Open science

no

Open science

Open science is a global movement and set of practices that aims to make research **transparent, accessible, reusable, and equitable** across the full research lifecycle—from idea and study design through data, code, methods, peer review, dissemination, evaluation, and education. As a philosophy, **Open science** emphasises that publicly funded knowledge should be a public good. As a toolkit, it promotes open access to publications, open research data and software, transparent methods and registered analysis plans, open peer review, interoperable infrastructures, and inclusive participation (e.g., citizen science). Because the focus keyword Open science is used by funders, universities, and publishers worldwide, its scope spans policy, infrastructure, culture change, and day-to-day practices in labs, libraries, and communities.

Open science
Also called	Open research; open scholarship; open knowledge (in the research context)
Core aims	Transparency • Accessibility • Reuse • Reproducibility • Equity • Participation
Key components	Open access (OA) • Open data • Open source/software • Open methods • Registered reports • Open peer review • Open educational resources • Citizen science
Guiding principles	FAIR data • CARE (Indigenous data) • Open by default (with exceptions) • Responsible metrics • Community governance
Major policies/institutions	UNESCO (2021) • cOAlition S/Plan S • NIH Data Management & Sharing • EOSC • OECD
Common infrastructures	Preprint servers • Repositories (data, code, publications) • PID graph (DOI, ORCID, ROR) • Registries (clinical trials, protocols)

Definitions, scope, and principles

Open science means that research materials, processes, and outputs are shared **as open as possible and as closed as necessary**, balancing transparency with legitimate constraints (privacy, security, IP, and Indigenous data sovereignty). UNESCO’s intergovernmental definition frames Open science as a set of values and practices that increase scientific quality and trust by making research methods and outputs openly available, enabling reproducibility and collaboration at scale.

Widely adopted anchoring frameworks include:

• **FAIR** principles (Findable, Accessible, Interoperable, Reusable) for data and metadata, emphasising machine-actionability.
• **CARE** principles (Collective Benefit, Authority to Control, Responsibility, Ethics) guiding Indigenous data governance and equitable benefit sharing.
• **TOP** Guidelines (Transparency and Openness Promotion) for journals and institutions to set policy levels on materials, data, and code sharing and on preregistration.

Historical development

Precedents and early open access declarations

The ethos of sharing methods and replicating results dates to early modern science; printing and learned societies enabled circulation. In the digital era, landmark open access statements—the **Budapest Open Access Initiative** (2002), **Bethesda Statement** (2003), and **Berlin Declaration** (2003)—articulated the case for free online access to research literature and reuse rights via open licenses.

Reproducibility crisis and methodological reform

From the mid-2000s, large-scale evaluations highlighted weaknesses in research credibility and reproducibility across domains, catalysing reforms that became central to **Open science**: preregistration, data and code sharing, replication projects, and improved reporting standards.

Policy mainstreaming and infrastructures

In the late 2010s and 2020s, funders and governments mainstreamed **Open science** through policy: cOAlition S launched **Plan S** for immediate open access to funded research; the NIH mandated data management and sharing; the EU developed the **European Open Science Cloud (EOSC)**; and persistent identifier (PID) infrastructures (DOI, ORCID, ROR, DataCite/Crossref) stitched together the research graph.

Components and practices

Open science operates across multiple, mutually reinforcing pillars:

Open access to publications

Open access (OA) makes articles free to read online, often with reuse rights via **Creative Commons** licenses. Models include **Gold OA** (publisher hosts open version), **Green OA** (self-archived in repositories), **Diamond/Platinum OA** (no author-facing fees), and **Transformative agreements** shifting subscription spending to OA publishing. Plan S requires immediate OA with a CC BY (or equivalent) license for funded work.

Open data and materials

Sharing well-described data, stimuli, and materials enables reuse and verification. FAIR-aligned repositories (e.g., Zenodo, Dryad, Figshare, institutional repositories) issue DOIs, manage metadata, and support embargoes or controlled access when necessary (e.g., for sensitive human data). Data management plans, standard vocabularies, and machine-readable licenses support reuse.

Open source software and computational reproducibility

Code sharing (e.g., on GitHub/GitLab) with open source licenses (MIT, Apache-2.0, GPL) and environments (containers, notebooks) allows transparent computation. Workflow capture (e.g., RO-Crates, workflow languages), continuous integration, and archival in long-term repositories (with DOIs) are best practice.

Preregistration, registered reports, and protocols

• • Preregistration** publicly time-stamps hypotheses and analysis plans to reduce p-hacking and HARKing; **registered reports** undergo peer review *before* data collection, with in-principle acceptance contingent on methodological rigour, not results. Protocol repositories (e.g., OSF Registries, ClinicalTrials.gov, PROSPERO) increase transparency and reduce duplication.

