You’re staring at a paywall, the paper costs $42, and three clicks away the same manuscript is sitting free on ResearchGate. Can you cite it? And if yes: which version, which URL, which DOI? These aren’t formalities — they decide whether your thesis survives a citation audit.
This post walks through the ways a scientific paper reaches the public, which of them are citable in a thesis, and how to choose when you have several versions in front of you.
The five ways a paper goes public
Almost every academic paper today exists in several parallel versions. If you don’t know the difference, you cite the wrong one by accident and produce misquotations, because page numbers or wording shift between versions.
- Preprint (author’s original). The manuscript before peer review. Hosted on arXiv, bioRxiv, medRxiv, SSRN. Fully citable, but you must label it a preprint.
- Postprint / Accepted Manuscript (AAM). After peer review, before the publisher’s typesetting. This is the green Open Access route: many publishers let authors deposit this version in their institutional repository 6 to 24 months after publication.
- Version of Record (VoR). The publisher’s final, typeset version with a stable DOI. If you have a choice, always cite this one.
- Gold Open Access. A VoR that’s free from day one because the authors (or their institution) paid an article processing charge. Legally unproblematic.
- Pirated copy (Sci-Hub, Libgen, often ResearchGate). Usually the VoR itself, without the publisher’s permission. The publication is citable; the URL is not — more on that below.
The two most common mistakes in undergrad theses: listing a ResearchGate URL instead of the DOI, or citing a preprint without noticing that the peer-reviewed VoR has since appeared with different numbers.
What does “citable” actually mean?
A source is citable when it is findable, verifiable, and stable. All three conditions have to hold. A blog post with a correct URL satisfies condition one, but condition three only weakly — URLs die. A book with an ISBN satisfies all three. A Sci-Hub link is findable and verifiable, but not stable (the domain changes every few months), which is why it doesn’t belong in a reference list as evidence.
Rule of thumb: if a reader can still locate your source in five years using only the details in your reference list, the source is citable. DOIs, ISBNs, and stable repository URLs (Zenodo, arXiv, institutional repositories) are what give you that stability.
Open Access — gold, green, bronze, diamond
Not everything free online is Open Access. The term splits into four flavours, and each one has consequences for how you cite.
- Gold Open Access. Released by the publisher on day of publication, usually under CC-BY. Cite as you would any classical journal article — DOI, journal, volume, issue, pages.
- Green Open Access. A postprint in a university repository or on a personal page. Citable — but give the DOI of the VoR, not the repository URL, or the reference looks second-hand.
- Bronze Open Access. Released by the publisher but without a clear licence — the paper can slip back behind the paywall at any time. Citable, at the risk of the link dying.
- Diamond Open Access. Free with no APC, funded institutionally. From a citation perspective identical to gold.
The colour matters less than three practical questions: is there a DOI, is the journal listed (DOAJ, JCR), and is the copy in your hands the published version? If yes to all three, you can cite it.
Paywalled sources — and how to open them legally
Citing a paywalled paper is fine — those are usually the standard references in your field. The real question is: how do you legally get the PDF?
- University library + VPN. The fastest route. Almost every university in Europe has bundles with Elsevier, Springer, Wiley — VPN on, paste the DOI into Google Scholar, full text one click away.
- Interlibrary loan / document delivery. If your library doesn’t subscribe, you order the paper through interlibrary loan. Turnaround: 1 to 7 days, cost usually $3 to $8.
- Emailing the authors. A message with subject “Reprint request: [title]” gets a PDF back in roughly 60% of cases within a week. Authors are allowed to send their manuscript for scientific use.
- Unpaywall. A browser extension that checks, for any DOI, whether a legal Open Access version exists (usually the green route in a repository). Estimated coverage: 30 to 50% of paywalled papers.
What you shouldn’t do: write the Sci-Hub URL into your reference list. That’s not a citable URL, and a citation audit will spot it immediately.
Citing preprints correctly
Preprints have become standard in fields like machine learning, physics, and biomedicine — the published version often follows 6 to 12 months later. Two rules:
- Label preprints as preprints. In APA 7, put
[Preprint]in square brackets after the title, plus the DOI-like identifier of the preprint server (arXiv and bioRxiv both issue their own). - Switch to the VoR the moment it appears. Before you submit your thesis, run every preprint reference through Google Scholar again. If the published version now exists, replace the DOI, journal, and pages. Your claims stay the same in 95% of cases — but your citations stay current.
Avoid: citing a preprint URL without marking it as a preprint. Examiners see the arXiv link and immediately know you never checked the peer-reviewed version.
When Open Access and paywall cover the same study
More common than you’d think: authors deposit the preprint on arXiv, the peer-reviewed version runs in parallel, and a year later the VoR appears in Nature. All three versions may carry slightly different numbers, because reviewers forced changes. The rule is uncomfortable but clear: cite the version you actually read. If your text says “0.42” and the VoR has since been corrected to “0.47”, your statement is wrong — not your source.
In practice: right before submission, revisit every central source once more, check whether a newer version exists, and reconcile the numbers. A citation-checking tool like Acurio verifies whether the claims in your text match the claims in the PDFs — and raises a flag whenever a number diverges.
Predatory publishers — the flip side of gold OA
Not every Open Access journal is legitimate. Predatory publishers sell publication slots without real peer review, pocket the APC, and produce papers that pass no gate. For you as the citing author the damage doubles: you cite an unchecked claim as scientific fact, and your thesis loses credibility.
A three-second test before citing any OA piece:
- DOAJ listing? The Directory of Open Access Journals only lists journals with documented peer review.
- Who is on the editorial board? Google two names. Do the people exist? Are they really at the institution listed?
- Journal Impact Factor / Scimago quartile? Not everything, but a strong signal.
If DOAJ is missing and the editorial board looks assembled at random, find an alternative source for the same claim — or drop the reference.
Short version: which version do I cite?
When you have to choose between several versions, use this ranking:
- Version of Record with DOI, regardless of whether it sits behind a paywall or is gold OA.
- Green Open Access postprint in a repository, referenced with the VoR’s DOI.
- Preprint with DOI, clearly labelled as a preprint.
- Everything else (a ResearchGate copy without version info, a Sci-Hub link, an author’s homepage): only cite if no other version exists, and note the fact.
If you consistently capture the VoR’s DOI at the “research” stage rather than the random URL where you found the PDF, you save yourself hours of reconciliation right before submission. Zotero does this automatically when you use “Retrieve Metadata for DOI/ISBN” rather than “Save Page”.