7

By a quick count, wikipedia lists almost fifty different open source / free software licenses.

Assuming (and this may not be valid) that many of these licenses were created in the early days of the movement, when it was trying to settle on common conventions, why do they all persist?

Given the seeming ideological split between the FSF and OSI, it is unlikely we would ever converge on a single common license. But much of the FLOSS movement is based on open standards, best practices, and so on.

What stops the FLOSS movement agreeing on a small set of licenses to cover, say, four or five levels of permissiveness, and everyone relicense to adopt them?

  • 4
    Are we sure that the small set isn't already there, in practice? How much code is licensed under each of these licenses, maybe the large majority already uses only a few already? – Michael Schumacher Jun 28 '15 at 20:21
  • The medfloss.org website provides a repository of OS software in health IT. Looking at the statistics of the licences of listed projects it shows the following picture: GNU GPL (159), GNU LGPL (40), BSD (35), Apache (18), GNU AGPL (13), Apache 2.0 (12), GNU FDL (12), MPL (8), MIT (7) with GNU-style licenses clearly being used most often Source: medfloss.org/project-wizard – Hans Demski Jul 3 '15 at 9:44
5

The major problem is one you've already noted: having everyone relicense. There are, literally*, millions of open-source projects. For each project, the author might have abandoned it, or be unreachable, or just not want to relicense.

Also, one organization or meeting can't hope to cover all the needs of every developer and every project - software must have different licenses to hardware, to drinks, etc., so you'll get people contacting you for years afterwards asking for the addition of another license to cover X.

The set of licenses will, over time, tend towards the smallest workable set, simply because humans are lazy.

That said, to some extent this has happened. If you ask someone involved in open source to tell you a few licenses, you can bet they'll say MIT, BSD, GNU GPL, and probably CC's set. Those are the accepted licenses, and people looking for a license will come across those first. Many of the others will (or have) fall out of use as people go for the big-name licenses.


* No, I don't know this for certain, but it seems exceedingly likely.

3

Somewhat sloppy quantative analysis on the current licences used on github shows that this is already somewhat happening. This is just a single datapoint, and it's not very solid data, but it does show that the MIT license, the BSD license(s), the GPL and the Apache license together make up for the vast majority of cases.

The long tail is very long, but not very deep. I don't know about any initiatives by the OSI or FSF to trim down the number of licenses they certify out of name of simplicity, but that would still leave a very large number of existing projects with another license and no real incentive to change it.

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.