Open data is data that anyone can freely access, use, modify and share for any purpose, subject at most to requirements that preserve attribution and openness. Among data fundamentals, the term describes a combination of legal openness, meaning an open license, and technical openness, meaning machine-readable formats available without unnecessary barriers.
Open data is published by governments, research institutions, international organizations and companies. Typical examples are public transport timetables, weather observations, census tables, company registers and scientific measurements. Its value comes from reuse by people the publisher never anticipated, which only works when the data is both permitted and practical to use.
What makes data open
Data counts as open only when several conditions hold together: an open license, free or at most reproduction-cost access, a usable format and no discrimination against fields of use or groups of users. A dataset that can be viewed on a web page but may not be reused commercially is publicly available, not open.
The widely used Open Definition formalizes these conditions. The table summarizes the core criteria and what typically breaks them.
| Criterion | Requirement | Common violation |
|---|---|---|
| Open license | Permits use, modification and redistribution | Non-commercial or no-derivatives clauses |
| Access | Free download, no registration barriers | Data only on request or behind payment |
| Machine readability | Structured format a program can parse | Tables published as scanned PDF files |
| Open format | Specification free of usage restrictions | Proprietary formats requiring licensed software |
| No discrimination | Same terms for every person and purpose | Reuse limited to research or one sector |
Openness is a property of the data and its terms together. The same table can be open in one release and closed in the next if the license changes, which is why reusers record the license version they relied on.
Licenses and legal basis
An open license is the legal instrument that turns access into permission. Without one, copyright, database rights or contract terms may still restrict reuse, even when the files are freely downloadable.
Most open data uses a small set of standard licenses, which keeps combining datasets legally straightforward. Custom national licenses exist as well and are often designed to be compatible with a widely used standard license.
| License | Main condition | Typical use |
|---|---|---|
| CC0 1.0 | None, a public domain dedication | Statistics, scientific and reference data |
| CC BY 4.0 | Attribution to the source | Government and research datasets |
| ODC-BY | Attribution, written for databases | Database publications |
| ODbL | Attribution and share-alike for derived databases | Collaborative map and community data |
Share-alike licenses require derived databases to be released under the same terms, which protects the commons but can conflict with proprietary products. Licenses with non-commercial or no-derivatives clauses do not meet the open data definition, even when they are free of charge.
Formats and levels of openness
The usefulness of open data depends heavily on how it is technically published. A common way to describe this is the five-star deployment scheme, which rates publication from an openly licensed document up to linked data.
| Level | Publication form | Example format |
|---|---|---|
| One star | Any format under an open license | PDF, image |
| Two stars | Structured, machine-readable data | Spreadsheet in a proprietary format |
| Three stars | Structured data in a non-proprietary format | CSV, JSON |
| Four stars | Things identified by URIs | RDF with stable identifiers |
| Five stars | Linked to other data for context | Linked data with external references |
In practice, three-star publication in CSV or JSON with good metadata is the most frequent target, because it serves both spreadsheet users and developers. Large or frequently updated datasets are often offered through bulk downloads and APIs, since an API alone makes complete analysis and archiving difficult.
Open data compared with related concepts
Open data is frequently confused with public data, shared data and open access, which overlap but are not the same thing. The distinction matters for what a reuser is legally allowed to do.
| Concept | Who can use it | Reuse rights |
|---|---|---|
| Open data | Anyone | Full reuse under an open license |
| Public data | Anyone can view it | Not necessarily reusable |
| Shared data | Named partners or members | Defined by agreements |
| Open access | Anyone can read publications | Often limited to reading articles |
| Closed data | Internal users only | No external reuse |
Open data also supports data democratization inside organizations and outside them, but it is not a synonym for it. Personal data is normally excluded from open publication, and data protection law applies regardless of any license.
Benefits, risks and quality issues
Open data creates value by lowering the cost of reuse: journalists verify claims, researchers reproduce results, developers build services and public bodies avoid duplicate data collection. It also strengthens transparency and accountability when governments publish budgets, contracts and performance figures.
The main risks are re-identification of individuals, when supposedly anonymous records are combined with other sources, and misinterpretation, when reusers lack context. Careful aggregation, suppression of small counts and clear documentation reduce both.
Publishers also face ongoing maintenance costs. A dataset that is released once and never updated quickly loses its value, so sustainable open data programs assign a named data owner, a fixed update schedule and a channel for reporting errors.
- Missing metadata leaves units, definitions and collection methods unclear.
- Irregular updates make published data stale without warning.
- Changing schemas break automated pipelines that depend on column names.
- Unclear provenance weakens trust in the data quality and lineage of a dataset.
A non-technical view
Open data is like a public library with unusual rules. Visitors may take the books home, copy them and build something new from them. Nobody asks for permission or payment.
Think of a city that publishes its bus timetables this way. Anyone can make a travel app, study delays or suggest a better route. The data is free to use and reuse. It is also published in a form that computers can read, so nobody has to type it in again.
A concrete example
A mid-sized city publishes its public transport schedules and real-time vehicle positions as open data under an attribution license, in a standard timetable format updated every 30 seconds. It also releases bicycle counter readings as CSV files every day.
Within a year, independent developers integrate the timetables into several journey planners, and a local university uses the vehicle positions to identify the ten stops with the largest delays. The city adjusts traffic light priority at those junctions based on the analysis.
Because the license and formats were clear from the start, no individual agreements were needed, and the city spends less time answering data requests than before publication.