Hacker News Reader

I am a lawyer, and I really appreciated your "free" versus "fair" line.

One tidbit on the GPLs: They require you to share alike if you share the software with others, but do not require you to share the software with others. You can make changes and larger programs entirely for your own or your company's internal use, and not share anything.

The GPLs mention this explicitly, and implement it using concepts like "distribute" and "convey". Your obligation to share alike doesn't kick in unless you distribute or convey.

That got GPL into trouble later on, when it turned out folks often ran software for others to use over networks, rather than sharing with users in copies for them to install locally. That led eventually to AGPL, a kind of fork of GPL, but outside FSF, also to licenses that simply require sharing back whether you distribute your work or not.

That is fascinating. Do you happen to know of any good resources for pitfalls/downsides of other popular licenses (bonus if layman friendly)?

This one is a good resource:

Software Licenses in Plain English https://tldrlegal.com/

You need to be careful though. For example for Commons Clause they accepted changes by Drew Vault (ironically the author of the article we are discussing here) which are misleading at best. Now it says that commercial use is not allowed [0], while it is really only selling that is not granted [1]. In other words, do read the license if you want to know about it.

[0] https://tldrlegal.com/license/commons-clause

[1] https://commonsclause.com/

For example this? https://writing.kemitchell.com/ ;)

(GP is the author - it contains some very good explanations on the subject of licenses, imho of course)

Thanks for the link :)

I have been looking for exactly this content for years! Thank you!

> That led eventually to AGPL, a kind of fork of GPL

Yes.

> but outside FSF

Initially. The current version was published by the FSF in 2007.

I wasn't clear. The reciprocal licenses that require sharing alike, with or without distribution, came from outside FSF.

Ah, gotcha. AGPL also came from outside the FSF in its original form, but was adopted (and later extended).

Pre-FSF Emacs from Stallman came with a request to send any/all changes made back to Stallman.

https://www.oreilly.com/openbook/freedom/ch06.html

> when it turned out folks often ran software for others to use over networks

aka the SaaS/cloud loophole

"ASP Loophole" or "The Google Problem" at the time.

I like the idea of a license that is "fair", but there's something about GPL that rubs me the wrong way where I feel it accomplishes the goal of fair in some antagonistic manner.

To me, GPL feels more like holding a legal gun to your head and say "You used my code, now play fair". It's not your Mum asking you nicely to share your toys with your brother. But then MIT is too far the other way where it's your Mum saying that if you don't want to give any toys to your brother even though he gave you a bunch then that's your ethical code and you have to live with it.

I wish there was a license that did feel like the "it's the right thing to do" argument without the threat, but I just don't know how you'd achieve that. Even something where you license the core as GPL and then the majority supporting libraries as MIT doesn't feel right to me.

As a lawyer, do you have any ideas along these lines?

It's not as if they spring the GPL licence only after you use the code, though.

It's more like "hi there, if you want to use my code, these are the terms under which I've made it available" than "you used my code, now play fair".

Think of it more as the price you pay for using someone else's work. Just like when you pay $1k for a license to ship some proprietary software, you agree to instead pay for using this code by sharing your source, too. You're free to not pay either price and write your own, of course.

> I like the idea of a license that is "fair", but there's something about GPL that rubs me the wrong way where I feel it accomplishes the goal of fair in some antagonistic manner.

It kind of is antagonistic: GPL started when RMS wanted to fix his printer and Xerox said, "No." IIRC the printer software in question drew on or was based on his own prior work.

The GPL is a weapon to pre-emptively undermine sort of venal mentality RMS had already encountered. Free Software is a reaction to a kind of enclosure of the commons.

GPL is pretty much a "it's the right thing to do" kind of license. GPL violations are widespread and the terms are rarely enforced. Even when they are, in 99.9% of cases there are no penalties(unlike most commercial licenses), just a requirement to publish your code and adhere to the original terms of the license.

