For a GNU dawn! For freedom!
March 18, 2015 8:22 AM   Subscribe

The GNU Manifesto Turns Thirty: Maria Bustillos profiles Richard Stallman, founder of the Free Software Movement and author of the GNU Manifesto, which was published 30 years ago this month: The GNU Manifesto is characteristic of its author—deceptively simple, lucid, explicitly left-leaning, and entirely uncompromising… Stallman was one of the first to grasp that, if commercial entities were going to own the methods and technologies that controlled computers, then computer users would inevitably become beholden to those entities. This has come to pass, and in spades… “With software,” Stallman still frequently observes, “either the users control the program, or the program controls the users.”
posted by Cash4Lead (113 comments total) 28 users marked this as a favorite
 
I used to live in San Francisco, and there was a stretch of pavement (sidewalk) in the Haight-Ashbury that had a spiral carved into the poured concrete and "LOVE GNU!" next to it.

This was before Debian had adopted the spiral motif.
posted by rum-soaked space hobo at 8:40 AM on March 18, 2015 [3 favorites]


I post a link to this short story, The Right to Read, every time Stallman comes up. It was posted in 1997, and say what you will about the man, it looks more prophetic every year.

(Written with emacs - or it would have been, if it was a longer comment.)
posted by RedOrGreen at 8:44 AM on March 18, 2015 [13 favorites]


You'll pardon me if I don't exactly sing his praises. From things like the virality clause and the commercial use poison pill, his 180 on Affero between GPL 2 and 3 (and then going back when push came to shove), the way he bent his own standards with bison (that was the point where I decided that he was full of it), the way he's taken over projects that were turned over to the FSF for safekeeping, his desire to control how people talk about software (read up on the bullshit he's pulled when someone doesn't use "GNU/Linux"), and last (but certainly not least) his history of misogyny...

Yeah, there's a reason I don't have a lot of respect for him.
posted by NoxAeternum at 8:44 AM on March 18, 2015 [7 favorites]


There's a lot to complain about with Stallman. And I've come to think that viral-type licenses are a bad idea. (Particularly for data, like the ODbL). But I have a huge amount of respect for how Stallman changed the very tenor of software development. Back in the 80s and 90s he took this thing programmers were doing, sharing their source without really thinking about it. And he made it into a movement, a specific action with concrete motives and agenda. I'm certain that without Stallman and the FSF the Internet and the Web would be a much poorer place today.
posted by Nelson at 8:50 AM on March 18, 2015 [38 favorites]


Basically what Nelson said. I don't always agree with Stallman but he's managed to change the whole dialog about software in the last three decades.
posted by octothorpe at 8:57 AM on March 18, 2015 [5 favorites]


Re: discussion above:

The man and his views are important regardless of whether you agree fully or are writing software that's compatible with the GPL, imo.

Even if he's not always right he's one of the only public figures that even brings up a lot of the problems with the increasingly proprietary walled garden software world we live in.

And the GPL is a useful tool for the right projects.
posted by unknownmosquito at 8:57 AM on March 18, 2015 [1 favorite]


Stallman is a complicated (and flawed) figure -- this is hardly unique among visionary ideological leaders -- but he pretty consistently tries to do the right thing (as he sees it) and his contribution to computing has been significant.

I work for a non-GNU open-source software project (which is distributed under a BSD-style license) and I prefer the additional freedom we allow our users but I understand the concerns which have motivated the successive GPLs even if I don't agree with the approach they have taken.

[On preview: I agree with Nelson that Stallman deserves a significant amount of credit for helping make free software into a movement and for working to sustain and defend it over the years. And let's please remember that the GNU project isn't just Stallman -- it represents the work and contributions of many accomplished and generous people who have made of their work a gift to mankind.]
posted by Nerd of the North at 8:58 AM on March 18, 2015 [6 favorites]


I post a link to this short story, The Right to Read, every time Stallman comes up. It was posted in 1997, and say what you will about the man, it looks more prophetic every year.

Oh, that's another one - he's an out and out hypocrite. Because the FSF does the exact same thing that he's decrying in that story - but they do it in the name of "freedom", so it's okay.
posted by NoxAeternum at 9:00 AM on March 18, 2015


Copyleft is unfairly maligned by people who have other plans than you do for your work. NoxAeternum can fume all week long about "virality" (a pretty weak attack on the principle of copyleft), but in the end it does one huge massively important thing:

Copyleft codifies the gift economy.

In a world where every kleptocratic corporation out there is trying to take things from you under the guise of a "sharing economy", copyleft ensures that the primary reason you contribute to such a commons remains true. The GPL is the greatest experiment in formalising the terms that a community of gift-givers can feel safe about their contributions. It expresses the kind of compensation that's needed at a minimum to keep the self-sustaining nature of the gift economy going.

When I choose to publish work under the terms of the GPL, it's not because I want to choose how your code is licensed. But you can't use mine without agreeing to my terms. A licence is just a grant of permission (nonsense about click-through contracts notwithstanding), and the GPL says "you're free to use and modify and share this if.." and all of the "if" clauses are about allowing use, modification, and sharing. If I said "you can copy my notes from class, so long as you pass them on when you're done" would that be seen as some kind of Bolshevik plot?

But copyleft is a method, not a goal. Sure, if your goal is to get code spread as widely as possible as quickly as possible, then a lax permissive licence will sometimes be the best fit. If your goal is to contribute code to the ZooptoCorp Spanlytron Project, you may need to use the ZooptoPublicLicence just to get commit merged.
posted by rum-soaked space hobo at 9:03 AM on March 18, 2015 [49 favorites]


rum-soaked space hobo: good point, though I'd quibble about the "use" part. If all someone does is use your software, the GPL shouldn't even be a factor. It's only if they modify and/or redistribute that the rights granted by the GPL come into consideration.
posted by idiopath at 9:12 AM on March 18, 2015 [6 favorites]


There's plenty to criticize about GNU or especially Stallman himself, but the fact remains that without their/his contributions to computing history we would neither have the open source community nor the kind of Internet we have today. So while I'm plenty critical, I'm plenty grateful too.
posted by trackofalljades at 9:15 AM on March 18, 2015 [8 favorites]


"Explicitly left-leaning" is such a weak, vague stand-in for a real political analysis here. I certainly don't really expect more than that from either Bustillos or The New Yorker, especially in a brief thumbnail sketch like this, but still, it'd really be nice if a profile like this could make at least a bit more of a gesture toward the complexity of both the technical and political sides of the subject. It's not exactly uncharted territory at this point.
posted by RogerB at 9:15 AM on March 18, 2015 [4 favorites]


idiopath: that's a good point. Copyright is about terms of distribution. It really only comes into effect when a work is communicated from one person to another. It's just handy to keep repeating "use, modify, and share" over and over again.
posted by rum-soaked space hobo at 9:15 AM on March 18, 2015


NoxAeternum: I'd assume that if the GPL bothers you, regular proprietary code would be even worse? The license is exactly that - a license to do something that would otherwise be illegal, and puts no restriction on the normal usage of software (unlike most proprietary software that comes with absurd and usually illegal / unenforceable EULA boilerplate).
posted by idiopath at 9:15 AM on March 18, 2015 [4 favorites]


Having talked about copyleft, now I'll talk about Stallman: the thing I find most admirable about his communication style is that he refuses to let people conflate issues. This is why people tend to resort to attacking his precision in language when they feel they cannot attack his ideas. It's true that he sticks to this in places where it does not make him appear sympathetic, but it's clear that he feels the issues he's illustrating are so important that they need to be laid out explicitly.
posted by rum-soaked space hobo at 9:18 AM on March 18, 2015 [5 favorites]


In a world where every kleptocratic corporation out there is trying to take things from you under the guise of a "sharing economy", copyleft ensures that the primary reason you contribute to such a commons remains true. The GPL is the greatest experiment in formalising the terms that a community of gift-givers can feel safe about their contributions. It expresses the kind of compensation that's needed at a minimum to keep the self-sustaining nature of the gift economy going.

But that's the whole issue - virality destroys that "self-sustaining nature" by eliminating the key element needed for that - choice. You're not any better than those "kleptocratic corporations" that you decry.

When I choose to publish work under the terms of the GPL, it's not because I want to choose how your code is licensed. But you can't use mine without agreeing to my terms.

Which just so happens to dictate how I have to license my code. Which I wouldn't have a problem with, if it was presented in a more honest fashion. But talking about giving people freedom, then turning around and forcing them to hand over their work as the price of admission - that sticks in my craw some.

If I said "you can copy my notes from class, so long as you pass them on when you're done" would that be seen as some kind of Bolshevik plot?

If that was all you were saying, that would be fine. But you're telling me that not only do I have to pass your notes along, but any addendum I add as well, no matter how much effort that may have taken me. That's the problem.
posted by NoxAeternum at 9:22 AM on March 18, 2015 [2 favorites]


As a commercial developer, I dislike the GPL, because it means I can't just (legally) take GPL code and make money from it. My impression is that this is most common objection to copyleft.

