Ask for a Diff and Rationale, Not a Silent Rewrite

What it is

An editing habit: ask ChatGPT to show what it changed and why — a marked-up list of edits with reasons — rather than returning a wholesale rewrite you have to reverse-engineer.

See what changed and why, so you approve edits instead of inheriting them.

Why it works

A silent rewrite hides its decisions: it may have fixed a real problem, but it may also have flattened a deliberate phrase or dropped a nuance, and a clean new version makes those invisible. A diff with rationale turns editing into review — you see each change, judge it, and keep control of your text instead of inheriting the model's choices blind.

When to use it

Editing writing you care about and know well — your own polished drafts, high-stakes documents, anything where losing a deliberate choice matters.

When not to use it

Rough first drafts you're happy to have wholesale-rewritten, or when you explicitly want a fresh take rather than a review of yours.

Prompt

Edit this, but don't just return a rewrite. List each change as: original → edited, plus a one-line reason. Group them (grammar, clarity, style). I'll decide which to accept. Text: <paste>.

Example

The change list reveals ChatGPT 'tightened' a sentence in a way that dropped a key caveat — you reject that one edit and keep the rest.

Common mistakes

Accepting a clean rewrite without seeing what it silently altered.
Losing deliberate word choices because the changes were invisible.
Asking for reasons but not actually reviewing them before pasting back.