Understanding the Scientific 'First Draft'
In science, speed can be critical. To get findings out quickly, researchers often post their work on public servers before it goes through the lengthy process of formal peer review. This early version of a paper is called a preprint. Think of it as a first
draft shared with the entire world for feedback. The practice became especially common during the COVID-19 pandemic, when the need for rapid information sharing was urgent. However, because these studies haven't been formally checked by independent experts, they have long carried a major caveat: their conclusions are preliminary and could be wrong.
A Surprising Finding on Reliability
A recent analysis, reported in July 2026, has challenged the conventional wisdom about preprints being unreliable. A study led by neuroscientist Ruslan Rust examined approximately 70,000 biomedical preprints posted on the server bioRxiv and compared them to their final, peer-reviewed versions. The findings were striking: in nearly 90% of cases, the central conclusion of the paper either didn't change at all or was only slightly revised. When claims did change, they tended to become more cautious, not more dramatic. The analysis also found that studies that started as preprints were retracted at roughly half the rate of those that didn't, although the authors note this is an observation, not proof of causation.
Why Caution Is Still Key
While the analysis is reassuring, it’s not a green light to treat all preprints as settled fact. For one, the study itself is currently a preprint, meaning it has not yet been peer-reviewed. Furthermore, some scientists have pointed out that the findings could be influenced by selection bias—perhaps researchers only post their most solid work as preprints—or that stretched-thin reviewers are catching fewer major issues. The world of scientific publishing is also grappling with a surge in fraudulent papers from so-called "paper mills" and errors introduced by generative AI, making critical evaluation more important than ever.
Your Checklist: 1. Check the Venue
The first step before sharing a scientific claim is to identify its source. Is the study on a preprint server like arXiv or bioRxiv, or has it been published in a peer-reviewed journal? Preprint servers are clearly marked and serve a valuable purpose, but they are not a stamp of validation. A peer-reviewed journal, on the other hand, indicates the paper has passed a formal quality control process where experts in the field have vetted its methods and conclusions. Knowing the difference is the foundation of responsible science communication.
2. Look for the Final, Published Version
If you encounter a claim based on a preprint, the next step is to see if a final, peer-reviewed version exists. Most preprint servers will link directly to the published paper once it's available. You can also search for the paper's title on databases like Google Scholar or PubMed. The published version is considered the current version of record. If the claims you saw shared on social media were from a two-year-old preprint, the final published paper might tell a very different story.
3. Compare the Core Claims
As the 70,000-preprint analysis shows, claims often become more nuanced after peer review. When a final version exists, compare its main conclusions to those of the preprint. Did the key finding hold up? Was the language softened? Did the authors acknowledge new limitations? Often, the most sensational claims made in early reports get toned down in the final publication. Sharing the updated, more cautious conclusion is always the more responsible choice.
4. Scan for Corrections or Retractions
Science is self-correcting, and even peer-reviewed papers can be updated or withdrawn. Before citing a study, always check for notices of correction or retraction. Journals post these notices on the paper's official page. An "expression of concern" is another red flag, indicating the journal has serious questions about the data or conclusions. In an era where AI-generated errors and fraudulent studies are on the rise, this final check is a crucial step to avoid spreading outdated or discredited information.
