Why is the MIT License unpopular in chess programming?
Posted: Sat Jan 28, 2017 3:44 pm
What follows is just a random rant, I'm not even sure if the topic is affecting me much (I've never done any serious programming and I'm not going to let my well-being depend on doing it anyway).
As per the Wikipedia:
2. Gull (PD in the original version 3, i.e. without Syzygy probing;
LazyGull and its derivatives are under MIT/Expat)
5. IvanHoe (Public Domain)
8. Cheng (zlib-compatible)
12. Arasan (MIT/Expat)
19. Murka (PD)
22. Strelka (PD)
23. Scorpio (BSD-new)
31. Zurichess (BSD-new) [Go]
34. DayDreamer (PD?) [Rust]
35. Donna (MIT/E) [Go]
41. NoraGrace (MIT/E) [C#]
42. GreKo (bare copyright notices in source files, no license file)
43. Floyd (MIT/E)
46. Maverick (bare copyright notices in source files, no license file)
49. Fridolin (a bare copyright notice in readme.txt, no license file)
This list has been made by comparing the CCRL list with the CPW list of GPL engines.
To clarify, I don't regard Crafty's and Delfi's [Pascal] licenses as permissive e.g. because they don't allow to use their derivatives in tournaments under another brand.
If my memory isn't failing me, the rest of the open-source top 50 are GPL.
ChessBrainVB [VBA/VB6], Cyrano, DanaSah, K2, Brutus, Pepito are actually GPL but their authors haven't tagged them as 'gplengines' in the CPW; at times, the license files are well-hidden in subdirectories of their distributions.
____________________________________
Now let me explain why I'm feeling a big difference between permissive licenses and protective ones like the GPL.
The GPL is notorious for requiring any software using a piece of code covered by it to be licensed under it too as a whole. What is meant by 'using' is debatable; a strict (but equivocal) viewpoint is that dynamic (or static) linking to a component covered by the GPL implies that the whole software distribution should be GPL.
Thus the GPL is putting an obstacle to the use of the covered software in a commercial product (though theoretically, it's not a big deal because it's normally legal to have the non-GPL part of a package access the GPL part by exchanging messages through the command line or a pipe), which looks like a strong agenda.
I'm surely benefiting from FOSS as a consumer as it drives the price of commercial software down, by increasing the supply of decent quality software to the end user. However, the GPL as opposed to the MIT makes it more difficult and longer for more motivated (profit-driven) developers to write premium quality products, thus increasing its production cost. Chess has become an exception somewhat because SF has reached the top, but the demand for strong engines playing different styles (e.g. Komodo and Houdini) will still exist.
I suspect that most of the authors of GPL software aren't retired but have regular programming jobs which are supporting them materially and allowing them to write FOSS in their free time with no fear of hunger. And the more they undermine the commercial software industry with their side projects, the worse is their expected future wage (by the boomerang principle) because it comes from their bosses' sales.
The reason why the equilibrium programmer wage may go down (if the GPL retains its strong presence) despite consumer prices not falling is that for-profit developers will be spending too much time reinventing the wheel.
Again, these are just random thoughts, I just shrug shoulders when people are so protective of their hobby projects that a simple acknowledgment of their contribution is not enough for them.
As per the Wikipedia:
Yet, on the CCRL 40/4 list of open-source engines (their best versions only) as of the time of writing, if I'm not mistaken, only 15 out of the top 50 (which is still enough engines to learn from without worrying about being ordered to open the source if the learned ideas are repeated in closed-source code with proper attribution, many thanks to their authors!) are distributed under permissive licenses, namely (please correct me where I'm wrong; the language of the source code is listed in square brackets and only when it's not C/C++):As of 2015, according to Black Duck Software and GitHub, the MIT license was the most popular free software license, with the GNU GPLv2 coming second. In June 2016 an analysis of the Fedora Project's packages revealed the MIT as most used license.
2. Gull (PD in the original version 3, i.e. without Syzygy probing;
LazyGull and its derivatives are under MIT/Expat)
5. IvanHoe (Public Domain)
8. Cheng (zlib-compatible)
12. Arasan (MIT/Expat)
19. Murka (PD)
22. Strelka (PD)
23. Scorpio (BSD-new)
31. Zurichess (BSD-new) [Go]
34. DayDreamer (PD?) [Rust]
35. Donna (MIT/E) [Go]
41. NoraGrace (MIT/E) [C#]
42. GreKo (bare copyright notices in source files, no license file)
43. Floyd (MIT/E)
46. Maverick (bare copyright notices in source files, no license file)
49. Fridolin (a bare copyright notice in readme.txt, no license file)
This list has been made by comparing the CCRL list with the CPW list of GPL engines.
To clarify, I don't regard Crafty's and Delfi's [Pascal] licenses as permissive e.g. because they don't allow to use their derivatives in tournaments under another brand.
If my memory isn't failing me, the rest of the open-source top 50 are GPL.
ChessBrainVB [VBA/VB6], Cyrano, DanaSah, K2, Brutus, Pepito are actually GPL but their authors haven't tagged them as 'gplengines' in the CPW; at times, the license files are well-hidden in subdirectories of their distributions.
____________________________________
Now let me explain why I'm feeling a big difference between permissive licenses and protective ones like the GPL.
The GPL is notorious for requiring any software using a piece of code covered by it to be licensed under it too as a whole. What is meant by 'using' is debatable; a strict (but equivocal) viewpoint is that dynamic (or static) linking to a component covered by the GPL implies that the whole software distribution should be GPL.
Thus the GPL is putting an obstacle to the use of the covered software in a commercial product (though theoretically, it's not a big deal because it's normally legal to have the non-GPL part of a package access the GPL part by exchanging messages through the command line or a pipe), which looks like a strong agenda.
I'm surely benefiting from FOSS as a consumer as it drives the price of commercial software down, by increasing the supply of decent quality software to the end user. However, the GPL as opposed to the MIT makes it more difficult and longer for more motivated (profit-driven) developers to write premium quality products, thus increasing its production cost. Chess has become an exception somewhat because SF has reached the top, but the demand for strong engines playing different styles (e.g. Komodo and Houdini) will still exist.
I suspect that most of the authors of GPL software aren't retired but have regular programming jobs which are supporting them materially and allowing them to write FOSS in their free time with no fear of hunger. And the more they undermine the commercial software industry with their side projects, the worse is their expected future wage (by the boomerang principle) because it comes from their bosses' sales.
The reason why the equilibrium programmer wage may go down (if the GPL retains its strong presence) despite consumer prices not falling is that for-profit developers will be spending too much time reinventing the wheel.
Again, these are just random thoughts, I just shrug shoulders when people are so protective of their hobby projects that a simple acknowledgment of their contribution is not enough for them.