5 ms·
My problem with the GPL philosophy is that I can't find the mechanism that results in more general access to code in the world, when releasing your software as
by epistasis 7y ago
My problem with the GPL philosophy is that I can't find the mechanism that results in more general access to code in the world, when releasing your software as GPL versus BSD/MIT.
What determines what's out there in the world, the decision to release source versus simply doesn't come down to "well I can use this GPL code as part of my project if I also GPL it, so I guess I'll share my code too." The decisions about whether to release code are made entirely independently of what libraries or code to reuse, and there's basically no code out there that's so special and unique that it could flip people one way or the other.
The decision to release code along with the software is made due to other considerations, so claiming that GPL results in more code access seems impossible to defend, to me.
- dTal 7y agoI don't really think any of what you said is actually true. There are many types of software project where the license doesn't matter at all - what matters is that the software exists. Someone is willing to pay to get it made, and they don't really give a hoot about "licensing". I believe this actually constitutes the majority of software - it's just invisible because no one really has much incentive to release it. But I don't really see how, say, a festival's business model is undermined by making their lineup app open source. Next, there really is GPL code that's simply too special to ignore. One obvious example is the Linux operating system. Designing an embedded device? You're almost certainly using Linux. Maybe Linux doesn't do quite what you want - say, support for a given bit of hardware. Guess what, you can't ship your patch without publishing it - and now every Linux user gets to use that hardware, if someone's motivated enough to merge it into mainline. This is a real thing that happens all the time. We're not there yet (and we may never get there if people stop believing in the GPL), but you can imagine that if we get a world where all the important software is GPL, the sheer mass of it will overwhelm any other licensing considerations. "You can either participate in our ecosystem, or you can rebuild modern computing from scratch. Your call."
- epistasis 7y agoIf somebody doesn't care about licensing, then it's trivial to convince them to use open source whether or not they want to tightly integrate with GPL code, so the GPL aspect isn't increasing the amount of open source software. Your hardware driver example is pretty much the only example I can find where somebody is actually incentivized to release code, and willing to go against other concerns; so I appreciate that example! However it's certainly not universal for all hardware, there's plenty of binary blob drivers out there on Linux unfortunately. GPL doesn't need to universally make all software open source for it to be more successful, so I think it's important to look for the concrete overall results. Let's contrast GCC and Clang here. For years, advancements in open source tooling was hindered by GCC's refusal to make improvements that could potentially be used by closed source software. Clang comes along, and all of a sudden an entire open source, tinkerable compiler/tooling/editing environment opens up, precisely because both open source and not GPL. People are willing to pay to improve it because they have the freedom to do what they want with their improvements, including the freedom to hold them back. Despite not having the stick of the GPL, MIT-style license resulted in a better ability to tinker with the code, expand it to more areas, and improve software. I don't see GPL as a crucial aspect for getting to that world where we have access to the important bits. My goal is to enable as much as can be done with software that also has source available. There's a different optimization goal, that says "let's make sure that all software out there has the source available." This is a drastically different goal, however, because it can be achieved by creating less software capability overall, and just eliminating all the growth in capability that originates first in proprietary software but then gets reimplemented as code-available software. IMHO, the GPL seems to be aiming for the "any software out there also has source available" goal.
- mcguire 7y ago20 years before clang, MIT style licenses resulted in a huge number of duplicate, incompatible systems along with a general inability to improve software, to the extent that Windows became the de-facto industry standard OS. There's a reason gcc became the best of breed compiler before llvm came along.
- mcguire 7y agoThe GPL is the one reason why there is one "mainline" Linux kernel and a number of roughly compatible distributions rather than the incompatible chaos of the Unix wars of the '80s and '90s. Take a look at imake and autoconf if you need convincing that the GPL is valuable.
- ndesaulniers 7y agoWell, I hate autoconf, so maybe poor choice of example. :P
- jake_the_third 7y agoI think that's his point.
- mcguire 7y agoYup. :-)
- epistasis 7y agoAre you saying that GPL holds off forks better than BSD/MIT? If so, by what mechanism? Of all the pro-GPL things I've heard, I've never heard that one before, so I'm quite interested to hear this fleshed out into something I can follow!
- mcguire 7y agoThe original license from Bell Labs for Unix wasn't BSD/MIT, but it had essentially the same effect (and in fact, BSD is a down-stream license of it): you could get the source code for Unix and do anything you wanted with it. Most people started with the BSD distribution (because networking), many later incorporated System V things from Bell. They then added their own changes and released the result under their own proprietary license. "They" being Sun Microsystems, IBM, HP, Silicon Graphics, and a host of smaller organizations including NeXT. Each of these had some really neat feature, like IBM's Journaled File System, SiliG's graphics, and Sun's hardware and the fact that they hired Berkeley's CSRG people. But they were all as incompatible as they could be made to be and still be Unixy. They were also business rivals and fought each other constantly. (I was part of IBM's effort with the "Open Software Foundation"'s DCE (https://en.wikipedia.org/wiki/Distributed_Computing_Environment https://en.wikipedia.org/wiki/Distributed_Computing_Environm...), which attempted to fight off Sun's lead with NFS.) This was at a time when Microsoft was pushing Windows on smaller machines and those smaller machines were getting powerful enough to be a threat in the Unix "workstation" market. As a result, Microsoft won the Unix wars. Linux, in contrast, is under the GPL, as are most of its basic user-space programs. You can modify the code to add some feature, but you have to distribute the changes if you distribute the whole thing. As a result, your feature gets incorporated into the mainline kernel, or you fork Linux. Forked Linuxes don't get a lot of traction, though, in part because nobody has a serious hardware advantage, but mostly because the mainline quickly grows a corresponding feature. It's very hard to lock your customers in to your own distribution.