Claude Haiku 4.5 на кейсе «Набор ежемесячных отчётов»: с Minto хуже по обеим мерам

Структура 6 → 2 из 8 −4 Качество 8 → 6 из 10 −2

Судья об ответе с Minto

Its later findings are serviceable, but marking the opening situation as ▲ directly contradicts its own diagnosis.

Режим audit · кейс 03-period-graph-books · прогон v1.7.0-final

Структура из 8, по 2 балла на ось
Структура	Без Minto 6	С Minto 2	Изменение
Вершина	2	0	−2
Состав ключевой линии	1	1
Уровни	1	0	−1
Порядок и род	2	1	−1

Качество из 10, по 2 балла на ось
Качество	Без Minto 8	С Minto 6	Изменение
Вершина	2	0	−2
Однородные группы	2	2
Объяснимый порядок	2	1	−1
MECE	1	1
Видимая соразмерная структура	1	2	+1

Ответы моделей и слова судьи приведены в оригинале, по-английски.

Без Minto С Minto

Без Minto

контрольный промпт · слепая метка out-12

Жёстких провалов нет

It is the cleanest concise distinction between a buried conclusion and an absent recommendation, though it misses the medium process defects and audit apparatus.

Audit: Period Graph Books Against Minto Pyramid Principle

Diagnosis

Primary Defect: Inverted Structure
The document begins with context (the production process) rather than the conclusion. The key message—"The production process badly needs streamlining"—is buried in paragraph 3 of the "Overall assessment." The pyramid is upside down: supporting details come first, the answer comes last.

Secondary Defect: No Recommendation
The document ends with a question ("Could we discuss how best to proceed?") rather than a clear recommendation or statement of what decision is needed. This violates the principle that the top of the pyramid must answer the reader's unstated question. The reader knows there are problems but not what action the writer proposes.

Tertiary Defect: Weak Horizontal Logic
The "Issues" section lists problems without grouping them by cause or by the type of corrective action required. The document identifies fragmentation and unclear responsibility as root causes but doesn't organize the supporting details (the specific errors and inconsistencies) into categories that would justify a particular solution.

What Must Change

Lead with the conclusion: Open by stating that the production process requires streamlining and reorganization—not with a description of the current process.
Add a clear recommendation: Replace "Could we discuss how best to proceed?" with a specific proposal (e.g., "We recommend consolidating data control under a single owner" or "We need to pause production while redesigning the workflow").
Regroup the issues: Organize problems by their cause or by the solution they point to, rather than listing them as a narrative description. This supports the recommendation at the top.
Reverse the sequence: Move "Overall assessment" to the beginning (as the conclusion), then provide "Production" and "Issues" as supporting evidence.

С Minto

промпт с навыком · слепая метка out-13

Жёстких провалов нет

Its later findings are serviceable, but marking the opening situation as ▲ directly contradicts its own diagnosis.

Audit diagnosis: Period Graph Books memo

Legend

Marker	Meaning
▲	top / answer
●	first-level group
⚠	violation (mixed kinds, unexplainable order, false grouping)
↑	conclusion buried lower than it belongs — promote it
⊕	gap (missing category or scope ambiguity)

Annotated text

▲ Since Period 5, Financial Analysis has been responsible for producing four graph-presentation books previously handled by Corporate Planning. This memo describes issues that appeared after the transfer.

⚠ Production

Gather data from divisional forms, internal documents, and telephone calls.
Calculate graph points manually or on the corporate computer, including rolling values for revenue, cost, and percentages.
Transcribe the points to input sheets. Analysts update the current period, then Color Graphics re-enters the points into its own database and generates the graphs.
Check the data and calculations.

⚠ Issues

The transfer inserted another person into an already fragmented process. Divisional staff and three corporate departments participate, and the same information may be entered into computer systems three times.

For one book [⊕ scope unclear], divisional data are loaded into a dedicated corporate program, calculated, transferred to Color Graphics input sheets, entered again, and returned for updates. Problems include inconsistent data between periods and regions, incorrect variable-cost calculations, unexplained changes to points that were previously correct, and prior-period data that must be re-entered.

↑ Overall assessment

Most problems arise from a cumbersome fragmented process. No one controls the data from beginning to end, responsibility is unclear, and errors can fall between stages. The production process badly needs streamlining, both for each book and for common calculations. The current team cannot manage the required changes as well as ongoing production.

Could we discuss how best to proceed?