There is a separate, less self-interested discussion to be had about "if the point of code is to be used, then use BSD" but that is a different goal - not a less or more valid one, necessarily.

I was talking with an academic recently: they were in a project, creating a resource. The resource would be freely available at the end of the project, but not for commercial use. So the commercial vendors - who make most of the software that might use it - wouldn't bother using it. Why negotiate another contract? So it won't be picked up and used as much as it might. This is a shame because the resource might, with greater use, supplant its existing commercial competitors, and therefore the Free Software project would be slightly further advanced. But with the copyleft clause, it won't be.

Then again, without the copyleft clause, would it just be taken by the commercial vendors but never develop further? It's impossible to say either way, but I think there's certainly a case that copyleft might restrict Free Software development as much as it enhances it.
posted by alasdair at 9:23 AM on March 18, 2015 [2 favorites]


NoxAeternum: in case you aren't aware, the GPL does not require distribution of derived works. It does require making the source code available under its terms if you distribute the derived work. If you don't like the license, and plan to distribute the code, then you can choose not to make a derived work.
posted by idiopath at 9:26 AM on March 18, 2015 [11 favorites]


Back in the 80s and 90s he took this thing programmers were doing, sharing their source without really thinking about it. And he made it into a movement, a specific action with concrete motives and agenda. I'm certain that without Stallman and the FSF the Internet and the Web would be a much poorer place today.

It's interesting to read Steven Levy's Hackers, with it's coda following Stallman, before Linux, with GNU just a glimmer in his eye, a final rallying cry against the world that left him behind. What a difference a couple of decades make.
posted by zabuni at 9:30 AM on March 18, 2015 [5 favorites]


Labeling the GPL "viral" is a kind of weird blood libel people seem to use to try to undermine it. Nothing about the GPL forces you to give away your intellectual property. Nothing. When you choose to adapt GPL'd software for your own use and then distribute the end result to other people the GPL just asks is that you treat them the way you have been treated. No part of this process is forced upon anyone.

Regarding Stallman's misogyny, I agree that it is disturbing but judging the entire movement based on his personal problems is misguided. Consider for example that one of the founders of the Open Source movement and the most outspoken supporter of "permissive" licensing is this guy.
posted by Poldo at 9:35 AM on March 18, 2015 [15 favorites]


last (but certainly not least) his history of misogyny...

The man is very, very clearly not neurotypical. To be unkind, he's bordering on unabomber-levels of craziness. Here's Stallman's list of things to know if you want him to give a talk:

Streaming the speech:

Streaming is a kind of Internet distribution, so everything in the
previous section applies. In particular, you must use only Ogg format
or Matroska VP8 (Webm).


What? He wants to dictate which codes are used to transmit recordings of his voice?

If you can find a host for me that has a friendly parrot, I will be
very very glad. If you can find someone who has a friendly parrot I
can visit with, that will be nice too.


I'm sure that particular sentence gets trotted out about Stallman all the time, but it's just so apt. No one wants to know this level of detail about his accommodation preferences. He's beyond hyper-specific. But THAT'S OK. He's clearly functional so if he's not neurotypical, geez, let it go. Is he a misogynist? I really don't know, maybe. But maybe he's just completely terrible at normal human interaction.

So when it comes to GNU I'd suggest you focus on the ideas and the license and less on the man because it's lazy and irrelevant.
posted by GuyZero at 9:41 AM on March 18, 2015 [5 favorites]


As a commercial developer, I dislike the GPL, because it means I can't just (legally) take GPL code and make money from it.

That's not true. There are non-commercial licenses out there, but the GPL isn't one of them.

The terms of the GPL might make it impossible to have a successful commercial product, but that's not the same as explicitly banning it.
posted by It's Never Lurgi at 9:43 AM on March 18, 2015 [3 favorites]


For example I don't think the GNU license would prevent a "pay what you want" model as long as 'nothing' was an option, then have the business model be giving priority on bug fixing to reports from people who paid above a certain cutoff, and have commissions where someone can pay to have their desired features added to the code.
posted by Zalzidrax at 9:49 AM on March 18, 2015 [1 favorite]


NoxAeternum: in case you aren't aware, the GPL does not require distribution of derived works. It does require making the source code available under its terms if you distribute the derived work. If you don't like the license, and plan to distribute the code, then you can choose not to make a derived work.

Not for lack of trying, hence that bit about his 180 on Affero. Besides, the derived work argument sort of illustrates the point - if you "give away" something with the clause "but in turn, you have to give away your additional work", I don't see that as giving it away. And then there's the rather...expansive view of what a derivative work is that the FSF has.
posted by NoxAeternum at 9:51 AM on March 18, 2015 [1 favorite]


if you "give away" something with the clause "but in turn, you have to give away your additional work", I don't see that as giving it away

As of now, this falls into "cry me a river" territory for me. They have explicit aims, those aims being the wide distribution of free (as in speech) software, right? If you don't want to stand on their shoulders, don't. They don't want you to make money off of their work in a way they don't approve of. I don't really see that there is room to kvetch about that.
posted by Steely-eyed Missile Man at 9:54 AM on March 18, 2015 [34 favorites]


I mean, if you want to make money, write your own code; don't use theirs. What's not fair about that?
posted by Steely-eyed Missile Man at 9:54 AM on March 18, 2015 [23 favorites]


The Linux kernel and git are GPL'd, so arguing whether the license prevents or even impinges on commercial use or profit is silly. Now you can argue that v3 puts some additional restrictions which have hurt adoption, but the GPLv2 is at this point has proven itself to be pretty much the best choice possible to balance contributor and end user freedoms. To me that is the fundamental difference between the GPL and its permissive alternatives. The GPL works to protect end user freedom (which is why v3 tried to plug the tivo-hole), while permissive licenses work only to protect developer freedom. Neither is "more free" in the way that people like O'Reilly disingenuously claim.
posted by Poldo at 9:56 AM on March 18, 2015 [7 favorites]


The viral concept from the GPL is one of the main reasons I still do respect Stallman, despite him being fairly personally unpleasant.
posted by atoxyl at 10:04 AM on March 18, 2015 [5 favorites]


Labeling the GPL "viral" is a kind of weird blood libel people seem to use to try to undermine it

I'd be grateful for a more neutral term. I agree "viral" sounds negative, but I don't mean it that way myself. The specific property of GPL I'm trying to describe is that a derived work is also subject to the GPL. It's "viral" in the sense of "contagion"; if you choose to use GPL code and then distribute the resultant combined work, then the combined work is required to also be GPL. I wish I could find words to describe this license feature that weren't negative disease metaphors. It's a very clever and unusual idea in intellectual property. I have a lot of respect for it.
posted by Nelson at 10:06 AM on March 18, 2015 [3 favorites]


The GPL is about demanding a certain kind of fair consideration for the value of your work. I think it's fair for the authors of software to have a say in how their work may be used and distributed. It seems uncontroversial (heh) that authors may demand compensation for their work, and to forbid unlicensed redistribution. The GPL is just specifying a different model of compensation. Rather than demanding cash, the GPL only asks that you respect it's contract. In exchange for giving you exceptional freedom compared to the usually restrictive terms of copyright and closed-source, the GPL only asks that if you do redistribute the software, that you do so under it's own terms, so that the value which is created by this software development community, remains available to everyone.

The fact that some developers find it irksome that they cannot profit by enclosing the commons and charging rents upon it, should really just illustrate the great value that has been generated by this community working in the spirit of the public good.
posted by rustcrumb at 10:09 AM on March 18, 2015 [25 favorites]


The only reason the GPL is palatable for any real software (Linux included) is because of the "distribution" requirement. Distribution is pretty much irrelevant for most software these days, because it usually runs on a web server. No problem, then, you can run on Linux using package-distributed GPL'd development libraries and so on, and it doesn't infect your own product because, hey, no distribution is taking place. If I were doing shrinkwrap (not that anybody is, now), or mobile development, I'd definitely use Apache and BSD licensed libraries.
posted by sonic meat machine at 10:12 AM on March 18, 2015 [1 favorite]


The time has come to stand up for the GPL: the Software Freedom Conservancy is suing VMware for violating the GPL.
posted by Bangaioh at 10:12 AM on March 18, 2015 [4 favorites]


And, no, the "rent-seeking" argument is bogus. If I link to, for example, a GPL'd database driver, there is no reason my software (which might involve orders of magnitude more complexity than the database driver) should be required to be under the GPL. It makes it sound like the main driver of commercial software is taking dev libraries, smacking a new version on them, and selling them; which is patently absurd.

Essentially, in RMS-land, there is no economic incentive for the creation of any software beyond hobbyist projects.
posted by sonic meat machine at 10:16 AM on March 18, 2015 [5 favorites]


Just call it what it is Nelson, it's copyleft.
posted by Poldo at 10:19 AM on March 18, 2015


The GPL doesn't go far enough for my taste, not even in v3, and not even in Affero. The rule should be that distribution is mandatory if more than one human person has a copy of the code or is using the code. The idea of corporate "internal use" is a pernicious loophole that unfairly advantages huge organizations and their cloud services. It's a driver for centralization, and that's a bad thing.

