Extended definition
Open science is the umbrella concept that gathers the practices aimed at making the research process and its outputs transparent, accessible, and reusable at every stage, not only in the final article. Vicente-Saez and Martinez-Fuentes (2018), in a systematic review proposing an integrated definition, summarize the concept as knowledge that is transparent and accessible, shared and developed collaboratively. Under this umbrella fall open access to publication, open data under the FAIR principles, study preregistration, open source, open peer review, and reproducibility. What unites these practices is the idea that science gains credibility and efficiency when its decisions and materials are exposed to scrutiny. Nosek and colleagues (2015) show, with the TOP guidelines, that this ideal is realized only when incentives change: since the reward system rewards novelty and not transparency, it is the policies of journals and funders that convert the stated value into effective practice.
When it applies
Open science applies throughout the research cycle, from design to dissemination. It applies at the start, with preregistration of the analysis plan, and during collection, with the data management plan and the opening of data. It applies at publication, through open access and the sharing of code and materials, and at evaluation, through open review. Munafò and colleagues (2017), in the manifesto for reproducible science, organize these practices into concrete measures on methods, reporting, reproducibility, evaluation, and incentives. It applies to institutional and funding policy, where openness mandates change the default behavior. It also applies to collaboration and equity of access, by lowering the barriers between those who produce and those who use knowledge, inside and outside academia.
When it does not apply
Open science does not apply as the indiscriminate opening of everything: sensitive data, under ethical or legal protection, require controlled access, and the correct formula is as open as possible, as closed as necessary. It does not apply as an automatic label of quality: an open study can be methodologically weak, and opening the materials does not fix a bad design. It does not apply without attention to cost and equity: open-publication fees can shift the barrier from the reader to the author, excluding those with fewer resources. It does not apply as a one-time event, but as a set of practices that need infrastructure, training, and incentives to be sustained; without these, as Nosek and colleagues (2015) warn, the value stays in the discourse. And it does not apply uniformly across fields, whose sharing norms and sensitivity vary.
Applications by field
- All disciplines: an umbrella concept that gathers open access, FAIR data, preregistration, open source, and reproducibility.
- Biomedicine: trial registration, data sharing, and open review as responses to the credibility crisis.
- Computational sciences: open, versioned code and data, linking openness and reproducibility.
- Science policy: funder mandates and editorial guidelines that convert the ideal into practice.
Common pitfalls
The first pitfall is confusing open science with total openness, ignoring that sensitive data call for controlled access. The second is treating the open label as a quality seal, when opening materials does not correct a flawed method. The third is ignoring equity: open publication can shift the cost to the author and exclude those with fewer resources. The fourth is adopting isolated practices without infrastructure and incentives, leaving the commitment on paper. The fifth is applying a single standard to all fields, without recognizing that sharing norms and data sensitivity vary across disciplines.