ConductScience · Open peer review
Good questions make
research stronger.
We believe peer review should be open to examination. Specific questions, evidence-backed answers and a visible decision history let readers see how a paper was assessed. Reviewers receive credit for the work they contribute.
Explore open reviews →From question
to decision
- 01
Authors choose public review
Authors authorize a public manuscript version and its review record. An editor checks the permission and manuscript link before opening it. A paper under review has not been accepted by the journal.
- 02
An optional initial assessment prepares the review
With separate author permission, the editor can request a private AI assessment of the manuscript text. ReplicateScience checks extracted quotations and methods details against that text. The editor checks proposed questions and relevant supplements before opening them to reviewers.
- 03
Reviewers address specific questions
Questions identify what needs checking and where to find it. A statistician might examine an analysis; a methods specialist might assess the study design. Anyone with a confirmed account can propose a question. Editors check reviewer expertise and conflicts before validating an answer.
- 04
Evidence and responses stay connected
Each published answer names its contributor and links to a particular paper version. Authors can respond, and the editor records how each question was resolved. A well-supported disagreement can earn credit.
- 05
Two reviewers assess the whole paper
Invited independent reviewers assess significance, interest, claims, design, methods, analysis, interpretation and ethics. Their reports summarize strengths and separate major from minor comments before giving an overall recommendation. Short answers alone do not satisfy this requirement.
- 06
The editor records a decision
All blocking questions must be resolved and two independent full reviews completed. The editor weighs the evidence, explains the decision and addresses any disagreement. Revised manuscripts need assessments of the new version.
Credit for useful work
An editor-validated answer earns one contribution point. A validated full-paper review earns five. The same question or full review can earn credit once per reviewer and manuscript version. Reproducing a result and identifying a problem both count when supported by evidence.
Points recognize contributions, not agreement, popularity or how quickly someone responds. Authors and handling editors do not earn reviewer points on their own papers. The leaderboard ranks completed whole-paper reviews, with answers shown separately. Rankings are optional; reviewers retain attribution for their published contributions.
See the reviewer leaderboard →A whole-paper assessment
Is the question important? Who would find the result interesting? Does the evidence support the claim? Reviewers explain these judgments, identify strengths and separate major issues that could change a conclusion from minor improvements to clarity or detail. Significance depends on the journal’s scope; careful replications and negative results can make valuable contributions.
Our framework adapts PLOS Biology’s reviewer guidance and the PLOS guide to writing a review. Automated checks supply evidence about specific items. Passing them does not establish scientific validity or replace a human assessment.
Open Review in your account →More people can help, in parallel
A specialist can answer a focused question without taking on the entire manuscript. Several questions can be worked on at once while the full-paper reviewers examine the research as a whole. We will measure how this affects review time and publish results when available.
A clear public record
Questions and answers become public after editorial checks. Reviewers agree to named publication of their contributions. Editors can return an answer for revision or revoke credit with a reason. Concerns about a decision can be raised through the editorial office.
Confidential review stays confidential
This process applies only to new records expressly authorized for open review. Existing private submissions, reports and discussions remain confidential. Opening a review does not publish a final journal article or confer PubMed or MEDLINE indexing.
People remain responsible
Reviewers must disclose any AI assistance and check their evidence. Unpublished material must not be uploaded to tools that cannot protect its confidentiality without the authors’ permission. Software scores, leaderboard rank and the number of comments do not decide whether a paper is accepted.
Open review policy 1.1.0. The journal peer review policy continues to govern confidential review and the final publication decision. Public posting under this process is limited to the specific material that authors and contributors authorize.