What Exactly Counts as Open Data?
What "Open Data" Actually Means
People often use "open data" to describe any dataset a government or company publishes online. But public and open are not the same thing. A PDF on a website is public. A machine-readable file with a clear license that anyone can reuse is open. The difference matters because it decides whether you can legally build a product, run an analysis, or reshare a table without asking permission.

The reference point is the Open Definition, maintained by the Open Knowledge Foundation. It states that data is open if anyone is free to use, reuse, and redistribute it, subject only to measures that preserve provenance and openness. That single sentence hides several concrete criteria.
The Core Criteria, One by One
The Open Definition breaks openness into practical requirements. A dataset must be available as a whole, at no more than a reasonable reproduction cost, and preferably downloadable over the internet. It must also be in a convenient and modifiable form, which usually means a machine-readable format such as CSV, JSON, or XML rather than a scanned image.
The license is the part people skip. Open data must be placed under terms that allow free reuse, including commercial use, and must not discriminate against persons, groups, or fields of endeavor. A license that forbids commercial use, or that only permits academic research, fails the test.
Public vs. Open: A Quick Comparison
The table below shows how the two categories usually differ in practice. Many portals describe themselves as open while failing more than one of these checks.
| Feature | Merely public | Truly open |
|---|---|---|
| Access | Visible on a website | Downloadable in bulk |
| Format | PDF or scanned image | CSV, JSON, XML |
| License | All rights reserved | Permits reuse and redistribution |
| Cost | Free to view | Free or nominal reproduction cost |
Common Traps and Half-Open Data
Some releases are open in spirit but closed in practice. A dataset may allow reuse but require registration, a password, or a click-through agreement that forbids redistribution. Others are technically open but practically useless because they change column names every month.
There is also the question of attribution. Requiring credit is allowed, but requiring permission for every reuse is not. A short quotation from the Open Definition captures the spirit:
"Open means anyone can freely access, use, modify, and share for any purpose."
How to Check a Dataset Yourself
You do not need a legal team to evaluate openness. A short checklist covers most cases:
- Can you download the entire dataset without an account?
- Is the format machine-readable and documented?
- Is there a license that allows commercial reuse?
- Can you redistribute it without asking for permission?
If you want a more formal process, follow these steps:
- Find the license statement on the source page.
- Compare it against the Open Definition criteria.
- Test whether the file opens in a spreadsheet or parser.
- Note any restrictions on redistribution or modification.
Open data is a promise about what others may do with the material, not just a statement that it exists. Applying the Open Definition keeps that promise honest.
