I'd like to ask a more general question and then I'd like to ask a more detailed one based on my specific case.

Suppose that I'm writing a software I'm planning to release under the terms of the AGPLv3 (or the GPLv3) and that software uses data taken from Wikipedia in order to work (basically, the software would take real data from Wikipedia and then use it to generate fake data. All the text on Wikipedia is under CC-BY-SA 3.0. I understand that I can use the data provided that I release any derivative work under the same (or compatible) license.

I now have the general question:

  1. If I put the data in a text file and then I tell my software to go fetch that data from that text file, is the software defined a derivative work of the data? What if I hard-code that data in a source code file?
  2. Is the data my software generates under the CC-BY-SA 3.0 license? I understand that the output of a (A)GPLv3 program is not under the (A)GPL, but is it so with this license?

Then to the more specific question: the data I'd like to use from Wikipedia is composed of:

  1. World country names
  2. World country flags (I took the main colors of the world country flags out either from the image
  3. City names of different countries
  4. Names and surnames of people of different countries (I'd like to use them even unmodified)

For the first point I'm pretty sure the list of world countries is public knowledge and hence in the public domain. For the second point, while the images of the flags are public knowledge (even Wikipedia acknowledges so), the description from where I extracted some color information (see, for instance the "flag colors" section of this page) would be under the CC-BY-SA license. For the third and fourth points... I don't know.

The problem is that I do not want to distribute my software under the CC-BY-SA license because even Creative Commons itself acknowledges it's not a good idea. Instead, I'd like to use either the GPLv3 or the AGPLv3 (if possible, both with the "or later" clause).

Thanks in advance.

1 Answer 1


The important point to focus on is that, with certain exceptions, the licence on a piece of software doesn't affect the outputs of that software. Instead, the outputs are generally a derivative work of the inputs, so the licence on the outputs (if any) will be governed by what rights and obligations the program's user has with regard to making derivative works of the inputs (s)he feeds to the program.

The exceptions generally come into play when the program copies large chunks of itself into its outputs, which doesn't seem to apply in this case.

I answer this general question first because hopefully I can then avoid the specific ones. You say your main concern is that you "do not want to distribute [your] software under the CC-BY-SA license... Instead, [you]'d like to use either the GPLv3 or the AGPLv3". In this case, my opinion is that you are perfectly free to do exactly that.

If you are distributing input data sets (flag images, city names, etc.) alongside your software, you will need to check that you have the right to do so, but as long as you do, you don't need to worry any further.

  • 1
    Thank you for your answer. Without the data from Wikipedia my program cannot work, hence the need to redistribute it alongside my program. I thought I could use a separate file for the data so that I can clearly state that the data is under CC-BY-SA 3 and my code is under GPLv3. As far as the output is concerned, I understand it is a derivative work (made by my program and further used by it for other purposes) and hence under CC-BY-SA 3, but I would be ok with that (the program will use the data to create a "playable story").
    – LuxGiammi
    May 12, 2021 at 17:48
  • What I initially read about this was this answer (opensource.stackexchange.com/questions/5334/…) where users suggested that, because the software used the CC-BY-SA data as training, it qualifies as a derivative work. I can do that, but I prefer not to and distribute the data along with the code to generate the fictional data, which will be the "Adaptation".In order to fix the problems with that previous answer was enough to read the training data from an external file - released under CC-BY-SA -, am I right?
    – LuxGiammi
    May 12, 2021 at 17:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.