The folks who regularly do GPL compliance work have a very reasonable set of principles they apply when doing that work. Legal action is a last resort and GPL violators are given every opportunity to comply. The only legal actions that have been taken have been from years of stonewalling or even blatant refusal to comply and the consequences of those legal actions have been reasonable.

https://fsf.org/licensing/enforcement-principles https://sfconservancy.org/copyleft-compliance/principles.htm...

I don't believe the AGPL requires distribution of the sources unless you are exposing the software in some way. So, a company is free to run modified AGPL software on their intranet without having to offer the sources to anyone but their employees.

The first paragraph of section 13 of AGPLv3 reads:

13. Remote Network Interaction; Use with the GNU General Public License.

Notwithstanding any other provision of this License, if you modify the Program, your modified version must prominently offer all users interacting with it remotely through a computer network (if your version supports such interaction) an opportunity to receive the Corresponding Source of your version by providing access to the Corresponding Source from a network server at no charge, through some standard or customary means of facilitating copying of software. This Corresponding Source shall include the Corresponding Source for any work covered by version 3 of the GNU General Public License that is incorporated pursuant to the following paragraph.

Mmm, I agree. I think I got tripped up by this line:

> That led eventually to AGPL, a kind of fork of GPL, but outside FSF, also to licenses that simply require sharing back whether you distribute your work or not.

...because I read it as "the AGPL requires sharing back whether you distribute your work or not", and felt like splitting hairs for the case where you expose the work, but only internally to a group.

Are there licenses that say if you run an altered copy, you have to contribute back sources, no matter who interacts with it?

> I read it as "the AGPL requires sharing back whether you distribute your work or not"

That's true, insofar as providing as a service isn't "distributing", or in (A)GPLv3-speak, "conveying". See section 13 of AGPLv3: https://www.gnu.org/licenses/agpl-3.0.en.html#section13

Compare how the Open Software License handles network services: by requiring they be treated as "distribution". https://zoo.kemitchell.com/OSL-3.0#5

> Are there licenses that say if you run an altered copy, you have to contribute back sources, no matter who interacts with it?

For an older license of this type, have a look at RPL-1.5. For a newer one, which I led, have a look at Parity.

> the MIT license gives you the freedom to make an awesome change to the free code you were given, and add it to your closed source application with no obligation to give those enhancements you made to the code to anyone else.

It's also the freedom to make zero changes to the free code you were given, reuse it in your closed source software, and pass it off as your own.

In a way, I feel like I should be a big enough person to be okay with this, but I'm not. The possibility bothers me.

Is this meaningfully different from being able to build a multi-billion dollar company like Red Hat on top of other people’s GPLed work?

Yes, because of the GPL, they have to give RHEL customers the source to all their patches, but what difference does this make in practice?

> but what difference does this make in practice?

they have to give RHEL customers the source to all their patches, and allow them to share it with the entire world.

In practice, that means everyone gets better software, and RedHat gets compensation for their hard work.

> but what difference does this make in practice?

CentOS is the difference; it is one of the most widely used distributions has the same stability and longevity characteristics and gets security fixes just as fast as RHEL.

That is a good point. Perhaps a better example would be FAANG making billions and running their servers on Linux, which (IIRC) is what things like the AGPL are designed to prevent.

The AGPL wouldn't prevent FAANG making billions, just ensure they respect end-user freedom while doing so.

The decision between MIT-like licenses and copyleft licenses all depend on the contextual usage of the code. Is it just some quick and dirty script I wrote to convert some obscure binary format to another obscure binary format? I literally do not care who uses it and why/where/when/how. The script was complete the day I made it, it does everything it needs to do and nothing it doesn't. School assignments fall under the same category, they were useful for that moment and will never be needed again. Writing my own linked-list in C++ stopped being useful to me the moment I turned the assignment in so I don't care who uses it. It's not anything special and licensing it any other way would (in theory) only limit someone trying to learn. Did I actually write something clever that could actually be used somewhere helpful? Is it something that I can see others would want to build upon and improve? That's definitely copyleft worthy.