But the idea that [on edit -- dynamic] linking creates a derivative work is silly.
posted by Hizonner at 10:32 AM on March 18, 2015 [1 favorite]


sonic meat machine, you seem to wish for the existence of like, an IP version of eminent domain, where if you have a complex enough, profitable enough project, you should just be able to annex the IP of people who stand in your way. If there is really that much at stake, why not just re-implement the functionality that you desire from whatever GPL'd code you covet?

I don't get the complaint about hobbyist stuff. It might be true that there is no economic incentive to create software, but clearly there are other incentives. If there were no incentives at all, how do you explain the existence of the huge, important, and valuable collection of GNU tools? These tools are basically standard kit on most operating systems. Compilers, linkers, debuggers, text editors, diagnostics, networking code, shells ... you can actually build serious, complex software with these tools. You can even sell the software you make with them, as long as you are not incorporating the tools themselves into your product.
posted by rustcrumb at 10:34 AM on March 18, 2015 [7 favorites]


sonic meat machine, you seem to wish for the existence of like, an IP version of eminent domain, where if you have a complex enough, profitable enough project, you should just be able to annex the IP of people who stand in your way.

No, he's pointing out the inherent ridiculousness of saying that the use of a single GPL driver binary turns a program into a "derivative work". Not to mention that doing so is the exact sort of "annexation" that you are complaining about.
posted by NoxAeternum at 10:45 AM on March 18, 2015 [1 favorite]


How can that be annexation? If you don't like the terms, you're free not to use the code.
posted by eruonna at 10:52 AM on March 18, 2015 [7 favorites]


No, he's pointing out the inherent ridiculousness of saying that the use of a single GPL driver binary turns a program into a "derivative work". Not to mention that doing so is the exact sort of "annexation" that you are complaining about.

Which was the idea behind the original GLPL, but that has mostly been dropped. I'm not sure if it was mostly ideology that caused it to be dropped or if it was just impossible to define in a legally coherent way. Or both.
posted by It's Never Lurgi at 10:53 AM on March 18, 2015


The authors of the GPL'd work are not verbing anything. They have verbed, and the result is the GPL'd item in question. Now, you are making your own thing, and your choice is clear: you can build GPL'd code into your project, or you can build your own version of the GPL'd thing to use. The consequences are clear: if you choose to use the GPL'd thing, your thing will have to be GPL'd.

It sounds like you are definitely unhappy with the idea of making your project GPL'd itself, but you also recognize that the GPL'd thing you want is very valuable. So it sounds to me that you wish that you could seize it, against the wishes if it's authors, harnessing the value of what they have created without any compensation to them. Is that not so? If not, what do you want?
posted by rustcrumb at 10:55 AM on March 18, 2015 [19 favorites]


Poldo: the term "copyleft" doesn't mean anything to outsiders. I'm looking for a word like "viral" that means something to people not in the culture. (Also the term "copyleft" is a bit confusing, since it's not so much an alternative to copyright as it is how one licenses a copyrighted work.)

My complaint with the viral licenses is they just don't work very well in practice, particularly for commercial use. The GPL is very complicated to explain to lawyers. And it's very unclear what "derived work" means; the GPLv2 makes little sense in the world of dynamically linked code, much less embedded script interpreters. And then there's the huge loophole that many companies don't distributed derived works at all, but simply use them to provide online services. There've been many attempts to patch this up – LGPL, GPLv3, Affero – but none of have been very successful.

The biggest success of the GPL, arguably, is the Linux kernel. Even that includes an oddball exception for binary device drivers. Also it's notable that Linux is working in a very nebulous world without copyright assignment; there's some 1000+ copyright holders for the Linux source. It's very different from the FSF projects that require assigning copyright to the FSF.

There are also many smaller GPL successes. One of my favorites is the Linksys router firmware. It took some doing, but Linksys was compelled to release the source because it included GPL products and that's spawned a long and productive world of router firmware like DD-WRT, OpenWRT, and Tomato.

I've moved to MIT or BSD licenses for my own projects. I want people to use my free software, even in proprietary and/or commercial closed source derivative products. I trust the open source community in general rewards giving back to the community enough that I don't feel the need to bind users of my software with the GPL. But it's nice to have a choice, and a framework to even have a discussion about license choices. That's what Stallman gave us.
posted by Nelson at 10:57 AM on March 18, 2015 [6 favorites]


It sounds like you are definitely unhappy with the idea of making your project GPL'd itself, but you also recognize that the GPL'd thing you want is very valuable. So it sounds to me that you wish that you could seize it, against the wishes if it's authors, harnessing the value of what they have created without any compensation to them. Is that not so? If not, what do you want?

So, how would you feel if a company said "yes, we'll let you use our code in your project, but in turn, you have to give us all of your work in turn"? Because that's fundamentally what the GPL asks for. And I find it to be a tad hypocritical for a group that talks a big deal about freedom to basically say "if you want to use our stuff, you have to give us your work."

In short, you're really no different than the companies you decry.
posted by NoxAeternum at 11:24 AM on March 18, 2015


To think in terms of "derived work" in the context of discussing GPL is flat-out wrong. That would be like saying a research paper that cites something becomes a "derivative" of that. It's just totally wrong diction in describing the kind of relation going on.

I get that software users may not share Stallman's political values. If one has libertarian tendencies I can see how one wouldn't agree with his approach. But attacking him instead of understanding these differences is just a way to fool oneself.
posted by polymodus at 11:27 AM on March 18, 2015


[Friendly reminder, this needs to not get personal. Thanks.]
posted by LobsterMitten (staff) at 11:28 AM on March 18, 2015 [2 favorites]


Which was the idea behind the original LGPL, but that has mostly been dropped. I'm not sure if it was mostly ideology that caused it to be dropped or if it was just impossible to define in a legally coherent way. Or both.

Oh, I'm pretty sure that it was ideology, considering that a) he wrote a lengthy ideological piece about why it should be dropped and b) he redefined what the acronym stood for, from Library GPL to Lesser GPL.
posted by NoxAeternum at 11:36 AM on March 18, 2015 [1 favorite]


I find complaints about the GPL really tedious and twenty years out of date, because the correct response is to implement a "better"-licensed alternative and give it away, just like the people writing GPL software did. For example, I respect Apple a lot for pushing the state of the art with LLVM+Clang almost entirely because they don't like that GCC is under the GPL. Talk is cheap.
posted by vanar sena at 11:39 AM on March 18, 2015 [13 favorites]


Furthermore, the "you have to surrender your own work if you want to use ours" system has a massive entrenching effect for the established players as well, which is something else that doesn't get talked about much.
posted by NoxAeternum at 11:40 AM on March 18, 2015 [1 favorite]


"My complaint with the viral licenses is they just don't work very well in practice, particularly for commercial use."

I must be imagining my paychecks....

"The GPL is very complicated to explain to lawyers. And it's very unclear what "derived work" means"

It's a standard term from copyright law. If your lawyers don't understand it, you're talking to the wrong lawyers. It's fuzzy, sure, but that's kind of inherent to the idea. There's nothing GPL-specific there. Though of course you have to think about it less if you're using more liberally licensed code.

"The biggest success of the GPL, arguably, is the Linux kernel. Even that includes an oddball exception for binary device drivers."

No, it's plain GPLv2. The only extra language there is about programs running on the linux kernel, which is more of a clarification than an oddball exception.
posted by bfields at 11:48 AM on March 18, 2015 [5 favorites]


So, how would you feel if a company said "yes, we'll let you use our code in your project, but in turn, you have to give us all of your work in turn"
Depending on the circumstances, I might be devastated. Or elated. Or miffed. It depends on what is on offer, and what I think I can make with it. But regardless of my emotions, I would have to accept that the company has the right to make that proposal, just as I have the right to take it or leave it, just as the authors of GPL'd code have the right to promulgate the GPL, whether it works for your purposes or not.
posted by rustcrumb at 11:53 AM on March 18, 2015 [3 favorites]


Really quick: whether or not Richard Stallman is neurotypical has nothing to do with misogyny, and would certainly not be an excuse for it. There are plenty of non-neurotypical feminists.
posted by en forme de poire at 12:06 PM on March 18, 2015 [7 favorites]


So, how would you feel if a company said "yes, we'll let you use our code in your project, but in turn, you have to give us all of your work in turn"? Because that's fundamentally what the GPL asks for.

