You have a preprint PDF in one folder, a published article in another, and a citation in your draft that points to neither clearly. Before you reuse a number, a method detail, or a reference, you need to know which version you are actually reading and what changed between them. The short answer: recheck the identifier and destination, the bibliographic fields, the claim-bearing text, and the citation record itself. A DOI resolves to a registered destination, but the fields and the source text still need to be checked for identity and claim support.
This guide is for students and early-career researchers finishing a course paper, thesis chapter, or first manuscript, and for anyone starting the next project after a break. It walks through the decisions, the artifacts to keep, and how to know the recheck is finished.
Why the version question surfaces now
Version confusion rarely appears at the moment you read a paper. It appears later: when a supervisor asks where a figure came from, when a reference manager exports a citation that does not match the published record, or when you return to a project in January and cannot tell whether the PDF you saved is the preprint or the final article.
Two facts make this a real task rather than a formality. First, a DOI can be resolved through doi.org to the current destination registered for that identifier, which helps you locate a work but does not by itself confirm that the bibliographic fields or the claims match what you have. Second, Crossref exposes publisher-deposited metadata through its REST API, including titles, authors, venues, dates, identifiers, licenses, funding, ORCID and ROR identifiers, abstracts when deposited, and post-publication updates from trusted sources. That metadata is a strong cross-check, but it is deposited by publishers and may not capture every change you care about in the body text.
So the recheck has two layers: the record layer (identifier, metadata, citation) and the text layer (what the paper actually says).
Diagnose your starting state before you compare versions
Before choosing a workflow, sort out what you already have. Most readers fall into one of a few states:
- Single PDF, unknown version. You downloaded a file and did not note whether it was a preprint or the published article.
- Two files, unclear relationship. You have both a preprint and a published version but have not compared them.
- Citation-only. Your reference list points to a work you have not opened in full.
- Stale project. You are returning after a break and your saved papers, notes, and citations may have drifted.
Each state needs a different first move. The single-PDF case needs identification. The two-file case needs a diff. The citation-only case needs retrieval and verification. The stale-project case needs a sweep across everything you plan to reuse.
A practical way to begin is to gather the works you plan to cite into one place. Saving or uploading papers and organizing them in the WisPaper library gives you a single collection to audit rather than scattered downloads.
What must you recheck on the identifier and destination?
Start with the identifier. A DOI resolves through doi.org to the current destination registered for that identifier. That tells you where the identifier currently points, which is useful for locating the work, but resolution alone does not confirm identity or claim support.
Recheck these items on the identifier layer:
- The DOI itself. Confirm the string matches the work you intend to cite, not a similar-titled paper.
- The destination. Follow the DOI and confirm the landing page corresponds to the same work.
- The version label. Note whether the destination presents a preprint or a published article.
- The metadata fields. Compare title, authors, venue, and date against your draft.
For the metadata comparison, Crossref's deposited fields are a useful reference point because they include titles, authors, venues, dates, identifiers, licenses, funding, ORCID and ROR identifiers, and abstracts when deposited. If your draft's author list or venue differs from the deposited record, that is a signal to look closer.
What must you recheck in the text?
Metadata can match while the text has changed. The text layer is where most substantive errors hide.
Recheck these in the body:
- Numbers and results. Any figure, percentage, or effect size you quote should come from the version you cite.
- Method details. Sample sizes, procedures, and analysis choices can differ between versions.
- Claims and framing. A conclusion may be softened, sharpened, or qualified in the published version.
- Tables and figures. Confirm the table or figure you reference is the one in your cited version.
- Limitations. Check whether limitations were added or revised.
When you have both a preprint and a published version, read them side by side for the specific passages you rely on rather than reading both cover to cover. Your draft tells you which claims matter.
Where does each version fail you?
Comparing the two options on the same task — supporting a claim in your draft — makes the trade-offs concrete.
| Criterion | Preprint | Version of record |
|---|---|---|
| Locating the work | DOI may resolve, but confirm the destination | DOI resolves to the registered destination; still confirm fields |
| Bibliographic accuracy | May lack final venue and date | Publisher-deposited metadata available via Crossref |
| Claim stability | Text may change before publication | Represents the published text you are citing |
| Post-publication updates | Not typically reflected | Crossref can carry updates from trusted sources |
| Citation record | May not match final citation | Matches the published record you will cite |
Where the preprint fails: it may not carry the final venue, date, or pagination, and its text may differ from what eventually appears. Where the version of record fails: it still requires you to verify that the deposited metadata matches your draft and that the specific claim you cite is actually supported in the text. Neither version removes the need to check the claim against the source.
A reasonable rule: cite the version of record when it exists and you can access it, and use the preprint to understand the work's development or when the published version is unavailable. Either way, record which version you used.
A combined workflow for a real project
Here is the workflow applied to a common situation: you are finishing a thesis chapter over winter break and returning in January to start the next project.
Step 1 — Collect. Gather every work you plan to cite into one library. The WisPaper search interface lets you find academic papers, and you can add results to your library for a single audit point.
Step 2 — Identify versions. For each work, note whether you hold a preprint, a version of record, or both. Resolve the DOI and record the destination.
Step 3 — Compare metadata. Check title, authors, venue, and date against the deposited record. Flag mismatches.
Step 4 — Compare text. For each claim you rely on, confirm the passage in your cited version.
Step 5 — Verify citations. Screen your BibTeX against real academic databases so that hallucinated or mismatched references are flagged before submission. TrueCite is built for this BibTeX citation screening step.
Step 6 — Ask targeted questions. When a paper's details are unclear, ask questions based on papers uploaded to My Library or added from search results using Library QA. This keeps the answers grounded in the papers you have collected.
Step 7 — Organize the next project. Create a workspace for the new project in Research Projects so the verified papers and notes carry forward.
For longer tasks — a literature review, inspiration gathering, or reproduction planning — the Scholar Agent area supports those use cases. If your next project needs reproduction planning from an uploaded PDF, PaperClaw is designed for that.
How do you know the recheck is ready?
You are ready when four conditions hold:
- Every cited work has a recorded version. Your notes say preprint or version of record, with the DOI and destination.
- Metadata matches. Title, authors, venue, and date in your draft agree with the deposited record, or you have documented why they differ.
- Claims are confirmed in the cited version. Each number and conclusion you rely on appears in the version you cite.
- Citations are screened. Your BibTeX has been checked against real academic databases.
If any condition fails, you have a specific next action rather than a vague worry.
A January project-start example
Suppose you return in January to a project you paused in December. Your library holds a mix of preprints and published articles, and your draft cites several of them.
You open your library and sort the works into three groups: preprint only, version of record only, and both. For the "both" group, you compare the specific passages your draft relies on. For the "preprint only" group, you check whether a published version now exists. For the "version of record only" group, you confirm the metadata matches your citations.
Then you screen your BibTeX with TrueCite and fix any flagged references. Finally, you create a new project workspace for the January work so the verified set carries forward. The recheck is done when your notes show a version for every citation and your claims are confirmed against those versions.
When to use search tools during the recheck
Search tools help you locate works and alternative versions, and they pair with a source check on the version you actually cite. Google Scholar documents searching by author and title, date sorting and date limits, Cited by and Related articles, citation export, library access links, email alerts, and ways to locate alternative versions of a work. Those features are useful for finding a published version of a preprint or an alternative version of a work you already have.
Use search to answer specific questions: Does a published version exist? Is there an updated version? What else cites this work? Then return to the identifier and text checks.




