A phone directory case set copyright's creativity floor

The writer's problem

A phone company wanted to stop a rival from copying names, towns and numbers out of its white pages. Rural Telephone Service's directory listings were compiled by ordinary business routine: canvass the subscribers, alphabetize the names, print the book. When Feist Publications copied thousands of those listings into its own regional directory, Rural sued for copyright infringement. The dispute forced the Supreme Court to state, in plain terms, what copyright actually protects when a work consists mostly of collected facts: not effort, but a minimal act of creative selection or arrangement.

What the documents show

The Court's opinion, decided 27 March 1991, held that copyright protection turns on originality, meaning independent creation plus 'some minimal degree of creativity,' not on the labor a compiler invested. It rejected the 'sweat of the brow' reasoning, which had protected effort itself, calling that approach incompatible with the 'most fundamental axiom of copyright law': facts and ideas cannot be owned by anyone. Rural's alphabetical listing of subscribers involved no creative choice at all; an alphabetical directory is exactly the arrangement a reader expects, and expected arrangements do not clear even copyright's low bar.

The Copyright Office's own Compendium of U.S. Copyright Office Practices builds on that holding, quoting Feist's 'minimal degree of creativity' language and its rejection of anything 'entirely typical' as the test examiners apply to every application. That same threshold resurfaces, without being named, in the Office's 2023 guidance on AI-generated content: a human who selects or arranges AI output can register a claim only if the result, 'as a whole,' constitutes an original work of authorship under the same compilation standard Feist construed. Feist itself says nothing about computers; the guidance imports its creativity floor, not its facts.

The editorial choice

An editor deciding whether an AI-assisted anthology or data-heavy manuscript can be registered is really asking a Feist question: did a human select or arrange this material in a way that clears a minimal, but real, creativity bar, or does it merely reproduce an expected order. This is a judgment about the specific manuscript, not a rule that any AI involvement disqualifies a work, and not a promise that any human touch is enough. The safer practice is to document what a human chose and why, since the standard rewards demonstrable selection over asserted effort.

What stays with the author

The author's own record of what they arranged and why is what a Feist-style review looks for. Neither the opinion nor the Compendium says how much AI-assisted selection is enough, and that gap remains a case-by-case judgment.

• Did a human choose or arrange this material, or does it follow the order any competent compiler would produce?
• Can the selection be described and dated, rather than only asserted after the fact?
• Would the arrangement look the same if a different person had made the same selection?

Feist never mentions a machine; the guidance that leans on it does, and the gap between an old ruling about phone books and a policy statement about generative tools is where an author's documented choices now do the work.