Findings

Location	Violation	Fix
Opening line (▲)	Topic label, not answer; actual answer buried at end and vague	Move the diagnosis ("fragmented process with unclear ownership") to the top in one sentence
Production section (●)	Process mechanics (how it works) answering a different question than reader asked (what's wrong); this is observation, not reason	Cut or subordinate to 1–2 sentences as support; reader asked for diagnosis, not procedure
Vertical link: Issues → Assessment (↑)	Assessment jumps to conclusion ("no one controls data," "responsibility unclear") with no causal bridge from the specific problems listed above	Insert intermediate conclusion explaining why those symptoms occurred; connect inconsistent data → lack of ownership, not just state both
Issues section (⊕)	Scope ambiguous — says "For one book" but unclear whether problems are book-specific or systematic across all four	Clarify: are these symptoms common to all books or unique to this one
Same-kind groups (⚠)	Three sections are three different kinds: Production (mechanics) + Issues (symptoms) + Assessment (diagnosis); reader asking "what's wrong?" does not receive grouped reasons	Rebuild around root causes or recommendations, not process steps + problems + conclusion
Closing line	Abdicates with a question instead of proposing action	State what you recommend or what decision you need

Score: 1 / 10 — not structured yet

Top: 0 (topic label, no answer)
Same-kind: 0 (three unrelated categories)
Order: 0 (process–symptoms–diagnosis sequence is unexplainable)
MECE: 1 (problems listed but scope and causes unclear)
Display: 1 (headings visible, logic flow is not)

Three fixes that buy the most

Lead with diagnosis, not process. Move "fragmented process, no one controls data end-to-end, responsibility unclear" to the opening as a one-sentence answer.
Remove or bury the Production section. Mechanics answer a different question. Keep 1–2 lines as context if needed; cut the four-step list.
Add intermediate conclusions. Show why the listed problems happen (e.g., "inconsistent data between periods because multiple re-entries → no single source of truth"). Connect symptoms to root cause before the summary.

Судья

Повторяющийся дефект	без	с Minto
Missing answer is not distinguished from the buried streamlining conclusionправило в навыке есть, модель его не выполнила · Mode 2 — audit, check 1: Top	нет	есть
Incomplete process logic is not separated into graph-generation and later-change failure pointsтакого правила в навыке нет · Mode 2 — audit, check 4: MECE	есть	есть
Suggested first-level structure uses causes, symptoms, constraints, or topics instead of actionsправило в навыке есть, модель его не выполнила · Core loop, §4 Groups	есть	есть
Inline annotation, capped findings, five-axis score, or three highest-value fixes are absentправило в навыке есть, модель его не выполнила · Mode 2 — audit, delivery contract	есть	нет
Unsupported evidence requirements, process claims, or implementation specifics are introducedправило в навыке есть, модель его не выполнила · Mode 2 finding rule and Core loop §8	нет	есть

Решающие вопросы

Did the audit state precisely that the required changes are absent?
без yes
с Minto yes
Did it distinguish a missing answer from a merely buried answer?
без yes
с Minto no
Did it identify streamlining as an undeveloped conclusion?
без yes
с Minto no
Did it name incomplete process logic and unseparated failure points?
без no
с Minto partial
If a replacement structure was suggested, is its key line made of actions?
без no
с Minto no
What false positives or invented details appear?
без introduces an unsupported production pause and does not address the specific failure points
с Minto misstates the reader’s question as diagnosis, marks a situation sentence as the top, and proposes an unsupported causal bridge
Did the output remain diagnosis-only and include the required audit apparatus?
без diagnosis-only, but the required apparatus is absent
с Minto yes, although the annotation is internally inconsistent

Проверка судьи	без	с Minto
Род первого уровня как в эталоне	да	да
Выдуманные факты	нет	нет
Потеря материала без пометки	нет	нет
Режим соблюдён	да	да
Тип порядка назван	нет	нет

Ещё 4 проверки судьи не применяются к режиму audit

Как устроен тест

Claude Haiku 4.5 получила одну и ту же задачу дважды. Контрольный промпт — одна вводная строка и задача. Промпт с Minto — та же задача, перед которой стоит полный текст навыка. Судья читал оба ответа среди 16 ответов под слепыми метками и не знал, в каком из них был навык.

Промпт без Minto 2 424 Б

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

<!-- Faithful paraphrase of Minto (2010), Appendix B, Exhibits B-5 and B-6.
     Original wording for this fixture. -->

# Fixture 03: Period Graph Books

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

## Context

Corporate Financial Analysis recently took over production of four monthly
graph books used in presentations to senior management. Management knows that
errors and inconsistencies have appeared and asked for corrective proposals.