> and pass it off as your own

No, the MIT license requires attribution by including the original copyright notice with any distribution of the software. You can claim credit for the parts you build around it, but you can't claim to have written the MIT-licensed part.

"In a way, I feel like I should be a big enough person to be okay with this, but I'm not."

One of the distinctions I like to draw strongly is between "nice" and "good". They are not the same, and at times, are actively at odds with each other.

It is certainly nice to give away your software to everybody, unconditionally, and not be bothered if your code does happen to end up being a fundamental component of some multi-billion-dollar company that gives you nothing.

But I'm less convinced that's necessarily good. In a world where you must live under the assumption there are bad actors, I think it's OK to put into your license that they can't necessarily do that. It is good to prevent the exploitation of anybody, and that goodness extends to preventing the exploitation of yourself, even if you also clearly have direct personal interests of your own.

I don't think "being comfortable with being exploited" is a component of "being a big person".

On the other hand, we often bring to mind mental examples of software where we slaved away for years and years to bring some unique new product to the world, only for it to be stolen by the big bad corporation for no compensation. My verbiage above about "exploitation" probably brought that to the reader's mind as the topic of discussion. If that happens, yeah, it's bad. On the other hand, that's not really the dominant case for software. There's a lot of open source put out there that's just a few day's work, which is still plenty valuable at scale, but if someone took one of these little things and incorporated it into their product, it's not really that big a deal.

I have several such small little things and they're all BSD. If a big corporation did somehow walk off with one of them, I'd legitimately be flattered, because I know that while it may have saved them a bit of time, it certainly isn't a core component of their product and had they needed to implement something themselves it would have been a trivial fraction of their time. I know I can say I'd be flattered because to some small extent, this really has happened, and I really was flattered and appreciative, so I'm not just hypothesizing about my future emotional states. I'm well aware on all levels that my ~two weeks total effort on the project in question, which I mostly did for my own purposes and for which the open source effort was essentially a bonus, doesn't entitle me to significant compensation for a product whose size is measured in person-decades.

> I don't think "being comfortable with being exploited" is a component of "being a big person".

In the context of the types of projects I was imagining, a company would likely just rewrite any code that wasn't available via MIT or similar. If I'm not getting anything in return either way, am I being exploited?

Logically, if the outcome is the same for me, there's no reason to make others redo work.

That's basically the logic I used too for my own code. As I hinted at but may not have been made clear, I was using "exploitation" quite deliberately as a loaded term in the beginning, only to try to defuse it in my final paragraphs as in many, perhaps even the vast majority, of cases you aren't really being that "exploited" no matter what anybody does. If somebody outright stole any of my projects and claimed them as their own work, I still would honestly be only annoyed, and maybe make a few posts in my README.md about the situation, and move on with my life without even a thought about engaging the legal system or anything.

But as rare as it may be, there have been many cases where an open source code base really was "exploited" by any reasonable sense of the term, with corporations slapping their logos on it and reselling significant works in violation of the license, and those cases are significant because of their size, even if they are rare.

> there's no reason to make others redo work.

As a simplified example, consider Corporation G and Corporation P. One will eventually lose market share to the other, go bankrupt, and be replaced by Corporation R that bases its business practices on the more successful one (then the cycle repeats).

Both are evil, as corporations are wont to be, and in particular will never release anything under a less restrictive licence than they have to (so they will never release software under a MIT licence). However, G will release software under the GPL if this allows it to save work, whereas P will always release software under a proprietary licence, even if this forces it to expend more resources rewriting GPL functionality.

Releasing code under the GPL exerts a selection effect that favors G over P, and in the long run results in corporations that release somewhat more software period - and significantly more software under FOSS licences - than if P was not forced to redo work.

Permissive licenses give freedom to the developers who use code under that license. The developer is free to restrict the users' freedoms.

Copyleft licenses grant freedom to the users of software: a developer cannot restrict a user.

>First, I'm not a lawyer, but I've always seen the MIT license as being more "free",

If you are a lawyer can you maybe be more clear about "more free" , more free relative to what? AFAIK GPL protects the user freedom and I think you are referring at the programmers freedom, GPL is a license that respects the users/society as a group, most people here are developers sothey will say that this is less free for them as an individual, less free to take freedom from others(the freedom to study,modify and distribute).