Open peer review and open evaluation

Open peer review encompasses models where reports, identities, or version histories are public; preprint commenting and overlay journals extend evaluation beyond traditional venues. Responsible metrics (e.g., DORA) encourage qualitative assessment and field-appropriate indicators over journal impact factors.

Citizen science and community engagement

Citizen and community science involve volunteers, practitioners, and communities as contributors and co-researchers in problem selection, data collection, interpretation, and dissemination. Co-production approaches emphasise mutual benefit and local knowledge.

Open educational resources (OER) and training

OER—freely available teaching and learning materials with open licenses—support capacity building in **Open science** methods (data stewardship, reproducible analysis, ethics, and governance).

Infrastructure and the open research graph

Open science relies on interconnected infrastructures:

• **Persistent identifiers (PIDs)** for entities and relationships: DOIs for publications/data/software (Crossref/DataCite), **ORCID** for researchers, **ROR** for organisations, **Grant IDs**, and **RAiD** for projects. PID graphs enable attribution, discovery, and automated compliance tracking.
• **Repositories and registries**: domain general (Zenodo, Figshare, OSF), domain specific (GenBank, PANGAEA, ICPSR), and institutional platforms; **DOAJ** indexes open journals, **OpenDOAR** indexes repositories, and **re3data** catalogues research data repositories.
• **Preprint servers**: arXiv, bioRxiv, medRxiv, SocArXiv, and others accelerate dissemination and enable open peer commentary.
• **Interoperability standards**: open APIs, OAI-PMH, JATS XML for articles, schema.org and DataCite metadata schemas, and open citation formats (e.g., CRediT for contributorship).

Ethics, equity, and inclusion

While openness can democratise knowledge, it may also exacerbate inequities if not designed with fairness:

• **Cost barriers**: author-facing article processing charges (APCs) can exclude researchers from low- and middle-income countries and unfunded fields; **diamond OA** and community-owned infrastructures mitigate paywalls on both readers and authors.
• **Privacy and safety**: human data require governance, de-identification, and controlled access. Responsible openness weighs benefits against re-identification risks and stigma.
• **Indigenous data sovereignty**: CARE principles recognise community authority over data derived from Indigenous Peoples and emphasise benefit sharing, ethics, and governance.
• **Attribution and credit**: contributor role taxonomies (CRediT), citation of data and software, and open peer review records recognise diverse contributions beyond authorship.
• **Accessibility**: open formats (machine-readable, screen-reader friendly), multilingual abstracts, and community review lower barriers to use.

Economics and governance

Open science is supported by a mix of public funding, institutional investment, community governance, and market services:

• **Business models**: subscription → transformative agreements; Gold/Hybrid OA; **Diamond OA** (no fees); platform cooperatives; overlay journals; institutional repositories.
• **Incentives and assessment**: responsible metrics (DORA, Leiden Manifesto) and narrative CVs shift evaluation towards contributions to **Open science** (data, code, peer review, replication).
• **Community ownership**: academy- or society-owned journals and repositories, open governance of standards bodies, and library-led consortia aim to align infrastructure with scholarly values.

Quality, credibility, and reproducibility

Open practices are associated with improved transparency and error detection. Shared data and code allow others to verify analyses; preregistration distinguishes confirmatory from exploratory work; registered reports penalise questionable research practices (QRPs). Meta-research finds higher credibility where open practices are normative, though uptake remains uneven across disciplines.

Disciplinary perspectives

• **Life and medical sciences**: clinical trial registration and data sharing; genomic and imaging repositories; preprints accelerate dissemination (e.g., pandemic response).
• **Physics and mathematics**: arXiv culture; open source computational tools; community codes of conduct.
• **Social sciences and humanities**: qualitative data governance, embargoes, and consent; open monographs; community archives.
• **Computer science**: code/data/model release conventions; reproducible ML (model cards, data sheets).
• **Environmental sciences**: community observatories; open geospatial data; policy interfaces.

Education and capacity building

Curricula for **Open science** include data management, licensing, reproducible analysis, version control, and ethical/legal frameworks. Community training (Carpentries, The Turing Way) and institutional support (libraries, data stewards, research software engineers) build capacity and sustain practice.