I am no expert, but this doesn't seem like an accurate comparison. It seems more like if a company gave something away for free, then told you, "You can incorporate this into your project, but if you do, you have to give your project away for free, too," which seems quite fair to me. The FSF isn't making money off of your work. It seems a little disingenuous to conflate what they are asking for with a for-profit entity taking your work to make themselves more money, which is the clear implication of the comparison.
posted by Steely-eyed Missile Man at 12:07 PM on March 18, 2015 [4 favorites]


So far as I know, nothing about the GPL precludes the author of a work from licensing it to you (specifically) under an alternative commercial license for a fee. Thus, the GPL stands in the way of commercial anything exactly not at all.
posted by emilyw at 12:09 PM on March 18, 2015 [1 favorite]


The FSF isn't making money off of your work. It seems a little disingenuous to conflate what they are asking for with a for-profit entity taking your work to make themselves more money, which is the clear implication of the comparison.

Why? I find the argument that they should get a free pass on asking people to surrender their own work because they aren't looking for money to be the disingenuous one, myself.
posted by NoxAeternum at 12:18 PM on March 18, 2015


I know that "derived work" is a standard term in copyright law and I've talked to some very good intellectual property lawyers working specifically on software. It's unclear exactly what a "derived work" means in the context of software and there's precious little case law to settle the question.

If I statically link a GPLv2 work to my proprietary code, am I required to follow the GPL for the derived work? Yes. If I dynamically link it, though, I'm not. FSF cleared that up for us about 15 years ago, but it's a little confusing as to exactly why. Now what if I write a proprietary binary device driver and link it to my GPL kernel? It's not clear, although the Linux kernel has made a specific exemption for that case. What if I distribute a bundle of proprietary Javascript running in a GPL browser? How about GPL Javascript running in a proprietary browser? Does that change if the Javascript is dynamically loaded from a third source instead of distributed together with the browser? How about if the Javascript VM is itself a dynamic library?

These are not theoretical questions; these are real live questions in real live software I use daily. They are mostly ignored by the community because most projects are small, or non-commercial, or people play fast and loose. But if you are building a for-profit company using GPL software you have to answer these questions very carefully, and the answers are not clear. (Also the FSF has not done a great job clarifying them; I think they prefer the ambiguity.) To some extent all software licenses have to grapple with this question of "derived work"; it's the foundation of copyright, after all. But the GPL is trying to do something clever in imposing license terms on the derived work that are much more complex than other open source licenses. That's why these complicated questions matter.

Again, none of this is meant to distract from what Stallman has accomplished with the FSF. I think it's been hugely influential in a positive way. I also have no principled objection to the added requirements of the GPL license on derived work. However I do find copyleft-type licenses are often very awkward to apply in practice, particularly in commercial contexts. Which is why I've chosen simpler open source licenses for my own work.
posted by Nelson at 12:20 PM on March 18, 2015 [2 favorites]


Part of my thinking on copyleft-style licenses has been about the ODbL and its application to OpenStreetMap. It's a bit of a derail, but the ODbL is basically an attempt to apply copyleft to data. Here's a good blog post about the problems of the ODbL, the related lawyer's white paper is also worth reading. The OSMF has tried to clarify these questions with some community guidelines, which helps, but still leads a lot of confusion and also prevents a lot of uses which I think would be beneficial to OSM's goals.
posted by Nelson at 12:20 PM on March 18, 2015


Why?

Most people find clear inherent differences between for-profit enterprises and entities which choose not to charge a monetary fee for their products or services, stemming mainly but not always entirely from the difference between motivations: self-interest vs. an interest in the common good.

I find the argument that they should get a free pass on asking people to surrender their own work because they aren't looking for money to be the disingenuous one, myself.

But why, exactly? Your position is clear, but it has not been explained. You seem to be the only one in this thread who finds the idea of an organization giving away software and requiring you to do the same if you incorporate their software into yours some kind of contradiction, and it's not clear why. Is it a contradiction if a for-profit entity creates a piece of software and requires you to pay them a licensing fee to incorporate it into your own work? The "freedom" of your own software is the licensing fee, here. They are forgoing the reaping of profit from their software. Their licensing fee for incorporating their work into yours is that you do the same. What is disingenuous about that? You are invoking "freedom" quite a bit, here, but it in all honesty seems to mean to you (in this circumstance) the freedom to do whatever you want, which is not usually how society works.
posted by Steely-eyed Missile Man at 12:26 PM on March 18, 2015 [14 favorites]


So far as I know, nothing about the GPL precludes the author of a work from licensing it to you (specifically) under an alternative commercial license for a fee. Thus, the GPL stands in the way of commercial anything exactly not at all.

This only works in the single hop scenario - if I am dealing with a multiple hop situation (that is, I want to use a piece of GPL'd code that it itself uses another piece of GPL'd code), then even if I negotiate a license with the first party, I'm still incorporating GPL code into my work and could be very well bound by it.
posted by NoxAeternum at 12:29 PM on March 18, 2015


FSF & Friends: "Hey, we have this big body of code that we've generated. It does useful stuff, you can modify however you like, add features you want that maybe we haven't thought of, take out stuff you don't like, use it for whatever you like. It's a gift economy, though -- you have to live by the same standards we created it by, and that means giving back/paying forward your work that's based on it, just like we're doing with you."

NoxAeternum & Friends: "Hypocrites! You're no better than the kleptocrats running Halliburton!"
posted by weston at 12:32 PM on March 18, 2015 [23 favorites]


Could we maybe just give a rest to the "free software is stealing other people's work" tangent? It seems like basically the intellectual-property version of the Rand Paul "public health care is slavery for doctors" talking point in the first place, trying to characterize a public good as the same thing as a private "taking" — but in any case it's surely been restated enough times here already.
posted by RogerB at 12:37 PM on March 18, 2015 [13 favorites]


But why, exactly? Your position is clear, but it has not been explained. You seem to be the only one in this thread who finds the idea of an organization giving away software and requiring you to do the same if you incorporate their software into yours some kind of contradiction, and it's not clear why. Is it a contradiction if a for-profit entity creates a piece of software and requires you to pay them a licensing fee to incorporate it into your own work? The "freedom" of your own software is the licensing fee, here. They are forgoing the reaping of profit from their software. Their licensing fee for incorporating their work into yours is that you do the same. What is disingenuous about that? You are invoking "freedom" quite a bit, here, but it in all honesty seems to mean to you (in this circumstance) the freedom to do whatever you want, which is not usually how society works.

That's a great argument for openness - that the code should remain available to all. But it's not really an argument for freedom - for people to use the code as they please.
posted by NoxAeternum at 12:40 PM on March 18, 2015 [1 favorite]


When discussing the GPL's "virality" I find it helpful to turn the tables and talk about how one might use code that's not licensed under the GPL. Say I'm writing a piece of software and there's a library available that would make my job much easier, but the author wants me to pay $5 per copy of my software that uses that library. I want to give my software away for free, so I'd end up having to pay the author of the library $5 for each user I had.

To me, that wouldn't make sense and wouldn't be worth it. So I'd have to either rewrite the functionality of that library myself or do without it. I don't think anyone would claim it was somehow unfair or greedy of the author of the library to want to charge for their software, and if I claimed that I should be allowed to use it for free without its price "infecting" my application just because I wanted to, people would laugh at me.

What's the difference between the condition of "charging $5 to use this" vs. "open your own code to use this"? Either way you are free to reject the terms and not use the library and rewrite it yourself or do without it. The author gets to set the terms and you can abide by them or not use the library as you wish. Claiming you get to do anything else just because you want it a way it's not being offered is simply childish petulance.
posted by fader at 12:41 PM on March 18, 2015 [10 favorites]


"If I statically link a GPLv2 work to my proprietary code, am I required to follow the GPL for the derived work? Yes. If I dynamically link it, though, I'm not. FSF cleared that up for us about 15 years ago"

That's not true.

"Now what if I write a proprietary binary device driver and link it to my GPL kernel? It's not clear, although the Linux kernel has made a specific exemption for that case."

Again, no. Linus:
There is NOTHING in the kernel license that allows modules to be non-GPL'd. The _only_ thing that allows for non-GPL modules is copyright law, and in particular the "derived work" issue. A vendor who distributes non-GPL modules is _not_ protected by the module interface per se, and should feel very confident that they can show in a court of law that the code is not derived.
A possibly useful article.
posted by bfields at 12:54 PM on March 18, 2015


To me, that wouldn't make sense and wouldn't be worth it. So I'd have to either rewrite the functionality of that library myself or do without it. I don't think anyone would claim it was somehow unfair or greedy of the author of the library to want to charge for their software, and if I claimed that I should be allowed to use it for free without its price "infecting" my application just because I wanted to, people would laugh at me.

Actually, I'd be willing to bet that there's a number of people who would argue that the company demanding such a license is unfair to you, especially since you are looking to give your program away for free.

What's the difference between the condition of "charging $5 to use this" vs. "open your own code to use this"? Either way you are free to reject the terms and not use the library and rewrite it yourself or do without it. The author gets to set the terms and you can abide by them or not use the library as you wish. Claiming you get to do anything else just because you want it a way it's not being offered is simply childish petulance.

But that's my point - there really isn't any difference between the two. It's just a difference in the terms people are asking for.
posted by NoxAeternum at 12:55 PM on March 18, 2015


That's a great argument for openness - that the code should remain available to all. But it's not really an argument for freedom - for people to use the code as they please.

