1 / 7
What the public record supports
"Peer reviewed" is treated as a single guarantee. It is not one thing, and different venues publish different parts of the process — which means the phrase alone does not tell a reader what was checked, by how many people, or whether anyone can see the reasoning.
| Property | The question it answers |
|---|---|
| Selectivity | What fraction of submissions are published, and against what bar? |
| Review model | Who reviews, whether identities are known, and what is published afterwards |
| Indexing | Which databases carry the record |
| Access model | Who can read it, and who paid for that |
That a result is true or will replicate; that reviewers agreed; that every concern was resolved; or what the editor privately weighted. Read only the checks and decisions the venue actually makes public.
2 / 7
Three records you can read in full
Nature · seed carriers
Bioinspired self-burying seed carriers. The published 96-page Peer Review File includes reports from three referees across three rounds and point-by-point author responses. The paper acknowledges two reviewers by name; the file does not map names to report numbers.1
Your instructor is a co-author.
eLife · earlier model
Suction attachment in mountain-stream insect larvae. A consolidated decision letter with numbered essential revisions, plus the author response.2
eLife · current model
Pterosaur tail-vane soft tissue, published as a Reviewed Preprint with an eLife assessment in controlled vocabulary.3
All three are bioinspired or organismal work, so the science is legible to this cohort. More importantly, all three publish the reasoning, which means every claim made in class can be checked against the record rather than accepted from a slide.
3 / 7
What each model makes public
| Nature peer review file | eLife, earlier model | eLife, current model | |
|---|---|---|---|
| Referee reports published | The article file includes reports and rebuttals across three rounds | Consolidated into a decision letter | Yes, as public reviews |
| Reviewer identities | Named in the acknowledgement when they consent; others anonymous | Anonymous; editors named | Anonymous unless the reviewer signs |
| Author response published | Yes, point by point | Yes | Yes, when provided |
| Accept / reject decision | Yes, before publication | Yes | Editorial invitation gate, but no post-review accept/reject decision; assessment plus versions |
| Controlled assessment vocabulary | No | No | Yes, for significance and strength of evidence |
This table is a snapshot. eLife changed its model substantially after 2022, and Springer Nature has extended transparent peer review across its titles. Before citing any row, open the journal's current policy page and confirm it.
4 / 7
Inside a referee report: three different lines of critique
The three Nature referees do not offer variations on one shared opinion. They challenge different dimensions of the work, and one report is strongly skeptical.
Referee #1 — technical
Raises specific questions about whether an empirical correction factor is labelled accurately, whether an equation is complete, and whether replicate counts and statistical variation are reported. Each concern points to a checkable manuscript change.1
Referee #2 — a necessity challenge
Questions whether the design provides a significant advance and asks why natural seed awns could not be used instead. The report also supplies literature showing that a three-tailed geometry exists in nature.1
Referee #3 — a venue-fit challenge
Calls the work interesting but asks whether its significance is broad enough for the venue.1
Referee #2's challenge questioned necessity, not merely presentation. The authors conceded the valid comparison, answered with quantities, tested real seed awns, and incorporated the cited natural example into the design rationale. The published sequence documents the response; it does not reveal which element the editor privately weighted.
5 / 7
A referee proposed the published title
Buried in Referee #1's list of minor points:
Referee #1 objected that “self-drilling” overstated what the seed did and proposed the more descriptive title “Autonomous Self-burying Seed Carriers for Aerial Seeding.”1
That is the paper's published title.
- A title is a claim. What does it mean that this paper's central claim was reworded by an outside referee?
- The referee's objection was semantic: the seed is not drilling itself. Find the equivalent objection in your own abstract — a word that asserts more than your evidence does.
- You will get an outside reader in the next 15 minutes. What will you ask them to attack?
One referee, one paper. This shows that a title can arrive from outside the author team, not that titles usually do, and not that accepting a referee's title improves acceptance odds.
6 / 7
Assessment vocabulary: intended order and reader interpretation
Before displaying eLife’s definitions, the instructor gives students the bare significance and evidence-strength terms. Each student ranks them individually and records confidence. This calibration task is completed before applying any term to a scientific abstract.
The pinned eLife assessment describes the contribution as valuable and the methods as compelling. Those are defined terms in eLife’s separate significance and strength-of-evidence vocabularies.3
"Valuable"
A significance term: findings with theoretical or practical implications for a subfield.
"Compelling"
An evidence-strength term: methods, data and analyses more rigorous than the current state of the art.
Hardwicke and colleagues asked 301 individuals holding doctoral or graduate degrees to rate short hypothetical statements on 0–100 scales.6 The implied ordering matched eLife’s intended order for 20% on significance and 15% on strength of evidence. A more explicitly ordinal wording produced much higher agreement.
- Compare each student’s recorded ordering with the intended order. Do not convert the individual study percentage into a prediction about teams.
- What further study would distinguish a problem with this word list from a problem with controlled vocabularies generally?
- Why might important, valuable, and fundamental be hard to order, when high, moderate, and low are not?
Hardwicke et al. tested individual interpretations of isolated hypothetical statements, not team deliberation, editorial prose, reviewer quality, or the overall value of eLife’s model. The assessment is also not a scoring rubric; applying terms to an abstract is a separate classroom exercise with no numerical comparison to that experiment.
7 / 7
Now write one
Write an independent review of another team's Draft 2. Five parts, each tied to text:
- Two-sentence neutral summary of the work as the draft presents it.
- One strength, quoting the text that supports it.
- One material weakness or uncertainty, quoting the text and naming what is missing.
- One specific, feasible request, stating what evidence would resolve it.
- One statement explicitly labeled as inference rather than as something the draft establishes.
"Figure 1 reports one trial. Either report n, mean, and standard deviation for each condition, or restrict the claim from 'reliable' to 'observed in a single run.'"
"The results section needs more rigor and the writing could be clearer."
Reviews address the draft, not the author. Do not speculate about a team's ability, effort, or intent. Do not evaluate who wrote which part.
Name one comment you wrote that the receiving team could act on without asking you a follow-up question. If you cannot find one, rewrite your best comment until you can.