Comment by smush | Hacker News Reader

Comment by smush | Hacker News Reader

This argues for a third counterpart to FSF-style "Free Software" and OSI-style "Open Source Software" terms.

Apparently, FSF is too preachy and OSI is too business-y and thus following "those who forget history are doomed to repeat it" style, a new, third movement of producing software that is free to copy, free to modify, etc. will naturally arise on its own.

There are allusions to brogrammers who know nothing more about OSI than 'It's on GitHub!!', I'm not sure how charitable that is but I can say for several years that no matter how many wiki walks I've taken down the history of Linux, GNU et al which is probably where some of those brogrammers fell as well. IDK the difference, does it really matter, might as well stop looking into it any further than that.

The schism and lack of unity between church of FSF and church of OSI has probably had a net detracting effect on getting software to be more permissible, even if the absolute effect has been way more FS/OS software than would be expected otherwis.

I could be totally wrong, I didn't show up for computing until the Win 95 days.

Thanks for reading!

> a new, third movement of producing software that is free to copy, free to modify, etc. will naturally arise on its own.

I think this is a bit too teleological. I'm not confident that such a thing will happen. I think that the environment is ripe for something to happen, but I don't know what that is. That's exactly because of

> IDK the difference, does it really matter, might as well stop looking into it any further than that.

I don't think that most people's conception of open source matches the FSF/OSI definitions. Does it really matter? I don't think it does for most people. I do think there's a values mis-match there though, and that means there's the possibility for a new thing that's a match. I think there are significant obstacles to that project, though. That's for the next post. I hope that I actually write it though, I've been trying to for over two years...

> I don't think that most people's conception of open source matches the FSF/OSI definitions.

To be honest, I'm still not sure what people think OSS is. I personally align heavily with the FSF model but that's mostly because I live in an industry (embedded) that obsesses over proprietary everything and thinks their shitty source code is the apex of valuable intellectual property (\rant) but I get that the story is very different on the other side where everyone is publishing Yet Another Javascript Library. Maybe this is all just a product of our environment?

I don't claim to know exactly either, and maybe you're right, and I'm over-correcting based on my environment :) But I have had a lot of conversations about this in the past few years...

(I personally prefer the FSF model, but think that the OSI has won, at this point. All my software is Apache2/MIT these days. And even though I prefer it in the abstract, I have a lot of criticisms of both.)

I think I forgot who I was talking to for a sec... I'm guessing there's quite a bit we agree on :)

Looking across the aisle, I do wonder from time to time if the AGPL was a sort of breaking point. GPL always "threatened" industry at the desktop level and lower but once everyone's beloved FAANG+/- was in the cross-hairs the discussion maybe degraded a bit.

:)

I think so, but I tend to think of it as not the AGPL's fault, exactly. That is, the GPL was a product of its time: it's very concerned with the details of writing non-networked C code. It pre-dates the web! It was impossible to forsee these kinds of changes, and it just so happens that it was not really prepared for them. The AGPL was an attempt to plug that hole, but it was too little, too late.

Yea, I don't think it's AGPL's fault. I think people became complacent with the landscape of GPL and what that meant for production applications but when GNU came around and tried to make it all apply uniformly (right after dot-com) OSI had already gained enough traction that a full GPL looked "weird" and turned into more of a signal that you didn't want anyone to use your code. That and Google et. al. banned it as it obviously threatened its model that some might see as abusing GPL.

Almost the right timeframe, but from my perspective the GPL2/GPL3 split is what broke everything. The GPL was a glorious hack as long as it was alone in its niche (supported by admonitions against license proliferation), but its success was predicated on it instantiating itself as a singleton. But the basic problem was simultaneously desiring (a) the ability to combine any collection of software and (b) preventing certain combinations of software. Now that we have to choose, the zeitgeist seems to have given up on (b).

Free Software seems like a model aimed at maximizing the available benefit for all people, while Open Source seems like a model that aims to maximize growth (in both quality and quantity) of that benefit, even if unevenly applied.

To me, this seems very analogous to some economic models I could mention. Communism sure sounds good, but I feel pretty confident in stating that the wild growth Capitalism has encouraged has made even the poor much better off than they would be in a system that was purely communistic for the same time period.

That said, I wouldn't call our current system purely capitalistic, which I think goes a long way towards explaining why the poor have an increased standard of living, and points towards the truth of what you're seeing (and what we've been seeing more recently with "interesting" business models using Open Source). The middle ground is often much more beneficial than the ideological extremes.

