Reading a Rejection: What Is Signal and What Is Noise

Rejection is the normal outcome of submission. Acceptance rates at selective venues mean that most submissions from most people, including excellent researchers, are rejected.

Knowing that does not help much on the day. What helps is a method for extracting the information, because the reports vary enormously in quality and some of what they contain is not about your paper. For a concrete operational example of how a workplace tool approaches a related coordination problem, see learn more.

First: wait

Do not read the reports carefully on the day they arrive. Skim for the decision, then close it.

Read properly after two or three days. The content will be the same and your reading of it will be different. Almost every regretted response to a rejection was written within a few hours of receiving it.

Separating the types of comment

Reports contain several different things, and treating them alike is the error.

A substantive flaw. The design does not support the claim, an alternative explanation is unaddressed, the data cannot bear the weight. This is signal. It will recur at the next journal.

A framing problem. The contribution is unclear, the paper appears to answer a different question than it does, the abstract oversells. Signal, and usually fixable without new work — and frequently the cheapest large improvement available.

A fit problem. Good paper, wrong venue. Not about quality.

A preference. The reviewer would have done it differently, prefers another method, or wanted a different paper. Weak signal.

A misreading. The reviewer misunderstood something. This is still signal, because if a trained reader misunderstood, the exposition failed. The paper needs to be clearer even though the objection is wrong.

Noise. Contradictory reports, obvious skimming, comments that could have been written without reading it.

The test that matters

Did more than one reviewer raise it?

If two independent readers hit the same thing, it is real, whatever you think of their reasoning. That is your revision priority regardless of anything else in the reports.

And the second test: does the comment recur across submissions? A concern raised at two different journals is a property of the paper, not of a reviewer.

Keep the reports. A rejection file from three venues, read together, tells you what to fix far more reliably than any single report.

What the editor's letter tells you

More than the reviews, and it is frequently skimmed.

A form rejection without reviews is usually about fit or priority, not quality. It says almost nothing about the paper.

A rejection with reviews and a short letter means the editor followed the reviewers.

A rejection with an editor's own reasoning is worth reading closely. It tells you what the editor thought the paper needed, which is the most reliable signal available.

"We would consider a substantially revised version" is not a rejection. It is an invitation with a low probability, and it is regularly misread as a refusal. If it is ambiguous, ask — editors answer this question.

Deciding what to do

Fix the substantive problems before resubmitting anywhere. The next venue's reviewers will find them too.

Fix the framing problems even if you disagree, because a paper that is repeatedly misread has a writing problem.

Consider the fit honestly. A paper rejected three times from top venues on contribution grounds may be a good paper for a different journal rather than a bad paper.

Do not resubmit unchanged. It happens, it wastes reviewer time, and in fields with a small reviewer pool the same person may see it again.

Do not immediately submit to the next journal down. The reflex is understandable; the result is a paper that never improves and a reputation for volume over care.

When the report is unfair

It happens. Reviewers are hurried, sometimes wrong, occasionally acting in bad faith.

A report that is unfair is still information about how the paper reads.

You can appeal, and it rarely succeeds. It is worth it when a review contains a clear factual error central to the decision — that the method is invalid when it is standard, that a result is absent when it is on page eleven.

How: a short, specific, unemotional note to the editor identifying the error and where it is. Not a defence of the whole paper.

Do not appeal because you disagree with the assessment. That is the reviewer's job.

And do not respond to the reviewer, even where the process would allow it. Nothing good comes from it.

The part that is not about the paper

Rejection is frequent and it does not distinguish good researchers from bad ones. Selective venues reject most of what they receive, including work that goes on to be influential.

The distribution of outcomes is wide and partly random. The same paper sent to the same journal with different reviewers gets different decisions, and studies of review agreement consistently find it is lower than people assume.

This is not a reason to dismiss reviews. It is a reason not to treat a single decision as a verdict.

And it is worth knowing early in a career, because the private experience of rejection is that everyone else is being accepted. They are not; they are not talking about it.

Practical habits

Keep every report, organised by paper.

Note which comments recurred, across reviewers and across venues.

Give yourself a fixed period to be annoyed — a day — and then work.

Talk to a co-author or colleague before deciding what to do. The first reading is unreliable and someone outside it will separate signal from noise better than you can that week.

And send the paper out again. The most common reason a paper is never published is that it was never resubmitted.

The short version

Wait two days before reading properly.

Anything two reviewers raised is real, whatever you think of the reasoning.

A misreading is still a signal — the exposition failed even though the objection is wrong.

Read the editor's letter closely, and check whether it is actually an invitation to resubmit.

Rejection is the normal outcome, the variance is wider than people admit, and the papers that never appear are mostly the ones that were never sent out again.

For an independent authoritative reference, consult COPE guidance for publication-ethics guidance for editors and reviewers.