Sort the Feedback

---
name: sort-the-feedback
description: "Organize feedback from critique partners, workshops, editors, beta readers, or AI into clear revision decisions. Use when a writer has multiple comments, conflicting recommendations, generic notes, or an overwhelming feedback document and needs to separate evidence from taste, recover the observation beneath a prescription, find meaningful convergence and disagreement, prioritize by leverage, or build a revision plan without surrendering authorial judgment."
---

# Sort the Feedback

Treat feedback as evidence to evaluate, not votes to obey. Recover what each reader experienced, distinguish signal from noise, preserve disagreement, and turn the useful material into a revision sequence the writer can consciously accept, test, question, or decline.

## Gather the decision context

Use:

- the feedback, with source labels preserved when possible;
- the draft or relevant excerpts, when available;
- the writer's intention and current concern;
- the draft stage;
- any decisions already made or areas currently out of scope.

If the draft is unavailable, sort the feedback on its own and clearly state that textual claims cannot be verified. Do not pretend a reader's quoted description is the passage itself.

## Normalize without erasing difference

Rewrite each comment into a short faithful note while retaining its source. Do not make vague feedback sound more precise than it was. Split compound comments when they contain different observations or recommendations.

For prescriptions, recover the experience underneath them:

- "Cut chapter two" may contain "I did not understand what changed during chapter two."
- "Make her more sympathetic" may contain "I stopped trusting her after this action."
- "Add more description" may contain "I could not orient myself in the room."

Keep the prescription and the recovered observation separate. The observation may be useful even when the proposed fix is wrong.

## Classify each note

Apply one or more labels:

- **Grounded observation:** Points to identifiable textual evidence.
- **Reader experience:** Reports attention, confusion, trust, expectation, or emotion.
- **Interpretation:** Infers meaning, motivation, or intention.
- **Taste:** Reflects a reader's aesthetic or genre preference.
- **Prescription:** Tells the writer what change to make.
- **Generic:** Could have been written without reading this draft.
- **Unverifiable:** May be accurate, but cannot be checked against the material supplied.

Do not dismiss reader experience merely because it is subjective. Subjective experience becomes more useful when its location and cause are clear.

## Build issue clusters

Group notes by the underlying problem rather than repeated vocabulary. Useful clusters may involve structure, scene function, character, pacing, orientation, voice, continuity, ending, line-level clarity, factual accuracy, or audience expectations.

For every cluster, determine:

1. What readers actually experienced.
2. Which passages or patterns prompted it.
3. Whether the readers agree on the observation, the diagnosis, the prescription, or only the topic.
4. How strongly the issue affects the writer's central intention.

Do not call two comments agreement merely because both mention pacing. Do not count feedback mechanically; three derivative AI notes are not necessarily stronger than one precise human observation.

## Interpret convergence and conflict

- **Convergence:** Independent readers notice the same specific effect. Treat this as strong evidence worth investigating.
- **Conflicting prescriptions:** Readers share an experience but propose different fixes. Keep the shared observation and return the fix to the writer.
- **Conflicting experiences:** Readers respond differently to an intentional choice. Frame an audience or aesthetic decision rather than a defect.
- **Lone precise note:** One reader identifies a well-grounded issue others missed. Preserve it for testing.
- **Lone generic note:** Set it aside unless the writer can connect it to evidence.

## Prioritize by leverage

Rank issues using:

1. impact on the work's governing intention;
2. structural reach;
3. quality of evidence;
4. recurrence across independent readers;
5. cost of postponing the decision;
6. fit with the current draft stage.

Resolve structural and scene-level decisions before polishing sentences that may disappear. Do not turn every valid note into a task.

## Deliver a Feedback Decision Sheet

Start with a table:

| Issue cluster | Sources | Shared observation | Evidence | Feedback type | Writer's decision | Priority |
| --- | --- | --- | --- | --- | --- | --- |

Then provide:

### Take into revision now

List two to four high-leverage issues supported by evidence.

### Test before deciding

List contested or uncertain issues and propose a small experiment for each. Do not rewrite the work.

### Set aside

Identify generic, out-of-scope, mistimed, or purely preferential notes and explain why they need not direct the current pass.

### Decisions only the writer can make

Name the conflicts involving audience, intention, ethics, genre promise, or acceptable difficulty.

### Revision order

Turn accepted issues into a short sequence of passes. Keep diagnosis separate from prescription and stop before revising the draft for the writer.