Open data licensing and terms of use for agency datasets

Reference article · 5 min read

An agency preparing to publish a dataset or a map service usually has the technical work finished before anyone asks what terms will sit next to the download button. The remaining question is what the agency is granting the person who clicks. Settle it before the first download, because terms are hard to change once copies are in circulation.

Public records status and licensing answer different questions. Records law tells the agency what it must disclose. It generally says nothing about what a recipient may do with the record afterward, and publishing without addressing that leaves reusers guessing.

Disclosure is not the same as a license

Rights in published data come from outside the records statute: copyright, terms attached by whoever supplied the source material, and in some states statutes governing particular record types.

• Copyright in government works varies by level of government. Works of the federal government are not protected by copyright within the United States. That rule does not extend to states, counties, cities, or districts. Whether your agency may hold copyright in its own works is a question of state law.
• Facts are weakly protected. United States copyright does not protect facts themselves. A table of parcel attributes may carry little protection, while documentation and cartographic styling may carry more. A license may be doing more signaling work than legal work.
• Some records carry use restrictions regardless of license. Certain categories of personal information are subject to limits on commercial use or resale. A license cannot grant what a statute withholds.

These rules differ by state and change over time. Confirm the approach with agency counsel rather than assuming another jurisdiction's language transfers.

The common options

• Public domain dedication. Instruments such as CC0 or the Open Data Commons PDDL waive rights as far as the law allows and impose no conditions, not even attribution. This is the lowest friction option and the easiest for aggregators and community mapping projects to accept. The tradeoff is giving up the ability to require credit.
• Attribution license. Creative Commons Attribution and the Open Data Commons Attribution License permit reuse if the source is credited. Many agencies prefer this because credit creates a path back to the authoritative copy.
• Share-alike terms. The Open Database License requires derived databases to be released under the same terms. It keeps derivatives open and is harder to combine with other sources.
• Custom agency terms. Local drafting can fit an unusual constraint, but the result is not machine readable, portals cannot interpret it, and every reuser's counsel reads it fresh. Conditions such as no commercial use also place the dataset outside what most catalogs treat as open data.

If the agency concludes it holds no rights to license, a public domain mark is more honest than a copyright license on unrestricted material.

Accuracy disclaimers, kept separate from the grant

A disclaimer limits what the agency promises about quality. A license states what the agency permits. Keep them in distinct sections so neither is read as qualifying the other.

• Condition. Provided as is, without warranty of accuracy, completeness, or fitness for a particular purpose.
• Positional accuracy. GIS layers are not surveys and are not intended for boundary determination or engineering design unless the agency says otherwise.
• Currency. State the update cadence and extract date, and note that downstream copies may be stale.
• Source and endorsement. Identify where the current version lives, and say that reuse does not imply agency endorsement of the reuser or the product.

Third party content inside GIS layers

This is where publication plans most often stall. A layer that looks like agency work may carry content the agency cannot redistribute: purchased aerial imagery, commercial road centerlines or address points, a licensed basemap, or contractor deliverables whose ownership was never specified.

1. Inventory each layer's sources and identify the agreement governing each one.
2. Read each acquisition contract for deliverable ownership and redistribution rights.
3. Separate restricted content from publishable content rather than fusing them into one service.
4. Consider a derived product where the source cannot be released, if the upstream terms permit derivatives.
5. Carry any upstream attribution requirement into your own terms, and record the decision in the metadata.

Attribution, redistribution, and where the terms live

Write the credit line you want, in the exact words, and ask reusers to note modifications and the retrieval date. Say plainly that redistribution is permitted, since silence reads as prohibition. Publish the terms where machines and people will both find them: a license field in the catalog record, the use and access constraints elements of the agency's metadata standard, a license file in every download bundle, and the service documentation.

How to start, and what to ask vendors

Pick one default license for the agency rather than deciding dataset by dataset, run it past counsel once, and publish it as policy with a decision date. When acquiring imagery, data conversion, or a hosting platform, ask vendors directly:

• Do the deliverables include the right to publish the results publicly under our open data license?
• Which components are licensed from third parties, and what do those licenses say about redistribution?
• Does the platform assert any rights in data we upload or in products derived from it?
• Can we export everything, including metadata, in a documented format at any time?
• Do the portal's terms of service add conditions to our data, and can we display our own license alongside them?

Put the answers in the contract. Rights to publish are cheapest at acquisition and most expensive once the dataset is built.

HQT works with state and local agencies on data platforms, GIS services, and the catalog and metadata work that carries licensing information to the people downloading data.

Frequently asked questions