RichStyle: The Angel is in The Detail.

Copyrighting Databases

Here is a brief summary of a discussion at GNU mailing-list about a GNU license for databases/dictionaries, and a primarily conclusion out of it.

“You cannot copyright databases in the US AFAIK. There was a case about a phone dictionary”. “why would we regard some dictionaries’ definitions as better than others? There is not a single, correct definition of any English word”.
“A dictionary requires as much work as a phone book and isn’t a very creative process”. “The amount of work isn’t important, it’s about the creativity. Writing all those definitions in the dictionary requires creativity, so you get copyright on the dictionary”.
“you cannot copyright the name + number in that phone book, since that is considered a ‘fact’.”
“a list (database) of genomes for a bunch of species isn’t copyrightable either”.

Yes, phone numbers, contact info, and genomes, are definitely facts. In this case, I don’t claim that they are subject to copyright, but maybe its archive’s (container’s) design.

By contrast, natural languages, so far, don’t have a structured architecture. Most people believe that its impossible or very difficult to put them in a structured shape.

So, here, in this case, in a dictionary case, when you build a bilingual dictionary —not just a word-list—, in fact, you’re trying to convert something unstructured to a structured thing. You’re trying to create and innovate structuring standards, and structuring contents according to these standards.

So, the activity that make databases subject to copyright is Designing, which structures something unstructured; since word definitions are not “Facts”.