You're only considering two parties to this transaction: the original author of the code and another developer who wants to make use of it. The "freedom" component comes in with the third party: the person who ends up using the derived work.

The GPL requires that the end user of the program must have the same freedom to examine the code and modify it that everyone else along the chain did. You are obligated to share your code, if you make use of GPL code, precisely so that you cannot deprive *your* user of the same freedom you have: to examine and modify the code they are running.
posted by Mars Saxman at 12:58 PM on March 18, 2015 [16 favorites]


there really isn't any difference between the two

There certainly is a difference, and for myself (and I would guess most people) it is a salient one: the license fee of fader's example stands in the way of the common good (more free software) in favor of a singular entity's good, whereas the GPL stands in the way of a singular entity's good (the developer/company that wishes to singularly profit from the work of others) in favor of the common good. The situations are not equivalent.
posted by Steely-eyed Missile Man at 1:09 PM on March 18, 2015 [14 favorites]


You're only considering two parties to this transaction: the original author of the code and another developer who wants to make use of it. The "freedom" component comes in with the third party: the person who ends up using the derived work.

The GPL requires that the end user of the program must have the same freedom to examine the code and modify it that everyone else along the chain did. You are obligated to share your code, if you make use of GPL code, precisely so that you cannot deprive *your* user of the same freedom you have: to examine and modify the code they are running.


Once again, this is an argument for openness, not really freedom per se. While it's being couched in terms of freedom, said freedom is derived from the openness of the code. Not only that, but in order to preserve this one specific freedom, many others have to be curtailed, which is why arguing that this is about freedom becomes problematic.
posted by NoxAeternum at 1:11 PM on March 18, 2015


I think it's a 'freedom to use' vs 'freedom to take' distinction. GPL stuff is free to use, modify, whatever, but ultimately it is not yours - if you choose to make your modifications available, you don't get to choose not to give them to certain people because the thing you modified never was yours, it was always everybody's. Whereas with a more permissive license you get to take a copy and keep it as yours and yours alone.
posted by Zalzidrax at 1:17 PM on March 18, 2015 [4 favorites]


Once again, this is an argument for openness, not really freedom per se.

I am not really seeing the distinction. Would this issue be resolved for you if they were called the Open Software Foundation instead?
posted by Steely-eyed Missile Man at 1:17 PM on March 18, 2015


Or is this another free-as-in-speech-not-free-as-in-beer problem of definitions? It seems almost impossible that we could have gotten this far into this conversation (in 2015, no less) for that to be the case, however.
posted by Steely-eyed Missile Man at 1:19 PM on March 18, 2015


There certainly is a difference, and for myself (and I would guess most people) it is a salient one: the license fee of fader's example stands in the way of the common good (more free software) in favor of a singular entity's good, whereas the GPL stands in the way of a singular entity's good (the developer/company that wishes to singularly profit from the work of others) in favor of the common good. The situations are not equivalent.

You'll pardon me if I tend to find "for the common good" arguments to be less than persuasive, as they more often than not are used against the powerless as they are against the powerful.
posted by NoxAeternum at 1:19 PM on March 18, 2015


You'll pardon me if I tend to find "for the common good" arguments to be less than persuasive, as they more often than not are used against the powerless as they are against the powerful.

I will pardon you if you can come up with specific reasons for that objection in this case rather than an extremely vague attempt at inciting guilt by association. Can you explain how, exactly, the GPL has been used against the "powerless"? To forestall pointers back to your earlier comments, I certainly do not (and will not) see developers as "powerless" in this circumstance, as they definitely do possess the power to write their own code if they do not like the GPL's terms.
posted by Steely-eyed Missile Man at 1:22 PM on March 18, 2015 [12 favorites]


"If I statically link a GPLv2 work to my proprietary code, am I required to follow the GPL for the derived work? Yes. If I dynamically link it, though, I'm not. FSF cleared that up for us about 15 years ago"

That's not true.


Yeah, dynamically linking against GPL .so's or .dll's still puts you under GPL. But it's easy to think of much murkier situations: for example, lets say you write a (GPL) web service that exposes some other GPL library to the web through http requests. Does any software that makes use of that web service fall under GPL as well? Or, lets say you write a proprietary piece of software and a shell script to pipe between that software and the GPL software: what's covered by GPL, the shell script, the proprietary software, both, or neither?

The GPL faq is not especially definitive on the matter (part of that cultivated ambiguity that Nelson mentions):
By contrast, pipes, sockets and command-line arguments are communication mechanisms normally used between two separate programs. So when they are used for communication, the modules normally are separate programs. But if the semantics of the communication are intimate enough, exchanging complex internal data structures, that too could be a basis to consider the two parts as combined into a larger program.
posted by Pyry at 1:25 PM on March 18, 2015 [1 favorite]


[NoxAeternum and Steely-eyed Missile Man, at this point, the back-and-forth between the two of you is kind of sucking the air out of the room, maybe take a step back for a little while?]
posted by LobsterMitten (staff) at 1:28 PM on March 18, 2015


If one has libertarian tendencies I can see how one wouldn't agree with his approach.

The agenda of the FSF is decidedly libertarian in the actual meaning of the word. Funny how that doesn't mesh with the so called libertarian US far right.
posted by uffda at 1:29 PM on March 18, 2015 [8 favorites]


I find characterizing a professional developer as a victim of the GPL amusing. How does this play out, exactly? Worst case you don't like the terms, and you proceed as if the software didn't exist. If you don't use an open source OS, the ecosystem will surely provide closed source tools that you can use for some price.
posted by idiopath at 1:30 PM on March 18, 2015 [12 favorites]


At best what you're arguing is that the RMS definition of "free" is arbitrary and idiosyncratic. Which it is - terminology is a weird cul-de-sac he's infamously stuck in. But I really don't see mustering the existing force of IP law in favor of openness as hypocritical for him and regardless it's something I think is a fundamentally sound idea - agreeing that there are some ambiguities that are important to work out - if that's your cause.

You'll pardon me if I tend to find "for the common good" arguments to be less than persuasive, as they more often than not are used against the powerless as they are against the powerful.,

do you have a problem with the idea that permission to use a pool of common resources might only be granted with the requirement that you contribute to that pool?
posted by atoxyl at 1:31 PM on March 18, 2015 [1 favorite]


bfields: "No, it's plain GPLv2. The only extra language there is about programs running on the linux kernel, which is more of a clarification than an oddball exception."

well, it's plain GPL v2, but Linus's interpretation is that a class of binary device drivers doesn't constitute a derivative work of the kernel, and thus doesn't come under the GPL umbrella. I think the distinction boils down to if the driver is written in isolation from the kernel, and then a shim is written to allow it to hook into linux (and that shim is then GPLed). Whether this interpretation is valid is subject to some contention.

I think this "freedom" argument boils down to freedom for whom. the GPL trades off developer freedom (to include a library/code in a proprietary derived work) for end user freedom ( to be able to freely modify derived works). BSD/MIT/Apache licenses go the other way: developers are freer, but end users can be constrained.

I can see how it might seem that the FSF has redefined freedom to fit their particular tradeoffs, but it seems to be that it is in fact a form of freedom. And this is coming from someone who rarely licenses stuff in GPL and is far from a huge fan of RMS.
posted by grandsham at 1:32 PM on March 18, 2015 [1 favorite]


I think the discussions about contributing back to a commons misses the central point of the manifesto.

A nice summary from HN user torrent-of-ions:

It's not really about "giving back". There's no obligation to give anything back at all and neither should there be. Rather, the GPL is designed to globally maximise the freedom to use, modify and distribute a piece of software. The only freedom the GPL does not provide is the freedom to take these freedoms away from future users.
posted by uffda at 1:41 PM on March 18, 2015 [12 favorites]


I think the discussions about contributing back to a commons misses the central point of the manifesto.

I maybe didn't choose the ideal phrasing for my comment - certainly you're not obligated to give back as an end user of GPL software. But if you wish to build directly on code from the GPL commons you do have to give back in effect - which is intentionally an alternative framing which appeals to my personal politics and which I thought might cut through the argument about what "freedom" means.

For the anti-GPL contingent, what's the angle here? If you are in favor of significant copyright reform, do you have a case to make that the GPL works against that in practice? If you are fine with the current system, then what's your problem with developers choosing these particular terms? You could argue that GPL code doesn't help small-time devs make a living as much as unrestricted open code does, but it's still more helpful than proprietary code and you still do get to use all the tools that happen to be GPLed. If your quarrel is with the exact boundaries of what counts as using the software versus making a derivative work that's fair enough but is that an unavoidable problem with viral copyleft or just with FSF?
posted by atoxyl at 1:59 PM on March 18, 2015


That's a great argument for openness - that the code should remain available to all. But it's not really an argument for freedom - for people to use the code as they please.

