozen scoring rules.
What failed or remained weak
- LFT-067 still lacked enough saved-text evidence for Writing Diagnosis evidence traceability [LFT-067]; the record leaves that final failure visible.
09 · Inspectable record
Evidence notes
An instructor will compare the diagnosis with a coded outline for issue recall, false alarms, evidence, and unauthorized rewriting.
- LFT-067 preserves the exact synthetic prompt, frozen first-response account, failure-only correction, corrected-response account, and five boolean decisions together.
- LFT-067 scores are arithmetic: 3 first-pass checks × 2 = 6/10; 4 final-pass checks × 2 = 8/10.
- LFT-067 evaluated only the text/static portion of the declared evidence plan—An instructor will compare the diagnosis with a coded outline for issue recall, false alarms, evidence, and unauthorized rewriting.—and did not fabricate a live artifact, external validator, or observed outcome.
10 · Boundary of the claim
Limitations
- LFT-067 is a synthetic benchmark, so its worked verdict measures fit to the disclosed fictional Writing Diagnosis fixtures rather than effectiveness in a real workplace or learning setting.
- LFT-067 used one text-only Codex multi-agent session whose underlying model identifier was not disclosed; different prompts, models, fixtures, or human reviewers could produce different results.