4 ms·

> If software is a process of evolution and continuously building on previous work, copyleft licenses represent an evolutionary dead end, while liberal licenses

> If software is a process of evolution and continuously building on previous work, copyleft licenses represent an evolutionary dead end, while liberal licenses represent branches of the tree that can live and progress much longer. This strikes me as not quite right and a bit out of place for an essay that's otherwise pretty thoughtful. Copyleft is a niche or a branch, not a dead end. There's a fundamental question facing anyone approaching the economics of a software project (and the economics of many other kinds of projects too): how much of the value of the project do you try to control/capture vs how much do you try to share in an effort to multiply the value? Source licensing questions are a subset of this: how much of the value of the project lies in certain levels of control of the source vs the prospect of a community around the source. The copyleft licenses are one answer to the question: if the utility value of software developed by an active community around the source strictly exceeds the value of any one player controlling access to the source, control the source to maintain community property of contributions, but let anyone work out their answer to the value capture question via the utility of the software or their knowledge about it. It isn't the only answer and I think it's fair to say that not every useful piece of software can be developed or maintained on that model, which is why I'd say it's an adaptation to a niche. A niche is pretty different than a dead end, though. This also addresses some odd conceptions of copyleft being incompatible with improving ones own lot or based in premature adoption of Star Trek values.

pauldix 8y ago
Heh, the Star Trek thing might have been a bit of an overreach based on people's reactions, but let me see if I can restate the point I was trying to make with that little bit. Licenses like MIT give the most possible freedom, particularly with how derivative works are commercialized. Copyleft licenses limit the freedoms of derivative works by forcing them to be copyleft (although with some hacks and exceptions). This limitation means that options for creating a derivative commercial product from copyleft sources are very limited. If you are able to get around the copyleft restriction (like SaaS products developed on GPL and AGPL'd software) then you are able to build a commercial product. But if you want to ship that software to a customer in a licensed product or a piece of hardware, you won't be able to do it without exposing your entire source code to be copyleft. I've heard people argue that copyleft is better because it forces this opening of source on everyone. It's a utopian ideal that I think limits its utility, particularly with regard to derivative works.
- yarrel 8y ago
  Licenses like MIT don't give "the most possible freedom". I am not more free to modify and repair the software running on an iPhone despite the plethora of BSD and other permissive licenses on it than I am on a laptop running GNU/Linux under the GPL. Likewise if you want to commercially exploit some MIT-licensed software that I have modified and am selling binaries of, you are not more free to commercially exploit that than you would be under the GPL. Copyleft doesn't "force" the "opening" of "source" on anyone. It's a pragmatic measure covering existing software in order to ensure recipients of that software are free to use it as they wish. The GPL doesn't limit utility. You are confusing use value and exchange value. The GPL ensures that software retains its utility for its users. All that said, if you want to give all your software to users under the MIT or Apache licenses knock yourself out.
  - SpicyLemonZest 8y ago
    The LGPL works as you're describing; I can license my own code however I want, as long as I let people freely modify all the LGPL libraries it links in. The GPL proper is very deliberately viral.
    - robocat 8y ago
      The LGPL is fine for libraries - but that limits where it can be used. I think the LGPL becomes something just like the GPL for scripting languages because there is no "linking" for many languages. For example an npm package for nodejs can't be made into a "library" - all the source is "compiled" together at runtime.
      - cyphar 8y ago
        If you can replace an LGPL component with a changed version then you don't need to license any additional code under the LGPL. Shared libraries are just one mechanism for doing this.
- orangeshark 8y ago
  While a license like MIT gives you the most possible freedoms, it also gives the freedom to derivative works to remove all those freedoms. A copyleft license doesn't grant those kind of freedoms so it can try and protect and keep the four essential freedoms[0] in derivatives. I do agree that it is more difficult to commercialize copyleft software. In an ideal world, we would be paying for copyleft software, not giving it away for free. [0]: https://www.gnu.org/philosophy/free-sw.html https://www.gnu.org/philosophy/free-sw.html
stcredzero 8y ago
Copyleft is a niche or a branch, not a dead end. What's a niche to some is a dead end to others. There are plenty of vibrant academic projects that will never, ever see commercial use. From the POV of someone only interested in making a profit, it could well be a dead end.
- cyphar 8y ago
  On the other hand, many academic software packages have absolutely awfully restricted licenses and likely make use of MIT or BSD code. In most cases the licenses were written by the academic themselves. If the majority body of software was GPL then the license choice would've been made for them and academic research would be far more vibrant.
  - stcredzero 8y ago
    I think Copyleft is very well suited for academic research, and for research produced by government institutions.