You are reminding me of any number of arguments I've had about anarchism, where I've laid out explanations of the kinds of societies and self-governing structures and governance anarchists have for almost two hundred years proposed and tried to implement only to have the other person insist that since those proposed structures and societies are anything other than Mad Max war of all-against-all, they're not really anarchy and don't count.
posted by Pope Guilty at 2:12 PM on March 18, 2015 [10 favorites]


Also AFAIK the earliest use of copyleft seems to be the Principia Discordia, which doesn't use the term "copyleft" but does declare itself to be "Ⓚ ALL RIGHTS REVERSED - Reprint what you like". Historia Discordia has an entry on the subject, and it comes up in Greg Hill's interview of himself that's often appended to the Principia, reproduced here.
posted by Pope Guilty at 2:36 PM on March 18, 2015 [4 favorites]


But that's the whole issue - virality destroys that "self-sustaining nature" by eliminating the key element needed for that - choice. You're not any better than those "kleptocratic corporations" that you decry.

Someone on Hacker News responded to this article by claiming that less restrictive licenses were more free than the GPL variants because they gave future developers more freedom.

In general, freedom as a whole is increased when certain freedoms are removed. Giving future developers the choice/freedom to make things un-free in the future is not actually a gain in freedom; freedom is better maintained when that option is simply not there.

(There are lots of analogical cases to this. It's better to have mandatory maternity and paternity leave than optional, so that people can't be coerced into not taking it. It's better not to have the freedom to sell oneself into slavery. There are all sorts of spurious "freedoms".)

(ETA I see uffda made this point above, quoting another HN user.)
posted by kenko at 2:36 PM on March 18, 2015 [13 favorites]


I had the distinct pleasure of interviewing Stallman when he visited Iceland some years ago. He was an altogether relaxed, easy-going guy (who apparently has a penchant for those Danish open-faced sandwiches). He was exceedingly patient in talking with this Linux noob, answering questions he may have answered a gajillion times before without getting flustered. When I told him I was using Linux Mint, no, he did not flip the table and bellow at me about proprietary graphic cards or wireless drivers. Instead, he pointed out that the GNU site has a list of Free distros I should check out. He was quite possibly the most pleasant movement leader (a term he'd doubtlessly disagree with) I've interviewed.

At one point during the interview, some German tourist walked up to our table, clearly star-struck, and stammered, "Excuse me but ... are you Richard Stallman?" Stallman nodded. The German nerdgasmed, giggling nervously, and said he'd seen Stallman speak some years ago. They shook hands and the German practically floated away on a cloud. It was actually pretty endearing, not least of all for how Stallman downplayed the whole thing.

I suppose this is only tangential to any discussion on the GNU Manifesto; I just wanted to share that my brief experience with the guy gave me the impression of someone living pretty consistently by their principles. More than I can say for a lot of prominent activists.
posted by Aya Hirano on the Astral Plane at 2:53 PM on March 18, 2015 [18 favorites]


Regarding the spiral in San Francisco: I always liked to imagine it was carved there by John Perry Barlow a quarter-century ago. I sometimes wonder if it's still there.
posted by rum-soaked space hobo at 3:14 PM on March 18, 2015


Poldo: "Labeling the GPL "viral" is a kind of weird blood libel people seem to use to try to undermine it."

I agree what you wrote, but please don't use the pejorative term "blood libel" this way.
posted by double block and bleed at 3:46 PM on March 18, 2015 [8 favorites]


NoxAeternum: I do not understand why you think the GPL removes choice. You still have the choice not to contribute to this free software given to you. We all still have the choice to use the GPL or some other licence. From your comments, it's hard to believe you've really thought this through very well, although I trust that you firmly believe the things that you are saying.

It's bizarre that the people who shout the loudest about "freedom of choice" in the "open source" licence debates tend to be the ones trying to argue against our choice of the GPL. I honestly can't internalise or comprehend it. It's not for lack of trying, but the perspective is alien to me.

At least the large business community around software released under the GPL has largely snuffed out the old "anti-commercial" canard. The GPL explicitly preserves your right to engage in commerce with the work.

It just makes it less trivial to obtain unfair advantage in that market. If anything, that feels more pro-commerce!
posted by rum-soaked space hobo at 4:03 PM on March 18, 2015 [1 favorite]


bfields, thanks for correcting me on dynamic linking and the GPL's interpretation of derived work. You're right, I was wrong. I was getting the GPL confused with the LGPL, which does make a distinction between static and dynamic linking. Not by redefining what "derived work" means, but rather explicitly allowing a certain kind of derived work.

I'm more confused than ever on the situation with closed source kernel modules being loaded in to the Linux kernel. I was thinking of Linus' 1995 view of the question. You've quoted Linus' 2002 rethinking of the question. This 2003 Linus email further muddies the waters, where he explicitly refers to kernel modules as a "grey area". I think that's in support of my larger point which is that the application of "derived work" to software is complicated, and the GPL relies heavily on that concept being applied correctly. (All of the sources I have found are years old and non-authoritative. Has the question about what the GPL means for Linux kernel modules been settled once and for all? It's not clear who can even speak authoritatively for Linux, since the kernel's copyright ownership is held among hundreds of people.)
posted by Nelson at 4:08 PM on March 18, 2015


When discussing the GPL's "virality" I find it helpful to turn the tables and talk about how one might use code that's not licensed under the GPL.

Oh, you mean like the MIT or BSD license instead? It's very easy to use that kind of open source code and using it carries very few requirements on the derived work; just some notifications to the user.

I totally agree that it's weird to talk about GPL code somehow impinging on my freedom. I still have the choice to simply not use the GPL code. But MIT and BSD licensed code gives me even more freedom of choice in what I do with the derived work than the GPL does. I have the freedom to choose whether to publish my derived source or not. Stallman's argument is that a BSD/MIT ecosystem leads to less Freedom overall, because it allows proprietary derived works from open source origins. But this is a very precise definition of Freedom and a very idealistic one.
posted by Nelson at 4:11 PM on March 18, 2015 [1 favorite]


Nelson: And yet when I choose a licence for my code, I think about what's best for me. That may not be best for you, and you may wish to negotiate other terms. You can always contact the copyright holder for a separate grant of permission!

But the original point you quoted seems to be arguing that it's unlikely that many proprietary licences are going to be in Nelson's best interest. That's just not how people draft these things.
posted by rum-soaked space hobo at 4:20 PM on March 18, 2015


Right, this isn't complicated. When I look for a library to use I have several choices. I can choose a MIT/BSD licensed library and do pretty much whatever I want with the code. I can choose a GPL licensed library and accept the requirement I distribute my own source code along with the derived work. Or I can use some proprietary library with some weird license that's likely to be very restrictive to me. Of the three I prefer using MIT/BSD first, then GPL, and proprietary a distant last.

The argument for the GPL being more Free is that it induces an ecosystem where everyone is compelled to share source code. That it mandates a sort of gift economy by the viral license. It's a very clever idea. I also think it hasn't worked so well in practice. As grateful as I am to Stallman and the FSF I'm also grateful to the BSD/MIT licenses for giving an alternate path. Also glad for the general culture of open source where sharing source is now just the reflexive thing a lot of people do in the culture do, no matter what the case. I mean seriously; if a Unix project is not on GitHub how seriously do you take it?

A lot of this sharing of code is enabled because the whole industry and practice is moving so fast. Success is about getting mindshare and an open source release is the best way to colonize that mindshare.
posted by Nelson at 4:42 PM on March 18, 2015 [1 favorite]


And let's please remember that the GNU project isn't just Stallman -- it represents the work and contributions of many accomplished and generous people who have made of their work a gift to mankind.

Speaking of others, amongst those in the public eye I like Eben Moglen not only for his legal work for the FSF, but for his oratory tying free software into freedom more generally: Having brought into being the tools of our liberation, it is now our privilege to use them to change the world around us. This is our special role in the long history of the struggle for freedom of thought.
posted by one weird trick at 5:05 PM on March 18, 2015 [7 favorites]


(Ah, here's the video)
posted by one weird trick at 5:17 PM on March 18, 2015


The thing is, I like free software; I create and contribute to it. I also use the MIT license (for my own projects) and prefer MIT/BSD/Apache-licensed projects when I contribute to others' projects. I just think the GPL is an ideological tool rather than a good thing for software, which I do not think has served very well. It's also dangerous, because it has been used as a weapon in the past, with people attempting to revoke the dual license of components of OpenBSD in favor of a GPL license, which would have caused all kinds of gnashing of teeth as the core components of OpenBSD linked to those components (effectively meaning that component would've needed to be excised). Fun times!

Keep this in mind: the FSF has delivered on none of its promises. It never created a completely free operating system (the Hurd), and at present the only userland tool that's GPL and "state of the art" is Emacs. (Vim, its primary competitor, is not GPL'd and has its own license exhorting users to donate to charity.) GCC has been superseded in many respects by the LLVM project (which Apple supported and which uses the University of Illinois license). The GIMP is the only other major userland tool which is currently in wide use, as far as I know.

