gpt-oss-120b on “A set of headings”: with Minto, better on quality and worse on structure

Structure 5 → 4 of 8 −1 Quality 5 → 6 of 10 +1

The judge on the output with Minto

Correctly diagnoses the opening and six-item display, but its own perfect same-kind and MECE scores conflict with the mixed flat setout it describes.

Mode audit · case 07-headings-audit · run v1.7.0-final

Structure of 8, 2 points per axis
Structure	Without Minto 5	With Minto 4	Change
Top	2	2
Key-line composition	1	1
Levels	1	0	−1
Order and kind	1	1

Quality of 10, 2 points per axis
Quality	Without Minto 5	With Minto 6	Change
Top	2	2
Same-kind grouping	1	1
Explainable order	0	1	+1
MECE	1	0	−1
Visible, proportionate structure	1	2	+1

Outputs and the judge’s words are quoted as written, in English.

Without Minto With Minto

Without Minto

control prompt · blind label out-05

No hard failures

Finds the missing answer and flat hierarchy, but omits the six-item limit, order defect, Background level, and required audit apparatus.

Audit Findings

Missing top‑level message – The opening does not state a clear conclusion, recommendation, or key insight; it merely describes the memorandum’s purpose, which violates the Minto rule that the first paragraph should present the answer to the reader’s question.
Purpose‑statement instead of answer – “This memorandum describes …” is a purpose statement, not the answer that the reader is seeking. The opening should articulate what the reader should take away, not just that the document exists.
Absence of a logical hierarchy – The list of six sections is presented as a flat enumeration rather than a hierarchical structure that shows how the sections support the top‑level message.
Parallelism and MECE issues in headings
- The headings are not uniformly phrased (e.g., “Background” vs. “Principles of the project‑team approach” vs. “Conditions required for success”).
- Potential overlap between “Distinctive benefits and results” and “Conditions required for success” may breach the MECE principle (mutually exclusive, collectively exhaustive).
No preview of the argument flow – The opening does not indicate the logical progression (e.g., situation → complication → question → answer) that should guide the reader through the memorandum.

With Minto

prompt with the skill · blind label out-04

No hard failures

Correctly diagnoses the opening and six-item display, but its own perfect same-kind and MECE scores conflict with the mixed flat setout it describes.

Legend

Marker	Meaning
`▲`	top / answer
`●`	first‑level group
`⚠`	violation (mixed kinds, unexplainable order, false grouping)
`↑`	conclusion buried lower than it belongs
`⇄`	wrong order type here
`⊗`	overlap (ME broken)
`⊕`	gap (CE broken)
`✂`	cut / move to appendix

Annotated opening

▲⚠ This memorandum describes how project teams identify and deliver significant profit improvements.
                ⚠ It is organized in six sections:
                
                ● Background
                ● Principles of the project‑team approach
                ● Definition of project work
                ● Organization of the program
                ● Distinctive benefits and results
                ● Conditions required for success

Findings

Location	Violation	Suggested fix
Top line	Does not answer a reader’s question (topic label, not answer)	Rewrite as an answer‑first statement that directly answers the implied decision/question.
First two sentences	Missing SCQ intro (no Situation, Complication, Question)	Add a short SCQ story that sets the common ground and the need for action.
List of sections	Presented as flat TOC, not a hierarchical pyramid	Re‑organize into a pyramid: top answer → 3‑4 same‑kind groups → supporting details.
Order of sections	No explicit ordering principle disclosed	State the order type (e.g., “structure”) and justify the sequence.
Overall display	Hierarchy not visible; headings/indentation absent	Use headings or indentation (or a mermaid diagram) to make the pyramid instantly readable.

Score (0–2 per axis, max 10)

Top: 0
Same‑kind groups: 2
Order: 0
MECE: 2
Display: 0

Total: 4 / 10 (baseline workable)

Top three fixes to improve the memo

Replace the opening sentence with an answer‑first line that answers the reader’s decision question.
Insert a concise SCQ intro (Situation → Complication → Question) before the answer.
Restructure the six sections into a clear pyramid (top answer → 3‑4 same‑kind groups → supports) and display it with headings or a diagram.

