Status: educational explainer. This page reports no experiment results. Project S.P.H.E.R.E. is at the proposal stage, hypotheses are not findings, and nothing here amends the registered LOC-1 protocol.

The core idea

Science is a process of producing reliable knowledge through systematic observation and criticism. It works best when observations can be checked. Open science is a collection of practices that make checking possible by removing barriers to access, scrutiny, and reproduction of research. At its core, the argument is straightforward: if the data, methods, and analysis that support a conclusion are unavailable to other researchers, those researchers cannot verify the conclusion, identify errors, or build on it with confidence. Openness is the structural prerequisite for effective scientific criticism.

The term "open science" encompasses a wide range of specific practices that operate at different stages of the research process — from how a study is planned to how its outputs are shared. These practices are increasingly required or encouraged by funding agencies, journals, and research institutions around the world, in response to evidence from the replication crisis that scientific findings are less reliable than had been assumed.

Five key open-science practices

Open access publication. Research papers that are freely readable by anyone — without subscription fees, paywalls, or institutional access — are open access. The practical effect is that researchers at institutions without expensive journal subscriptions, practitioners, policymakers, journalists, and the interested public can all read and evaluate the primary literature. Two main models exist: "gold" open access, where the journal makes the article immediately available but typically charges the author or their institution an article-processing charge; and "green" open access, where the author deposits a preprint or accepted manuscript in a public repository (such as arXiv, bioRxiv, or OSF Preprints) regardless of the journal's position.

Open data. Raw or processed measurement data is shared with documentation sufficient to understand its structure, collection method, and limitations. Formats matter: a spreadsheet emailed to individual requesters on an informal basis is not the same as data deposited in a versioned, citable repository with a persistent identifier. Well-documented open data enables independent reproduction of analysis, meta-analysis across studies, and discovery of errors in the original data reduction. Standards like FAIR (Findable, Accessible, Interoperable, Reusable) provide a framework for evaluating data quality.

Open methods and code. Sharing enough procedural detail to repeat the work is fundamental to science but rarely achieved in practice. Published "Methods" sections routinely omit critical details that researchers know from experience but do not think to write down. Sharing analysis code addresses part of this by making the computational choices explicit and reproducible. Well-documented code can be run by others to verify results, and the discipline of writing it forces the researcher to make every decision explicit rather than leaving it implicit in the mind of the analyst.

Preregistration. A time-stamped, publicly deposited record of the research hypothesis, sampling plan, and analysis procedure made before data collection begins. Preregistration creates a public commitment that prevents post-hoc modification of these elements in response to what the data show — the core mechanism of p-hacking and other forms of analysis flexibility. See preregistration explained for a detailed treatment. Preregistration does not prevent exploratory analysis; it requires clearly labeling exploratory results as such, separate from the confirmatory tests specified in advance.

Open peer review. Traditional peer review is anonymous and private: authors do not know who reviewed their work, reviewers do not know whether their comments were acted on, and the exchange is never seen by readers. Open peer review makes some or all of these elements public — review reports, reviewer identities, author responses — creating a citable, accountable record of the evaluation process. It also enables post-publication review, where anyone can comment on a published paper and those comments persist alongside the paper.

Decentralized science and open science

Decentralized science (DeSci) is a related but distinct movement that applies distributed-infrastructure approaches — public blockchains, decentralized repositories, on-chain governance mechanisms — to research funding, publication, and data provenance. DeSci and open science share the goal of making research more transparent and accessible, but they differ in their emphasis and methods: open science is primarily about practices and norms, while DeSci is primarily about infrastructure and incentive mechanisms.

For the purposes of this project, the relevant overlap is in open data and open publishing. Project S.P.H.E.R.E. is structured as a DeSci project, but its research-integrity commitments are those of open science: preregistration, open data release, published uncertainty budgets, and public protocols. The word "decentralized" in the project name refers to its independence from institutional affiliation and to its commitment to open publishing, not to blockchain mechanisms or token-based governance.

Limits and trade-offs

Openness involves real costs and genuine trade-offs, and good open-science practice acknowledges them.

Data sharing requires investment in documentation, format standardization, and repository infrastructure. Raw experimental data that is meaningful to the researchers who collected it may be incomprehensible to anyone else without a data dictionary, a processing manual, and documentation of measurement conditions. Producing that documentation takes time that competes with other research activities.

Some data cannot be shared freely due to privacy obligations, participant confidentiality, national security restrictions, or contractual constraints. Open science does not require sharing data that cannot legally or ethically be shared; it requires being explicit about what is and is not shared, and why, rather than defaulting to opacity.

Sharing does not guarantee scrutiny. Most shared datasets are downloaded rarely; most shared code is never run by anyone other than its author. The mere availability of data does not produce the verification benefit that motivates sharing — that requires a community with both the incentive and the expertise to actually check the work. This is an institutional problem, not a reason to avoid sharing, but it tempers expectations about how much openness alone can accomplish.

Open science at S.P.H.E.R.E.

Project S.P.H.E.R.E. is designed around open-science principles from the outset. The registered LOC-1 protocol is publicly deposited on the Open Science Framework (OSF) with a permanent identifier at osf.io/5b6mj. The project whitepaper is archived with a DOI on Zenodo at 10.5281/zenodo.23583932. The website and codebase are publicly accessible on GitHub. Budget and transparency information is published on the transparency page.

Everything published so far is a proposal — a protocol, a whitepaper, a set of research pages outlining the plan. No measurement data has been collected. When data collection begins under the LOC-1 protocol, data and analysis code will be released under open licenses as part of the standard reporting process. The project invites outside review and collaboration through the collaboration page. More background is available at Wikipedia: Open science.

Keep learning