Preprints.org’s Sam Rye defines a preprint as a “draft or in-progress version of a scholarly article.” That sounds like a really nice definition, especially if coupled with the traditional view of the academic publication flow. Consider for example the Publishing Status Ontology. Under green open access, it seems to define a preprint as “the version of the article as first submitted for publication.” Note the the article here, referring apparently to the fact that a preprint is intended for future publication in a peer-reviewed journal.

In my applied academic reality, though, I have been struggling to understand whether the difference stands. The issue is exacerbated by the fact that if a researcher fancies self-publishing something virtually in 2026, multiple options fortuitously present themselves:

• A legit preprint server. ArXiv, medRxiv, ChemRxiv, you name it. These apparently are platforms that intend to host publications intended to be submitted for publication in a peer-reviewed journal. Or so reads the name.
• Various researcher-oriented data repositories, including Zenodo, Dryad, or the Open Science Framework (OSF). These vary in terms of their inclusion criteria but overall seem to operate under the policy of “anything goes” (unless thematically constrained) or even offer the researcher a variety of options to identify with when they submit their document.
  • Within this category, one might observe a class of platforms called registers. These seem to exist to register research documents in some particular, register-specific sense. For example, my interpretation is that a research document published on Zenodo would not be considered “registered” due to the fact that Zenodo does not position itself as a register, despite the fact that metadata for all Zenodo registrations are nevertheless registered with DataCite.
• Non-research-oriented repositories might also be legitimately used. For example, GitHub, GitLab, or even Google Drive have all been used extensively to host public research documents.
• Distributed hosting solutions are also not off the table. These include the InterPlanetary File System (IPFS), Arweave, BitTorrent, and many other similar protocols and services.
• Not least, the researcher may opt to self-host their research document. What that means varies depending on whether they self-host it on their own physical computer, on a rented host, on the said distributed nodes, and so on.

Ultimately, it seems to me that the distinction between these options is often more societal than it is technological. For example, publishing a document redundantly through both a scientific data curator (e.g., DataCite) and through the IPFS makes a lot of sense provenance-wise because while the first avenue offers a trusted timestamping authority, the second avenue offers a strong cryptographic fingerprint of the data while also enabling self-maintenance. At the same time, publishing a protocol through Zenodo while also registering the same document with the OSF would probably be more of a political consideration because technologically, both platforms will register their data through DataCite, but one is U.S. and the other E.U. based, and they also have different external data preservation arrangements; while on the other hand, one or the other could be more or less familiar within a particular scholarly community.

This concludes my journey into the preprint-verse, for now. I hope that next time, I will surely select the right distribution channel for the right publishing status. The traditional, paper-based academic publishing has been there for hundreds of years, and as still-pioneers of the digital it’s important that we get this right.

The Start | The End

---

Written by Pavel Zhelnov on May 19, 2026. Last revised May 19, 2026.