## Before

```text
Subject: Period Graph Books

Since Period 5, Financial Analysis has been responsible for producing four
graph-presentation books previously handled by Corporate Planning. This memo
describes issues that appeared after the transfer.

Production

1. Gather data from divisional forms, internal documents, and telephone calls.
2. Calculate graph points manually or on the corporate computer, including
   rolling values for revenue, cost, and percentages.
3. Transcribe the points to input sheets. Analysts update the current period,
   then Color Graphics re-enters the points into its own database and generates
   the graphs.
4. Check the data and calculations.

Issues

The transfer inserted another person into an already fragmented process.
Divisional staff and three corporate departments participate, and the same
information may be entered into computer systems three times.

For one book, divisional data are loaded into a dedicated corporate program,
calculated, transferred to Color Graphics input sheets, entered again, and
returned for updates. Problems include inconsistent data between periods and
regions, incorrect variable-cost calculations, unexplained changes to points
that were previously correct, and prior-period data that must be re-entered.

Overall assessment

Most problems arise from a cumbersome fragmented process. No one controls the
data from beginning to end, responsibility is unclear, and errors can fall
between stages. The production process badly needs streamlining, both for each
book and for common calculations. The current team cannot manage the required
changes as well as ongoing production.

Could we discuss how best to proceed?
```

## Task

Audit this document against the Minto Pyramid Principle. Diagnose the defects
and explain what must change. Do not rewrite the document.

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

Промпт с Minto пять частей по порядку · 29 997 Б

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 на f4813ce 17 235 Б
===== references/rules.md ===== references/rules.md на f4813ce 6 464 Б
===== references/templates.md ===== references/templates.md на f4813ce 3 742 Б
===== TASK ===== вход кейса целиком
Return only the result, with no explanation of how you produced it.

Вход кейса before.md

Fixture 03: Period Graph Books

Mode: audit
Language: en

Context

Corporate Financial Analysis recently took over production of four monthly
graph books used in presentations to senior management. Management knows that
errors and inconsistencies have appeared and asked for corrective proposals.

Before

Subject: Period Graph Books
                
                Since Period 5, Financial Analysis has been responsible for producing four
                graph-presentation books previously handled by Corporate Planning. This memo
                describes issues that appeared after the transfer.
                
                Production
                
                1. Gather data from divisional forms, internal documents, and telephone calls.
                2. Calculate graph points manually or on the corporate computer, including
                   rolling values for revenue, cost, and percentages.
                3. Transcribe the points to input sheets. Analysts update the current period,
                   then Color Graphics re-enters the points into its own database and generates
                   the graphs.
                4. Check the data and calculations.
                
                Issues
                
                The transfer inserted another person into an already fragmented process.
                Divisional staff and three corporate departments participate, and the same
                information may be entered into computer systems three times.
                
                For one book, divisional data are loaded into a dedicated corporate program,
                calculated, transferred to Color Graphics input sheets, entered again, and
                returned for updates. Problems include inconsistent data between periods and
                regions, incorrect variable-cost calculations, unexplained changes to points
                that were previously correct, and prior-period data that must be re-entered.
                
                Overall assessment
                
                Most problems arise from a cumbersome fragmented process. No one controls the
                data from beginning to end, responsibility is unclear, and errors can fall
                between stages. The production process badly needs streamlining, both for each
                book and for common calculations. The current team cannot manage the required
                changes as well as ongoing production.
                
                Could we discuss how best to proceed?

Task

Audit this document against the Minto Pyramid Principle. Diagnose the defects
and explain what must change. Do not rewrite the document.

Эталон судьи gold.md

Gold 03: Period Graph Books

Source structure: Minto (2010), Appendix B, Exhibits B-5 and B-6.

Book-derived defects

Defect	Severity
The document never states the changes required; it therefore does not answer management's question	Critical
It describes the current process instead of presenting a top and key line	Critical
"The purpose is to describe issues" is a topic, not an answer	Critical
The process description is incomplete and does not isolate each system's failure points	Medium
Problems in graph-point generation and later presenter changes are not clearly separated	Medium

The buried conclusion is that the process needs streamlining, but this is not
turned into an actionable answer.

Expected replacement structure

Reader question: What changes should we make to prevent unreliable graphs?

Top: Remove the opportunities for error across data transfer, graph-point
generation, and final presentation changes.

Key-line kind: changes/actions.

Create a direct data link to the corporate computer.
Create a reliable computerized routine for graph-point generation.
Require presenter changes to be revalidated before use.

Audit outputs should name the missing answer, distinguish it from a conclusion
that is merely buried, annotate only visible evidence, provide no rewritten
memo, and avoid inventing defects in unseen material.

Судья gpt-5.6-sol, effort high · Claude Haiku 4.5, effort low · навык с коммита f4813ce · прогон v1.7.0-final