[License-discuss] Can OSI take stance that U.S. public domain is open source?

Karl Fogel kfogel at red-bean.com
Sat May 3 19:00:53 UTC 2014

Richard Fontana <fontana at sharpeleven.org> writes:
>This work's authors seem to explicitly say that they are dedicating it
>to the public domain, not merely (or explicitly at all, as far as
>I can see here) relying on the notion of statutory public domain for
>US government works. I'd argue those are two different concepts of
>"public domain" (one of which is really something more akin to the
>effect achieved by CC0). 

No, I took away the opposite impression: they are US gov't employees
working on a gov't project on gov't time, and that they contend they
therefore have no choice about its public domain status.

That's what they seem to be saying in the comments, and it matches the
owner of the GitHub account, which is the US National Geospatial Agency:


>With statutory public domain works, you can't be sure out of context
>what the status of the work is when published outside the US. See e.g.
>http://www.cendi.gov/publications/04-8copyright.html#317. I've found
>that many US government lawyers dealing with open source seem to assume
>that 17 USC 105 operates worldwide (this sometimes comes up in the form
>of a refusal to sign CLAs because 'there is no copyright to license').
>Also with statutory public domain works you have the same old MXM/CC0
>inconsistency problem in a different form. Consider the case of public
>domain source code created by a US government employee, having features
>covered by a patent held by the US government.

The patent issue would apply just as much if it were MIT- or
BSD-licensed, though, and we'd call it "open source" then, right?

http://www.cendi.gov/publications/04-8copyright.html#317 seems to
indicate that we'd need an explicit notice that the U.S. government will
not claim any copyright on the work in jurisdictions outside the U.S.

If the US government were to publish such notice on a given work -- say,
if standardized language for doing so were approved by the OSI :-) --
then would there be any sense in which the work would not be compliant
with the OSD?  E.g., would its open-sourceness be materially different
from an MIT-licensed work?

It wouldn't have any attribution or no-warranty clause, but the
*absence* of those clauses is not a problem w.r.t. the OSD.


More information about the License-discuss mailing list