If “users” = “not programmers” (which I think must be the definition here) then neither license gives users anything except the freedom to hire programmers to do something they want more cheaply. (I include buying software and downloading free-as-in-beer software in “hiring programmers”.) With MIT/GPL the programmers can afford to charge a lot less to do the same job.

It’s fundamentally programmer freedom in either case. Any user freedom is an indirect side effect.

You are correct if I am using an MIT/BSD program as a user I have all the freedoms the issue is if someone forks it and closes it up making it proprietary.

You need to think at the larger picture, compare Linux kernel and BSD and how the licenses affect the projects and all of the users. With Linux GPL prevents parasitic use, like you get Linux, put your proprietary thing on top, close it off and give it to some users, this users will have no freedom to know what the code they run does, no freedom to change it and distribute it. The fact that a user does not code is irrelevant because they ca always pay for others to modify things.

But if the proprietary fork is analogous to the original software the "user" can just seek out the MIT-licensed project and start from there. If the fork contains significant functional modifications then it seems fair to let the author of those changes choose how to distribute them. This doesn't restrict user's freedoms anymore than allowing greenfield projects to be proprietary already does. GPL is for people who are concerned with forcing their viewpoint into wider adoption. MIT is for people that are interested in truly gifting their work (and only their work) to society.

That is only possible sometimes, say I create ffmpeg, it can decode/encode 100 video formats, then Google/Apple forks it and support for a new proprietary format that only they use and not contribute it back, Now Google/Apple use this new properitary format and only their applications will be the only ones that can read all the videos formats.

Maybe a bad analogy, GPL - I make knives and if you want to use my knives you have to agree not to use them to injure/kill people because this taking their freedom

BSD - you can do whatever you want with my knives, you can kill people I am fine with it because I care only about you my customer freedoms but not about any other people freedoms(lives). Just don't remove my name from the knife.

No, the GPL is primarily about user freedom, and programmer benefit is a virtuous side-effect. The GPL is entirely made up of restrictions on programmer freedom, which is why BSD/MIT partisans rail against it. It attempts to define the maximum amount of programmer freedom that can be allowed without impeding the user's freedom in any way.

Actually, that's not really true either. Usually, the company that holds the copyright can afford to charge more for proprietary software, but they also can afford to pay less for software developers, because they hold both the monopoly for the software the user wants, and the monopoly for the work the programmer wants to do/is qualified for. With free software, both sides can potentially just skip the middleman.

Should go and watch any video in which RMS explains what free software aims at achieving.

I was looking at it from the programmer's perspective. Specifically, from the perspective of someone who built something themselves and distributed it.

From what I can tell, the user's freedoms (assuming no changes) are pretty much the same. They can use the software without worry.

The user does NOT have the same freedoms under MIT because there is no obligation to release the sources you have modified and distributed as a binary. A user may also be a programmer but if they have no source their options are limited.

Yes, the direct users of teh MIT program have the freedoms, the issue is the project can be hijacked and made proprietary and then sold to the users. As an example the fact Webkit is opensource is because it was impossible to close source the KHTML fork.

If I make a (A)GPL program, I can be certain that EVERY downstream user has access to the source code and the freedom to recompile, reuse, remix, etc.

That is not the case with MIT, where any developer downstream may choose to re-license as they see fit.

From a user's perspective, the rights are very, very different. When you choose a copyleft license, you are making a specific choice that downstream users should have their rights protected, even if that comes at the expense of the "freedom" of downstream developers--who are now prevented from doing anything which restricts user freedom.

Even within the libre-sense there's differing axes of 'free'. You're describing one. Another is from the perspective of the software itself. Similar to the phrase "information wants to be free" (which sounds cheesy but gets the message across), if changes made to the software are kept isolated it's less free to spread which is the motivation for free software before it got fractured into open-source that's less free in that sense.