Policy landscape

Policy body	Key instrument	Salient features
UNESCO	Recommendation on Open Science (2021)	Global policy framing; inclusive and equitable focus; open infrastructures
cOAlition S	Plan S (2018→)	Immediate OA with CC BY; repository routes; rights retention; transparent pricing
NIH (US)	Data Management & Sharing Policy (2023)	Data management plans; data sharing by default with justified exceptions
EU	EOSC; Open Data Directive	FAIR data; public sector information reuse; research infrastructure interoperability
OECD	Open Science Policy Recommendations	National strategies; incentives; skills; cross-border collaboration

Implementation: a practical workflow

Research phase	Open science actions	Tools/examples
Idea & design	Pre-registration; protocol publication; ethics & data governance plans	OSF Registries; ClinicalTrials.gov; PROSPERO; institutional RECs
Data collection	Consent for sharing; standardised metadata; PID assignment	Consent templates; DataCite schema; ORCID/ROR
Analysis	Version control; literate programming; containers; notebooks	Git/GitHub; R Markdown/Jupyter; Docker/Singularity
Dissemination	Preprints; OA journals; data/code deposition; machine-readable licenses	arXiv/bioRxiv/medRxiv; Zenodo/Dryad/Figshare; CC BY/MIT/Apache-2.0
Peer review & evaluation	Open peer review; registered reports; responsible metrics	Journal policies (TOP/DORA); review recognition (Publons/ORCID)
Teaching & outreach	OER; citizen science; multilingual summaries	The Carpentries; citizen-science platforms; lay abstracts

Critiques, risks, and mitigations

• **APC inequities & consolidation**: High fees and market concentration shift paywalls from readers to authors; mitigation includes diamond OA, funder/institutional agreements, and community-owned platforms.
• **Predatory publishing**: Low-quality outlets exploit APC models; responses include whitelists (DOAJ), journal transparency criteria, and local mentoring.
• **Sensitive data**: Open human data can risk re-identification; solutions include controlled-access repositories, differential privacy, and tiered consent.
• **Misinterpretation of preprints**: Media and policy uptake before peer review may mislead; clear labelling and rapid, open review help.
• **Administrative burden**: Compliance without support can be performative; investment in data stewards, research software engineers (RSEs), and automation reduces friction.

Future directions

• **Living science**: continuous updating of preprints, datasets, and guidelines; versioned knowledge objects and open knowledge graphs.
• **Machine-actionable openness**: FAIR-plus practices; executable papers; interoperable provenance (PROV-O, RO-Crate).
• **Global equity**: multilingualism, local repositories, regional networks, and funding for diamond OA; stronger alignment with CARE and community governance.
• **Reward reform**: embedding open contributions in hiring/promotion; narrative CVs; recognition for replication, curation, and review.
• **Open hardware**: community-validated designs for lab equipment, sensors, and field devices to reduce costs and expand access.
• **Policy coherence**: aligning research security, privacy, and IP with open goals through risk-based, proportionate controls.

Representative timeline

Year	Milestone	Significance
2002	Budapest Open Access Initiative	Sets out OA for literature with reuse rights
2003	Berlin Declaration; Bethesda Statement	Consolidates OA principles and adoption
2005	Ioannidis on false findings	Spurs metascience and reform
2012–2015	TOP Guidelines; Replication projects	Institutionalised transparency norms
2018	Plan S announced	Funders mandate immediate OA
2021	UNESCO Recommendation on Open Science	Global, intergovernmental commitment
2023 →	NIH DMS policy; living reviews/guidelines scale up	Data sharing by default; continuous synthesis

Glossary

**APC**

Article processing charge paid to publish OA in some journals.

**Diamond OA**

Open access model with no fees for readers or authors.

**DOI**

Digital Object Identifier for persistent citation and linking.

**FAIR**

Principles for data/metadata to be findable, accessible, interoperable, reusable.

**CARE**

Principles for Indigenous data governance.

**Registered report**

Journal format with peer review before data collection.

**Repository**

Platform that preserves and provides access to publications/data/code.

**Responsible metrics**

Assessment practices that avoid misuse of journal-level indicators.

**Transformative agreement**

Contract that repurposes subscription spend to enable OA publishing.

See also

• Open access
• Open data
• Open-source software
• Reproducibility
• Citizen science
• Creative Commons license
• Data management
• Preprint
• Registered report
• Research data management

Further reading

• Open Science: One Term, Five Schools of Thought, Springer, 2014
• Open Access, MIT Press, 2012
• Data sharing by scientists: practices and perceptions, PLOS ONE, 2020
• Automating Inequality, St. Martin’s Press, 2018
• Open Science Training Handbook, FOSTER, 2018
• The oligopoly of academic publishers in the digital era, PLOS ONE, 2015

External links

• UNESCO — Recommendation on Open Science
• cOAlition S — Plan S
• DOI Foundation
• Crossref
• DataCite
• ORCID
• Research Organization Registry (ROR)
• Directory of Open Access Journals (DOAJ)
• Sherpa Romeo — Journal OA policies
• Open Science Framework (OSF)
• Zenodo
• Dryad
• Figshare
• arXiv
• bioRxiv
• medRxiv
• GO FAIR — FAIR Principles
• DORA — Responsible research assessment
• The Turing Way — Open, reproducible data science