When I said earlier that there would be no economic incentive for the creation of complex software, this is what I meant. Without economic incentive, you wouldn't have software to do your taxes; you wouldn't have software to run your car; you wouldn't have the software that makes medical imaging possible. All of this is "proprietary" software, and is against the ethos of Stallman, et al., and I also happen to think that it is generally good. It might be expensive, but that makes sense for a product that can take incredible amounts of work to create.
posted by sonic meat machine at 6:36 PM on March 18, 2015 [2 favorites]


userland tool that's GPL and "state of the art" is Emacs.

Git. I don't know what's "state of the art" but there's some other very popular stuff like VLC. WordPress is still huge.

I agree that the GPL has lost a lot of relevance, but with it goes the relevance of the complaint that RMS doesn't want anybody to make money from software. And I'm not convinced the viral license was bad for open/free software 25 years ago, before commercial developers caught on that there was value to them to having this stuff out there. Am I wrong, was it irrelevant the whole time?
posted by atoxyl at 7:13 PM on March 18, 2015 [2 favorites]


I use Linux on the desktop, Android in my phone and have the source code from my router open (in every sense) in a window right now. I use Audacity (oh, on preview from atoxyl, VLC and Wordpress too) in my work, and I lack for nothing in their functionality.

Thirty years ago, I was hand-disassembling an OS to find out what the hidden magic was I needed to make my code work. That there was hidden magic was clear, as the OS manufacturer used it itself in its other products, but it was undocumented and risky to make any assumptions about, and if by some miracle our product had in any way become seen by that OS manufacturer as competitive or threatening, then history shows that they would have no compunction about using such things as weapons. We spent a lot of time and worry on things like that. And I was lucky in that I reviewed dev tools for magazines, which meant the start-up got stuff to play with for free that would have been too expensive to experiment with.

I had ten more years of that sort of nonsense.

Whether there'd be a Linux, an Android or an open (very advanced) router in equivalent form if GPL hadn't happened, I can't say. Perhaps a better Stallman would have pushed a better GPL, and we'd be even better off in some way, there'd be more good software I could use or entire industries would be paying less for Adobe/Oracle/CA like functionality (to pick three companies who prosper mightily from lock-in, to the expense of most of the things I value). But I can't really subscribe to the 'it did nothing of use' or 'it poisoned the well' point of view, when my life is substantially different and better through the GPL.

If I wanted to go and be a programmer again and had something of worth to donate, I could choose the GPL, or one of a whole spectrum of alternative permissives, or slap a "This is secret and you will die if you copy it" EULA on my work. I'm not sure I'd have that range of choices if Stallman had decided to be a parrot keeper. There are a lot of programmers out there who produce stuff under all of the above and manage to eat at the same time, and while you can't say much from things that haven't happened, I do wonder what class of product the GPL has prevented from appearing.

Things could be better. But they could be a hell of a lot worse, and my actual life is full of actual evidence that GPL has been a big part of that not being true.
posted by Devonian at 7:19 PM on March 18, 2015 [17 favorites]


True, git is excellent, but I don't know that I'd call it a "userland" tool (it's another development tool). VLC is good, too. WordPress gets around everything via the "loophole" I called out earlier, which is the server/client deployment. People can write plugins that are simply never distributed, and thereby escape the GPL.

The thing that I've come to realize about open source userland is that it only admits a certain level of abstraction. So, you have emacs/vim, which work directly on plain text files. You have git, which works on plain text files and the underlying file system. VLC translates a file to an audible sound (i.e. File → Memory → Sound Subsystem). GIMP translates a file into memory and allows you to manipulate it, then return it to disk (File → Memory Bitmap → [mathematical transformations] → File). More complex software, where data manipulation might encompass a vast array of possibilities, external integrations, robustness requirements, and so on, simply isn't "emergent" enough for this process to work.

I don't assert that the GPL has prevented this from occurring, Devonian, but it must be noted that for Stallman and the FSF, they view proprietary software as bad. The software that does do these things (tax preparation, business process automation, medical imaging, on-board automotive software, ...) is less "morally correct" than GPL tools.

Personally, I think that without the FSF, many of the same "base products" would have emerged. They might have looked different, but there would be free operating systems, free development toolchains, and so on, simply because it makes economic sense for people to collaborate on them (without necessarily the ideological baggage of Stallman/the FSF). In essence, if there was no Linux, someone would create Linux (because, of course, Linus did, and was not part of the FSF, and the actual FSF operating system has never materialized).

If you want to see why I think the FSF is actually harmful, the gcc saga is a good one. It started out as unifying the fragmented landscape, but after a period of time RMS actually became hostile to adding necessary features (exporting the abstract syntax tree) because he thinks it might help proprietary software. This is a good overview. Thus, libclang/llvm, Apple's support, the UoI license, etc.
posted by sonic meat machine at 8:07 PM on March 18, 2015 [1 favorite]


So no one likes to say this out loud, but part of RMS's initial impact was he was an amazingly good programmer. Both gcc and emacs were incredibly good technology in their heydey. But they aren't any more. No one makes that level of significant technical contribution over 20+ years. (With maybe an exception for Rob Pike). RMS got hamstrung by his wrist problems, and maybe the usual slowing down of getting older, and also not keeping up with changing technologies. LLVM is a significant technical improvement over gcc's AST just the way gcc was a significant improvement over BSD cc. I also think the failures of Hurd and Guile were hugely painful to FSF and RMS. Even if they won't say so themselves, the bitterness about the "GNU/Linux" name speaks volumes.

I guess I'm saying the FSF's time as a technical leader has past. That's OK, the legacy is hugely influential both technically and politically. It's possible to respect their contribution while also recognizing the world has moved on.

(And I wince at holding up VIM and Emacs as "state of the art". They are both ancient, inhumane TTY-based user interfaces that are hideous. I think SublimeText strikes the best balance of hackerware editor that's also got some semblance of a modern UI. But of course it's not open source.)
posted by Nelson at 8:39 PM on March 18, 2015 [3 favorites]


Nelson, I still think that the muscle memory you can acquire using vim makes it superior for text manipulation to the newer, sexier editors. Worse on first experience than Sublime, but superior after six months, which is a good trade for anyone who spends hours a day staring at their text editor.

(But this way lies holy war.)
posted by sonic meat machine at 8:44 PM on March 18, 2015 [5 favorites]


The true way™ is evil mode vim emulation within emacs. Needing to touch the mouse in order to change text, or chords for any frequent task, is a huge usability fail in a textual domain.
posted by idiopath at 9:05 PM on March 18, 2015 [3 favorites]


I don't assert that the GPL has prevented this from occurring, Devonian, but it must be noted that for Stallman and the FSF, they view proprietary software as bad.

I don't particularly share his ideology, I was just surprised at the hostility toward the very idea that he would stand behind anti-commercial principles, considering that the destruction of commercial software at the hands of the FSF isn't exactly impending.

I have to admit there was probably never a time someone said "I'll use this library in my produc-- oh shit it's GPL guess it's time to give way my code now." But I do kinda think it popularized the idea among people who were already hobby programmers that it was in their interest as well as everybody's to make their work free and open source.
posted by atoxyl at 9:54 PM on March 18, 2015


emilyw - I don't think I've seen anyone address this, and it's a point worth discussing. When you say "So far as I know, nothing about the GPL precludes the author of a work from licensing it to you (specifically) under an alternative commercial license for a fee. " we bump into a thorny little problem, because the author is implicitly a unitary enity, be it one person or one corporation. But the goal of openness is distributed authorship where changes are returned, integrated and re-shared. The upshot- on any non-trivial open project there will be no one entity that owns copyright to the whole project and can re-license it without onerous steps. One may reject outside contributions, require copyright assignment of contributions, obtain permission at the time of relicensing, or just hope nobody who matters cares.
posted by wotsac at 10:14 PM on March 18, 2015 [1 favorite]


Needing to touch the mouse in order to change text, or chords for any frequent task, is a huge usability fail in a textual domain.

I am perfectly happy with regular vim + tmux as opposed to evil-mode emacs, but I totally agree with the sentiment here.
posted by en forme de poire at 10:27 PM on March 18, 2015


It's received wisdom among some programmers that keyboard only interfaces like vim and emacs are the ideal, but it strikes me as a rather dubious assertion given that the rest of the open source / free / libre / GNU/whatever ecosystem is not known for its usability.
posted by Pyry at 11:14 PM on March 18, 2015


I think linear scales of usability are problematic when you cross paradigms.

I can log into my Virtual Private Server from my phone, and from there tunnel back to my home desktop machine. This is without leaving any incoming ports open - just a tunnel connection that I initialize from the desktop side. From that connection I can load my emacs session, with all the files I have open, the output of any build commands, and an interactive interface to my code in my dev process. I can also access any files I might need, and copy them up to the VPS if I want a friend to be able to download them.

What this means is that no matter where I am, if I have my phone, I can access all my important files, including all my work in progress, and use all of the software I use to do my job.

