5

In a comment on my question What if anything is wrong with the Apache License 2.0? someone (Zimm) reminded me that its considered bad practice to use a less widely known license.

How wildly known is the COIL license? How wildly used are copyfree licenses in general?

Jeff Atwood's recommendation of the WTFPL license in https://blog.codinghorror.com/pick-a-license-any-license/ is perhaps the highest profile recommendation of a copyfree license I can find.

How do you go about answering the question of how widely known is wide enough?

These questions are too broad for an easy answer so what I'm looking for is some pointers on how to going about getting a good enough answer to them.

  • I may be perverting the use of the "license-recommendation" tag here. This is about how you decide to recommend a license or not rather than an actual question asking for one. – Bruce Adams Aug 24 '16 at 1:04
  • The linked article by Jeff Atwood seems unfortunately misguided and misleading. I wouldn't put too much weight on it. (E.g. it confounds the notions of "copyright" (which is inalienable in many jurisdictions) and "license" (which is a thing you choose). Also, the WTFPL is probably a reason for others to not use your code because it's vague and unclear (in the sense that it's unclear whether a user might get sued eventually). – Kerrek SB Aug 24 '16 at 12:40
  • I actually agree with you there. Its a relatively high profile recommendation but definitely not a good one. – Bruce Adams Aug 24 '16 at 13:59
3

As mentioned in an answer to your other question, while "copyfree" is not a widely used term, some "copyfree" licenses are very much used: CC-0, the MIT license, the Simplified BSD license, the Unlicense.

I make this answer based on my own experience of checking the license of each software I come accross (I have personnally never seen the COIL license) and based on the recommendation of the FSF http://www.gnu.org/licenses/license-list.en.html, the OSI https://opensource.org/licenses/category and Github http://choosealicense.com/.

There are ways to make statistics on the usage of various licenses but this usually starts by choosing a reference database. Whether this database is Sourceforge or Github, it will bias the data towards GPL or towards MIT/Apache.

  • A problem with some of those database's is people not bothering to select a license at all or doing so without thinking about it enough. The selector at github fixes the first by encouraging the second. The age of a license will create a bias, as will is provenance. Some of that is useful of course. A well established Mozilla, GNU or Apache license for example. I prefer to choose based on facts not other peoples choices which could just be what's fashionable but I can see with this facts are quite slippery (like the teams of lawyers involved?) – Bruce Adams Aug 24 '16 at 9:10
  • Yes it create a bias, but is a fair competition between licenses really something that has to be encouraged? Sometimes, having a standard, even if not the best one, is preferable instead. In the end it is up to you to decide. My final advice would be: if you have good reasons to think that license A is better than license B, then choose license A. If licenses A and B are very similar and you cannot really decide which is better, then choose the more popular of the two. – Zimm i48 Aug 24 '16 at 9:19
  • I recall I wanted the MIT license originally then came across this news.ycombinator.com/item?id=3402450 A problem with the common licenses is the lack of a patent clause (though I am in the EU where this shouldn't matter, my software might not stay there). – Bruce Adams Aug 25 '16 at 0:16
2

What would be good reasons to not use code just because of a less common license? I can't imagine a developer doing that, so it must stem from considering the legal aspects. So I guess it comes down to whether its percieved as a "crayon" license.

Criteria for deciding that could be:

  • was it drafted by a competent lawyer or legal team?

  • is it recognised by any respected licensing bodies (osi, fsf, copyfree)

Please improve on my answer if you can. For less popular licenses that information can be harder to come by.

The COIL license looks like it was drafted by a lawyer but there is no proof on its website (http://coil.apotheon.org/) Its too new or too rare for the OSI and FSF to have bothered with an opinion on it yet.

  • 1
    This answer seems to address "why (not) use existing code under an uncommon license?" while the question seems also to include the additional concern of "why (not) license your brand new code under a less common license?". This seems like a perfectly good answer, but I'm concerned about the apparent (to me, anyway) mismatch between these concerns. Should the question be narrowed, or the answer broadened? Or did I just misread something here? – apsillers Aug 24 '16 at 2:54
  • I never said that you, as a developer or a user, should not use code just because the license is less common. I only said that when you have to decide on a license for a new software, you should strive for a better known license unless you have good reasons not to. For very well known licenses, see this list choosealicense.com/licenses. Some companies have a strict policy about open source. Each license must be reviewed by their attorneys. If an employee finds a new software and it is under an already approved license, they can start using it right away. If not, it has to be reviewed. – Zimm i48 Aug 24 '16 at 7:32
  • I know but you did make a good point that I seemed to have forgotten. Hence this question. – Bruce Adams Aug 24 '16 at 8:00
  • @apsillers I was trying to address "why (not) license your brand new code under a less common license?" in terms of "why (not) use existing code under an uncommon license?" as that is one of the criteria. Though it is not the only one. – Bruce Adams Aug 24 '16 at 8:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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