To bring this back to Free Software and Open Source, do we think there would be nearly as much high quality free software if Open Source had not introduces the large talent pools and funding of businesses? Would Open Source have had anything to entice those businesses with had not Free Software provided the base upon which to build the current ecosystem? They both provide a large benefit, but for different aspects. I think a mix of them is the most beneficial way forward, whether that's a third way, or a vacillation between these extremes as more licenses are developed and tweaked that change how the market for software of this type responds.

I don't agree that the middle of extremes is better, but I do think this is a very interesting framing, thanks for this. Your first paragraph, especially is pretty insightful, I think.

Other than moral connotation and RMS being the way he is, Open Source is just business friendly rebranding of Free Software. Is there anything else to the "FSF model" that is different from OSI proposition?

I don't buy the FSF position that Open Source is only about efficiency, collaboration etc and that Free Software is superior - it is the same thing, albeit different language but the concepts are identical. Playing moral superiority is only counterproductive to FSF. Btw I wish them to succeed even if that might literally mean becoming a tax exempt church so that we could all be praying by the holy book of Emacs.

For GPL'd software you're required to share your changes, for open source software you aren't. That's a big difference, especially for a business that wants to sell their changes.

So when people say open source is business friendly, they don't mean that the vibe is business friendly, they mean that it removes a restriction and allows more avenues to make money.

Being similarly FSF-aligned, all I know is that in gamedev it's not unpopular to consider things like Unreal Engine "open source" simply because you can get access to their source code on GitHub :(

I agree with both of your points.

I will add that if the difference between the FSF and OSI definitions of OSI don't matter...perhaps they should merge and join forces? Get the (not that it sounds great to say it this way) least common denominator between the two definitions, unify the Orgs, and present a unified front to prospective learners. There's propriety software and FOSS, the former word is defined in the dictionary, the latter word/acronym is defined at freesoftwareisopensource.stallman , rthar than the "it's over here at the FSF websites, on your right, past the "Linux should be called GNU/Linux arguments" aisle."

Maybe, but I can't possibly see it happening. Two decades of bad blood is non-trivial to resolve, and I'm not sure that the two projects really are trying to do the same thing. From the FSF side, see https://www.gnu.org/philosophy/open-source-misses-the-point.... for example.

>if the difference between the FSF and OSI definitions of OSI don't matter...perhaps they should merge and join forces?

I don't see how merging would be possible because FSF and OSI have incompatible philosophical differences to sharing/distributing software.

E.g. OSI is compatible with permissive licenses such as MIT License and BSD License. That's totally against what FSF and Stallman is about and their GPL license reflects that philosophy.

The MIT and BSD licenses are also Free Software licenses, see https://www.gnu.org/licenses/license-list.en.html

> This license is sometimes called the MIT license, but that term is misleading, since MIT has used many licenses for software. > > This is a fine license for a small program. A larger program usually ought to be copyleft; but if you are set on a lax permissive license for one, we recommend the Apache 2.0 license since it protects users from patent treachery.

and

> This is a lax, permissive non-copyleft free software license with a serious flaw: the “obnoxious BSD advertising clause”. The flaw is not fatal; that is, it does not render the software nonfree.

Note that some software licences may qualify as being a free software license, yet be not compatible with the GPL. Those are two different things.

> OSI is compatible with permissive licenses such as MIT License and BSD License. That's totally against what FSF and Stallman is about

Essentially all of the permissive OSI-blessed “open source” licenses are also FSF-blessed “free software licenses”; they are not “totally against what the FSF and Stallman is all about.”

>Essentially all of the permissive OSI-blessed “open source” licenses are also FSF-blessed “free software licenses”; they are not “totally against what the FSF and Stallman is all about.”

To both dragonwriter and steveklabnik,

Saying "totally against" was too strong. Let me try to clarify.

Yes, the GNU license list has "The following licenses qualify as free software licenses, and are compatible with the GNU GPL."[1]

To me, that isn't really about BSD & MIT being compatible with underlying philosophy of FSF and Stallman. That just says using those licenses as part of GPL projects is acceptable. However, they don't really fulfill Stallman's strategy and objectives.

An example of what I mean by BSD & MIT not aligning with Stallman's objectives would be past comments about him missing the chance to include Chris Lattner's LLVM/clang project in GCC.[2]

Presumably, Stallman would have not let llvm/clang become permissive-BSD license like Lattner did. He would prefer llvm/clang's valuable capabilities to be licensed as copyleft-GPL just like GCC. (More commentary on that style of thinking.[3]) To him, the permissive licenses "helps the enemies create proprietary software" (paraphrasing previous Stallman comments).

That's why I believe FSF and OSI have fundamental incompatibilities. The MIT & BSD licenses being on the GNU "approved" list doesn't really solve that.

EDIT to add reply to: >The FSF and OSI are broadly aligned on what freedom software should be provide.