But of course this is a total "usability fail" because it's all terminal UI. And funny enough the fact that it is terminal UI is also why it works so well with so little setup, and why I can access all of this so quickly (plain text is very low in bandwidth usage).

I don't know if you can directly compare things that are immediately useful with no training or learning process, and things that allow unparalleled flexibility, productivity, and power given the requirement to learn some skills, a few general principles, and a bunch of weird special cases before it all makes sense.
posted by idiopath at 11:29 PM on March 18, 2015 [2 favorites]


Anyone mind if I stick a non-programmers prospective in here? As a scientist, GPL is great. There are three types of scientific software:

$40K/license things the university has one copy of, if that, or knows someone with a super computer we can pay to run it. Example: Gaussian. These tend to have very strict license terms, and you regularly see people publishing with outdated versions, since they can't afford to upgrade. Now, some of it only costs a few thousand dollars, but it still gets tied to one PC, so hope you never want to move it to something newer. My labmates are glaring daggers at our weak, general use office PC that someone tied some monty carlo simulation software too, while our beastly PCs sit idle, since we can't move the license.

Software that comes with equipment: Hell on a PC. Never works right, since they already got your money buying the gear. Always uses some special format, so at best you'll have to save it in that to keep the metadata and then once as a CSV to back it up. Oh, and it is usually tied to some specific combination of windows version and hardware, which is why my lab has working Windows 3.1 and Windows 95 computers and a stock of floppy disks, because it would cost a fortune to move the software to anything newer. If the computer breaks and you can't replace it, hope the company is still in business, makes that product and hasn't changed all the interfaces and won't charge more then a few hundred dollars for barely usable software.

Then there is GLP and other FOSS software. The interface normally sucks, and you might have to compile it yourself, but it will usually run on most hardware, and you don't have to pick one computer in your lab to run it on. But you know some prof, somewhere will keep updating it forever (Or at least until they die --I know of one bit of software that is being allowed to die, since there are better alternatives out there and the maintainer died). I don't have to worry about one computer dieing and taking my data with them. They usually give me nice, easy to parse output. Hell, sometimes you get a choice of interface to use with them, and you can usually write your own programs to work with them.
Hobbiest software? Lets see, Shelx is the industry standard crystallography software, and an XRD runs in the several hundred thousand to millions range. Its main competitor? Crystals, which was forked off of Shelx when another prof thought he could do a better job.

SO yes, I like nice, viral software. Because if you don't keep it around, software devs will screw you in a heartbeat to make a buck, using every form of vendor lock in and annoying DRM you can imagine. Want to see it in action? Find any IR, UV-Vis or Ramen spectrometer and try and get a file from it to open on any other computer with all the metadata intact, or ask the chem or physics dept tech person why they have so many antiques running.
posted by Canageek at 11:30 PM on March 18, 2015 [29 favorites]


(Oh, I should say: The one exception I've encountered is Bruker. You do need to license their software, and it costs a lot, but you can get floating licenses, the software works really well and gets updated a lot. Also they work with the FOSS community, provide updates and linux and mac versions of most of their software.)
posted by Canageek at 11:33 PM on March 18, 2015


Academic open source definitely has a great deal of value, Canageek! I certainly wouldn't cast aspersions on it—but it is kind of a special case, because it has built in, usually-technical users who rely on it. You can get developers to work on crystallography software because it matches their research aims; it's harder to get people to work on software for automating insurance applications for free.
posted by sonic meat machine at 2:32 AM on March 19, 2015


I think the greatest evidence that non-copyleft licenses actually reduce freedom is the fact that, when Apple were using the GPL/copyleft gcc, they released Objective C as open source. Its no coincidence that - now they're using the non-copyleft LLVM instead - their new programming language, Swift, is proprietary.
posted by axon at 3:04 AM on March 19, 2015 [10 favorites]



Canageek, there's another kind of scientific software, which I call "academicware". It's licensed free-of-charge for use in education or by educators or students. It might seem a noble sentiment to license things this way, but the persistence of this kind of licensing 30 years in usually makes me wonder what the intent really is.

Is the intent compatible with FOSS licensing but the licensors just don't know how to use these licenses? This I think is often the case--it's not unusual to read the COPYING file in GPL licensed code to see that the metasyntactical example licensor name and address at the bottom of the text have not been changed. I've also seen code that in one place is said to be under the GPL, but in another place is said is meant to be for non-commercial academic use only. Or, code where there's some statement pretty clearly FOSS-sympathetic on the web site for the software, but no license text anywhere in the software source code tarballs or directories themselves.

Is the intent to force any commercial use into different licensing? If so, often this is made explicit with an offer of other, paid licensing, but not always. One gets the sense that folks don't really know if or how they'll try to collect from licensing, but they're hedging their bets. This isn't that surprising of course, the constant gnawing of academic funding hunger being what it is.

Is the intent to just stick it to "the man" in some vague way?

Since you mention Gaussian, I'll note the rather famous (in the field) example of software written as an explicit, academic-ware alternative to Gaussian, a project founded explictly in reaction against Gaussian's licensing policies. However, digging through all the discussion of the history and the licensing terms, one finds that this alternative is, itself, still not free and open source software, at least in the way I understand the criteria used by the FSF, or the Open Source Initiative, or the Debian Free Software Guidelines. The GPL, it seems, would be the perfect fit for a project like this, where the intention is *clearly* that Gaussian not benefit uniquely from any of the code created for the project. Or, at least, that's the way it was last time I looked. I'd be delighted to learn that something has changed and one can now just "apt-get install" it.

There sometimes is the intent to control the software, out of some stated concern for quality or priority. Nevermind that bitrot guarantees quality will suffer, once the benevolent-dictator-for-life of a software package is unable or unwilling to keep maintaining and porting the software--I know a couple of packages where this has happened and its a bear, for instance, getting academics to give up their old PowerPC or at least Rosetta-capable Macs because that's the last known architecture for a given package. Ugh.

Academic-ware licensing is especially pernicious because it acts as a sort of competitive inhibitor to the development not only of more straightforwardly commercial, proprietary solutions, but also to the development of FOSS solutions, because to the lay user free-of-charge for working on just this one problem for just this one paper or grant application or assignment is free enough. Then, they're invested in it, and it becomes entrenched.

What's worse is when the academicware comes with source code--at that point, anyone who has worked with the source can (or at least, should) be further constrained in developing a FOSS alternative to avoid creating a derivative work of the non-FOSS, source-available work.

Bringing this all back around to Stallman and the GNU Manifesto, the trouble one sees even today in licensing software in the academy make it clear why he had to leave MIT to do this.



The mishmash of permissions and restrictions one sees really makes me appreciate the clarity and wide use of the GPL and the modern MIT/BSD licenses both.
posted by one weird trick at 5:23 AM on March 19, 2015


it's not unusual to read the COPYING file in GPL licensed code to see that the metasyntactical example licensor name and address at the bottom of the text have not been changed

It's an example and there's no need to change it. Note there are no changes in the Linux COPYING file.
posted by grouse at 8:01 AM on March 19, 2015


I use emacs and it is amazing and far better than any IDE I have ever used, so yeah, no, it counts as state of the art for a bioinformaticist/programmer.

Also, the academic license is there so that academics can use it without having corporations use it without paying the maker. There are plenty of these and I have no problem with them.
posted by lizarrd at 5:28 PM on March 19, 2015


Also: I buy plenty of software that, if you look deep into its folder structures, will have a 'source' or 'GPL' folder with source code in it. They don't advertise that it is there, but it complies with the license terms, and sure, you COULD compile it yourself. Or just pay them to not have to deal with compilers, dependencies, etc.

one weird trick: Yes, I've seen that type of software as well. As I recall some of the Shelx frontends follow that. However, even if they don't last forever, they tend to at least be available on multiple computers, various operating systems, etc. There is a lot less incentive to try and get you to pay through the teeth in that case. Also, I don't know that it discourages competition: How many SXRD solving programs are there? (Shelx, Crystals, OLEX2/CGS, and at least one other one). How many NMR programs are there at different cost points? The difference in those cases is that the data isn't locked down as tightly to one machine. Not sure why Bruker didn't go that route, unlike other gear companies, but you have your choice of FOSS, closed source and commercial software in that area.
posted by Canageek at 6:01 PM on March 19, 2015


Richard Stallman is absolutely brilliant. We'd all be slaves to Sun, M$, Apple, etc. by now if his movement hadn't taken off. I highly recommend his political notes section of his webpage too.

I love the GPL itself too of course. I'd prefer that copyright did not apply to closed source programs, ala the source code is copyrighted but if you do not share the source then the binary does not derive the copyright.

Amusingly, there are many categories of software, like cryptography, for which closed source software is basically worthless, but dual licensing under the GPL and a commercial license has proven profitable for open source projects like Qt.
posted by jeffburdges at 9:16 PM on March 19, 2015 [2 favorites]


« Older "I feel like my own identity was born in those...   |   Self Help Books You Wish Existed Newer »


This thread has been archived and is closed to new comments