Only accepted work is recordedAboutGuidesContactSign in
CitePepCitePep
Guides/Start here/Writing a correction that gets accepted
Start here · 3 of 414 min read · Beginner · Updated Sep 2026

Writing a correction that gets accepted

An editor deciding on your correction has a few minutes and a page they are responsible for. They accept what they can verify quickly and apply without rewriting. This guide shows how to give them exactly that, and why most declined corrections fail on the same few points.

Exactthe change can be applied as written
Sourceda reader can check it in one click
Narrowit fixes one thing, and only that

1Before you write

Five minutes of checking saves a declined correction. Before proposing anything, confirm three things.

CheckHowOtherwise
The claim is really wrongOpen the page’s cited source, if it has one. The page may be quoting it correctly for a different period or definition.Propose a clarification instead, e.g. adding the period.
It’s a new proposalOpen the page record and look under "Pending".Add a source to the existing proposal instead.
You’re independent of the pageDo you work for, advise or compete with the publisher?Declare it first, or leave this page to someone independent.

2Anatomy of a correction

Every correction has five parts. Here is an accepted one, annotated.

Correction · whennotesfly.com/overthinking-and-ruminationAccepted 9 Aug 2026
1LocationParagraph 2, sentence 1Quote enough of the sentence that the editor can find it without searching.
2Current text"Global mean sea level is rising by about the published rate per year."Copy it exactly, word for word.
3Proposed text"Global mean sea level is rising by about the current rate per year (2006–2018)."Change only what is wrong. Adding the period makes the figure checkable.
4ReasonThe page gives the 1993–2010 average. The current assessed rate for 2006–2018 is the current rate per year.One or two sentences. State the error itself.
5SourceIPCC AR6 WG1, chapter 9, section 9.6.1.1Primary, current, and precise enough to open at the right place.

Notice what the proposal doesn’t do: it leaves the paragraph, commentary and tone as they are. The editor can accept it without touching anything else on the page.

3Five kinds of correction

Most corrections fall into one of five kinds. Each has its own trap.

KindExampleThe trap
Wrong figure3.2 → the current rate per yearUsing a figure from a different period or definition. State both.
Outdated guidanceMetformin start dose 1,000 → 500 mgCiting a summary rather than the current guideline itself.
Misattributed quote"The commute is the killer": no source foundProposing a new attribution you haven’t verified. Suggest removal instead.
Wrong date or name2020 → 2015 (Bloom et al.)Correcting the date but not the citation that depends on it.
Broken or weak sourceDead link → IPCC AR6 WG1Replacing it with a source that says something slightly different.

4Choosing a source

Editors weigh sources the way reviewers do. The closer to the original evidence, the faster the decision.

Strongest
Primary research, official statistics, the current guideline or standard
Good
Systematic reviews, assessment reports (e.g. IPCC), regulator publications
Weak
News articles, secondary summaries, textbooks older than the claim
Not accepted
Other web pages without their own sources, AI-generated summaries, your own opinion
Make the source easy to check
  • Link to the exact page, table or section, not the report’s front page
  • Prefer a free version when one exists
  • For a figure, name the table or figure number

5What the editor sees

Your correction arrives in the publisher’s dashboard looking like this. Everything the editor needs to decide is on one line, plus your track record in the topic.

Publisher dashboard · Incoming work
Correction · 2h agoOverthinking and rumination projections explainedthe published rate per year → the current rate per year (2006–2018)IPCC AR6 WG1, ch. 9 · by Melik Can Sariyer · Verified · 92% accepted · Sea-level authority 78AcceptAsk a questionDecline

Your trust score and topic authority are shown next to your name. The editor decides on the evidence, and a clear, sourced correction from someone with a record in the subject is quicker to accept.

6After you submit

Step 1SubmittedThe publisher is notified.
Step 2Question (optional)The editor may ask you for detail. Reply on the same thread.
Step 3DecisionMedian 2–3 days. You’re notified either way.
Step 4RecordedIf accepted: added to the page log and your profile.

If the editor asks a question

Answer on the same thread, with the specific detail requested. Keep it in one correction so the conversation and decision stay together.

If it’s declined

Declined work is kept privately for integrity; your profile and score reflect accepted work. The editor may give a reason. If you believe the page is still wrong and you have new evidence, you may propose again; if you believe the decision breaches the standards, use the complaints process.

7Common reasons for decline

DECLINED"This section could be clearer."Not a correction: no exact change, no error identified.
DECLINED"The figure is outdated, please update."No proposed text and no source. The editor would have to do the work.
DECLINEDRewrites the whole paragraph to fix one number.Too broad. The editor can’t accept the fix without accepting everything else.
DECLINED"According to Wikipedia, it’s 3.7."Secondary source. Cite what Wikipedia cites.

8Checklist before you submit

Was this guide useful? Yes · NoFind a page to correct