pauldix 8y ago

Heh, the Star Trek thing might have been a bit of an overreach based on people's reactions, but let me see if I can restate the point I was trying to make with that little bit. Licenses like MIT give the most possible freedom, particularly with how derivative works are commercialized. Copyleft licenses limit the freedoms of derivative works by forcing them to be copyleft (although with some hacks and exceptions). This limitation means that options for creating a derivative commercial product from copyleft sources are very limited. If you are able to get around the copyleft restriction (like SaaS products developed on GPL and AGPL'd software) then you are able to build a commercial product. But if you want to ship that software to a customer in a licensed product or a piece of hardware, you won't be able to do it without exposing your entire source code to be copyleft. I've heard people argue that copyleft is better because it forces this opening of source on everyone. It's a utopian ideal that I think limits its utility, particularly with regard to derivative works.

yarrel 8y ago
Licenses like MIT don't give "the most possible freedom". I am not more free to modify and repair the software running on an iPhone despite the plethora of BSD and other permissive licenses on it than I am on a laptop running GNU/Linux under the GPL. Likewise if you want to commercially exploit some MIT-licensed software that I have modified and am selling binaries of, you are not more free to commercially exploit that than you would be under the GPL. Copyleft doesn't "force" the "opening" of "source" on anyone. It's a pragmatic measure covering existing software in order to ensure recipients of that software are free to use it as they wish. The GPL doesn't limit utility. You are confusing use value and exchange value. The GPL ensures that software retains its utility for its users. All that said, if you want to give all your software to users under the MIT or Apache licenses knock yourself out.
- SpicyLemonZest 8y ago
  The LGPL works as you're describing; I can license my own code however I want, as long as I let people freely modify all the LGPL libraries it links in. The GPL proper is very deliberately viral.
  - robocat 8y ago
    The LGPL is fine for libraries - but that limits where it can be used. I think the LGPL becomes something just like the GPL for scripting languages because there is no "linking" for many languages. For example an npm package for nodejs can't be made into a "library" - all the source is "compiled" together at runtime.
    - cyphar 8y ago
      If you can replace an LGPL component with a changed version then you don't need to license any additional code under the LGPL. Shared libraries are just one mechanism for doing this.
orangeshark 8y ago
While a license like MIT gives you the most possible freedoms, it also gives the freedom to derivative works to remove all those freedoms. A copyleft license doesn't grant those kind of freedoms so it can try and protect and keep the four essential freedoms[0] in derivatives. I do agree that it is more difficult to commercialize copyleft software. In an ideal world, we would be paying for copyleft software, not giving it away for free. [0]: https://www.gnu.org/philosophy/free-sw.html https://www.gnu.org/philosophy/free-sw.html

stcredzero 8y ago

Copyleft is a niche or a branch, not a dead end. What's a niche to some is a dead end to others. There are plenty of vibrant academic projects that will never, ever see commercial use. From the POV of someone only interested in making a profit, it could well be a dead end.

cyphar 8y ago
On the other hand, many academic software packages have absolutely awfully restricted licenses and likely make use of MIT or BSD code. In most cases the licenses were written by the academic themselves. If the majority body of software was GPL then the license choice would've been made for them and academic research would be far more vibrant.
- stcredzero 8y ago
  I think Copyleft is very well suited for academic research, and for research produced by government institutions.