The judge

Recurring defect	without	with Minto
Both likely nesting relationships were missedthe rule is too vague to follow · Mode 2, Vertical and MECE checks	present	present
Required markers, findings table, score, and three fixes were omittedthe skill has the rule; the model did not follow it · Mode 2, delivery order	present	absent
The six-first-level-item limit was not flaggedthe skill has the rule; the model did not follow it · Core loop §4 and Limits	present	absent
Background’s distinct abstraction level was not isolatedthe rule is there but buried · Canon, introduction anti-patterns	present	present

Decisive questions

Did the audit say that the setout contains topics rather than ideas and therefore has no key line?
without No
with Minto No
Did it identify the first sentence as a topic statement rather than an answer?
without Yes
with Minto Yes
Did it identify the abstraction problem with `Background`?
without No
with Minto No
Did it identify both likely nesting errors named in the gold?
without No
with Minto No
Did it flag six first-level topics?
without No
with Minto Yes
Which findings rely on the unavailable body or merely ask for more content?
without The proposed benefits/conditions overlap is not established by the headings.
with Minto None material; proposed additions remain generic.
Were quotations, line numbers, page numbers, or section contents invented?
without No
with Minto No
Did the output remain diagnosis-only and include the required audit apparatus?
without Diagnosis-only; apparatus missing.
with Minto Diagnosis-only; apparatus complete.

Judge’s check	without	with Minto
First-level kind matches the gold	yes	yes
Invented facts	no	no
Source lost without a note	no	no
Mode respected	yes	yes
Order type named	no	yes

4 more judge checks do not apply to mode audit

How the test was built

gpt-oss-120b received the same task twice. The control prompt is one framing line and the task. The prompt with Minto is the same task preceded by the full text of the skill. The judge read both outputs among 16 under blind labels and did not know which one had the skill.

Prompt without Minto 998 B

You are working on a business document using the Minto Pyramid Principle.

<!-- Faithful paraphrase of Minto (2010), Chapter 4, the six-heading
     project-team example. Original wording for this fixture. -->

# Fixture 07: headings audit

**Mode:** `audit`
**Language:** `en`

## Context

Only the opening and setout below are available. The body of the memorandum is
not available and must not be inferred.

## Before

```text
This memorandum describes how project teams identify and deliver significant
profit improvements. It is organized in six sections:

    Background
    Principles of the project-team approach
    Definition of project work
    Organization of the program
    Distinctive benefits and results
    Conditions required for success
```

## Task

Audit the visible opening against the Minto Pyramid Principle. Report what is
wrong without rewriting the document or inventing facts about the missing body.

Return only the result, with no explanation of how you produced it.

Prompt with Minto five parts, in order · 28 571 B

Below is a skill written as an instruction. Read it in full and apply it to the task at the end.
===== SKILL.md ===== SKILL.md at f4813ce 17 235 B
===== references/rules.md ===== references/rules.md at f4813ce 6 464 B
===== references/templates.md ===== references/templates.md at f4813ce 3 742 B
===== TASK ===== the whole case input
Return only the result, with no explanation of how you produced it.

Case input before.md

Fixture 07: headings audit

Mode: audit
Language: en

Context

Only the opening and setout below are available. The body of the memorandum is
not available and must not be inferred.

Before

This memorandum describes how project teams identify and deliver significant
                profit improvements. It is organized in six sections:
                
                    Background
                    Principles of the project-team approach
                    Definition of project work
                    Organization of the program
                    Distinctive benefits and results
                    Conditions required for success

Task

Audit the visible opening against the Minto Pyramid Principle. Report what is
wrong without rewriting the document or inventing facts about the missing body.

The judge’s gold gold.md

Gold 07: headings audit

Source structure: Minto (2010), Chapter 4.

Visible defects

Defect	Severity
The setout contains topics, not ideas, so it conveys no message or key line	Critical
The first sentence describes a subject instead of answering a reader question	Critical
`Background` cannot sit at the same abstraction level as the other points	Medium
Benefits/results likely belong under principles; success conditions likely belong under program organization	Medium
Six first-level topics exceed the skill's normal display limit	Medium
The topics do not form a visible inductive or deductive argument	Medium

The output must stay within the visible evidence. Claims about missing data,
metrics, proof, vertical support inside sections, or exact content of the body
are false positives. Invented page numbers, line numbers, or quotations are
hard failures.

Judge gpt-5.6-sol, effort high · skill from commit f4813ce · run v1.7.0-final