I disagree because it seems that Stallman has taken great pains to explain why they are not aligned on the freedoms that software should provide. There is some overlap between FSF and OSI but that's different from alignment.

[1] from: https://www.gnu.org/licenses/license-list.en.html

[2] https://lists.gnu.org/archive/html/emacs-devel/2015-02/msg00...

[3] https://www.gnu.org/licenses/why-not-lgpl.html

> To me, that isn't really about BSD & MIT being compatible with underlying philosophy of FSF and Stallman.

Anything the FSF recognizes as a Free Software license (whether or not compatible with th GPL) is compatible with the FSF philosophy—It may not be compatible with the FSF strategy, because the FSF has a strategy, but that's a slightly different issue.

The FSF and OSI are broadly aligned on what freedom software should be provide. They differ on how problematic they view software that does not do that, and (largely as a consequence of the preceding disagreement) on how to best move toward a view where the benefits of software providing the freedom they agree on is broadly enjoyed.

> Anything the FSF recognizes as a Free Software license (whether or not compatible with th GPL) is compatible with the FSF philosophy—It may not be compatible with the FSF strategy, because the FSF has a strategy, but that's a slightly different issue.

I don't think this is accurate. Anything the FSF recognizes as a Free Software license can be relicensed under the GPL because nothing in it contradicts the GPL. BSD, MIT, Apache, etc. licenses don't put any restrictions on how you can relicense the software, therefore they can be relicensed under the GPL.

The philosophy if the FSF is that you should have access to the code of the software you run, and that all software should be GPL. It helpfully offers a list of software that can be used as a base for GPL'd software.

In this case, "compatability" should actually be read as in "IBM-compatable" not as "simpatico" or "synonymous."

> perhaps they should merge and join forces?

I wish that too, the problem off course is that FSF as an organization is absolutist and that would never work.

This recent install-fest threads illustrates this vividly: https://news.ycombinator.com/item?id=19435791

Finding a viable third way is going to require identifying a set of broadly agreed-on values that differ from those espoused by FSF or OSI. Maybe my thinking is limited by what's already out there, but I have a hard time seeing a third free-as-in- beyond speech (FSF) and beer (OSI). Perhaps thinking in terms of free-as-in- is too constrained and we need to propose an entirely new social contract around participatory software development?

> Maybe my thinking is limited by what's already out there, but I have a hard time seeing a third free-as-in- beyond speech (FSF) and beer (OSI).

We, OSI and FSF are both free-as-in-speech; they both use definitions centered around th freedom of the licensee.

I don’t want to spoil too much, but yes, I think “participatory” may play a part. I have more questions than answers, though.

Let me spoil it a little then :)

I use the word "participatory" because it seems like a lot of complaints about the current state of free and open source software hinge on a lack of reciprocity. You have the individual maintainer of a small project, who feels burned because there are people out there making money on her back. You have companies like Red Hat making life harder for CentOS because Oracle are a big pile of freeloading jerks (old news, I know). You have other companies like Google strategically open sourcing some of their technologies for commercial advantage. We have complaints about maintainers who orphan their projects, and we argue about whether it's OK to complain about not getting something for nothing any more.

It seems like all of this strife comes from disagreements about what expectations we should have for all of the stakeholders -- owners, users, contributors, and maintainers. Not all of this can be spelled out in a license, as we've seen in the rise of CLAs and COCs. And so I use "participatory" in the sense that participating in third-way open source would impose some obligations in addition to the entitlements enumerated by the license.

I get the impression that the Rust community is wholly burdened in this third wave mindset. Everything is dual licensed Apache / MIT to guarantee maximum developer freedom. There have probably been a billion man hours poured into developing the Rust ecosystem at this point and I personally really think its destined for greatness. But the licensing style is already coming back to haunt all the projects - with growing popularity game studios, big megacorps, Mozillas own competitors, etc have started adopting Rust.

And they give nothing back. No money, no developer hours. They take the fruits of a fledgling ecosystem and experience great productivity and performance improvements in their commercial products and reap increasing profits on the backs of people who naively just thought "I just want people to be able use my code is all".

The third wave does not recognize the tremendous costs associated with giving away productivity to corporations. It costs society broadly, because it empowers business interests to exploit workers and customers for greater revenues not through their own ingenuity or innovation but through the labors of others given away freely. That isn't being generous on the parts of the developers who wrote the code for those that most need generosity.

Several of our production users do give back developer hours. Some have asked to give money, but we don't have a way to do it yet. We're working on it...

> The third wave does not recognize the tremendous costs associated with giving away productivity to corporations.

I think they do, and that's where the unrest is coming from. This realization has come after, not before, actually contributing.

Reply on news.ycombinator.com