[License-review] Review for the NIST Software License

Rob Landley rob at landley.net
Tue Sep 30 19:22:26 UTC 2025


On 9/30/25 03:57, Carlo Piana via License-review wrote:
> Lucas,
> 
> if I understand correctly, this should not technically be a license, since the software is not subject to copyright in the USA as far as it has been created by NIST employees. I think that if software is not given protection in the state of first publication is not protected even elsewhere, under the Berne Convention, therefore this is basically a dedication to public domain, whose primary scope is the liability disclaimer(s).
> 
> However, the "provided that you keep intact this entire notice" is technically (US lawyers please help) a condition, that means this is a license with conditional grant, after all. The other condition-like provision uses the verb "should", which is more of an invite, at face value.

This "stuttering problem" was the main reason I did the public domain 
equivalent "zero clause BSD license" in 2013 (and walked it through the 
SPDX approval process in 2015 when Samsung asked).

https://landley.net/toybox/license.html

Android's "toolbox" (which my project was aiming to replace) was under a 
conventional public domain adjacent license that had "copy this blob of 
text into derived works", but the license file in the android repository 
had over 20 repeated copies of the same license (concatenated together), 
and when I asked _why_ the Android guys said "the copyright date and/or 
who changed, and our lawyers said a strict reading of the license 
requires..." Let alone trying to combine Apache and MIT together in the 
same codebase. (Which license applies to which parts? I can't re-license 
inherited code, so license text can only increase in derived works. The 
"kindle paperwhite" had over 200 pages of stuttering in its 
about->license pulldown.)

Notice how busybox's networking/ping.c put a GPLv2 dedication at the 
start and moved the original BSD license at the end, presumably in hopes 
that the contradiction between the two wouldn't be as obvious. (GPLv2: 
no additional restrictions are allowed. BSD: you must drag this blob of 
text around or you're in violation.)

The main distinguishing factor of a public domain equivalent license is 
it does NOT require copying a specific blob of text into derived works, 
thus making it compatible with other public domain equivalent licenses 
and rendering most "apache vs ISC" arguments irrelevant.

https://en.wikipedia.org/wiki/Public-domain-equivalent_license

Rob


More information about the License-review mailing list