6 ms·

I agree with the overall reasoning of Paul, and we are having the same conversations inside Redis Labs. For instance the reason why Redis remained BSD licensed

I agree with the overall reasoning of Paul, and we are having the same conversations inside Redis Labs. For instance the reason why Redis remained BSD licensed is that, while it's odd that cloud providers are exploiting open source system software in this way, yet to look just at that does not capture the whole story. Open source software changed our world completely, so the real balance must factor in also what happened to IT because open source exists. Even what now is a problem for OSS producers, the cloud providers, is in some way a product of the fact that software was democratised in this way by the OSS movement (and note that without the cloud to productize OSS is harder). Personally even the fact I write software is because I installed Linux 23 years ago or alike. So the point is to also have something that is interesting enough and that can be released in a commercial license and is complementary to the original project. Because the part that needs to be commercial, must go totally commercial (it's a long story but AGPL for sure, and probably also other attempts including SSPL and Common Clause are not just confusing, they may not be enough). However what this requires is a very good balance between what's open and what not. In the case of Redis this is a bit a natural process because I'm very unwelcoming to a lot of things that users want because I feel they are not a fit. Proof is that for instance, two of my popular open source projects, that are linenoise and dump1090, are currently both mostly distributed via forks made by other people. So Redis Labs can just jump in and do a number of very useful things like graph database, CRDTs, on-disk support and so forth without any conflict of interests. Let's see what happens. Right now the trend is to distribute the non-open parts with source code available and with very few limitations for end users: if this works is great, and it looks like is the only way because it is very unlikely that users will have the need of something as a service if they do not use it already in some form. DISCLAIMER: This comment was written incrementally and modified several times adding things that came to mind while writing. About BSD,MIT,... vs CopyLeft. We can say what we want, but if we just stick to facts, MIT-alike licenses provide more freedom for the final user. If one would like to dig more I think that the canonical example would be the Linux kernel and asking questions about, what if it would be BSD licensed.

pauldix 8y ago
Yeah, I'd be interested in an analysis of open source projects that are copyleft licensed and successful with the "what if" question posed about if they were MIT or similar. I'm definitely not the person to put forward an educated opinion on those.
zeotroph 8y ago
> Linux kernel [..] BSD licensed. It would be another BSD with not many commercial players contributing to. BSD-ish licenses mean free labor for companies, but are problematic for cooperation. The easiest way is always to "defect": Take the sources and never contribute back - it might help a competitor. The Copyleft licenses work around that and make cooperation between of adversaries possible.
- Joky 8y ago
  LLVM is an example that this black-and-white view of the world does not generalize.
  - antt 8y ago
    LLVM only exists because companies can't steal from gcc. The second gcc disappears LLVM will be forked a dozen times with no one seeing any source code again.
    - Joky 8y ago
      Can you clarify what is this claim based on? I have been working on LLVM for years and this is absolutely not the dynamic in LLVM...
  - robocat 8y ago
    Context for anyone who wants to understand what parent is referencing: http://llvm.org/foundation/relicensing/ http://llvm.org/foundation/relicensing/ https://news.ycombinator.com/item?id=18699228 https://news.ycombinator.com/item?id=18699228
    - Joky 8y ago
      Actually, I was trying to address "The Copyleft licenses work around that and make cooperation between of adversaries possible" to show that a more "BSD-like" license like LLVM have many "adversaries" collaborating: researchers, enthusiasts, and otherwise competing companies. This is a rebuttal to "It would be another BSD with not many commercial players contributing to. BSD-ish licenses mean free labor for companies, but are problematic for cooperation". https://www.wired.com/2013/07/apple-google-llvm/ https://www.wired.com/2013/07/apple-google-llvm/
nickpsecurity 8y ago
One of reasons Im ditching traditional FOSS is Ive found via all the lobbying and lawsuits that we need money to protect software freedom. Lots of it. I rarely see this discussed. I wrote more here: https://lobste.rs/s/0h98nl/it_s_not_okay_pretend_your_software_is_open#c_zim2ks https://lobste.rs/s/0h98nl/it_s_not_okay_pretend_your_softwa... Follow up: https://lobste.rs/s/kbcjnx/open_source_confronts_its_midlife_crisis#c_yokqtn https://lobste.rs/s/kbcjnx/open_source_confronts_its_midlife... Whats your thoughts on using more paid, shared-source stuff to generate money to protect software freedom and fund development of FOSS infrastructure?
blackorzar 8y ago
I would like to release a software with Commons Clause, but I don't entirely like the "Commons" name. Would it be accurate to call it "Apache 2 + Do-not-resell Clause"?
- wmf 8y ago
  Saying "Apache 2 + anything" is disrespectful of the Apache brand because you're adding additional restrictions that don't exist in the original; likewise with "GPL + restriction".
  - cyphar 8y ago
    Though with GPL + restriction, the license explicitly states that such restrictions can be removed. So it's less of an issue.
zzzcpan 8y ago
> MIT-alike licenses provide more freedom for the final user There is this freedom for end users to modify and run software, sort of anti lock in freedom. And there is this corporate freedom allowing corporations to restrict freedoms of end users and lock them in. The second one is what MIT-alike enables. It doesn't make sure end users keep any freedom at all. So the choice is to either restrict corporations or let corporations restrict users. Letting corporations restrict users also lets them capture all profits and make sure original creators won't be able to compete with them. No matter how you look at it, the idea of unrestricted use for corporations will die. Startups, small companies, independent creators are either going to switch to restrictive licenses or won't be able to sustain development and abandon it. Things could have been different if tech companies weren't allowed to get so big.
- antt 8y ago
  You are getting very close to the only idea I think makes sense. A license that is only to be granted to human beings. You can use, distribute, copy, modify and sell my software to your hearts content as long as you are doing it on the behalf of a single human being. If you are doing it for a legal entity of any description, you need to start paying license fees.
bhickey 8y ago
> cloud providers are exploiting open source system software This keeps popping up from database maintainers and unless addressed we're just going to keep talking past each other. I'd like to challenge the notion that cloud providers are somehow competing with the likes of MongoDB, et al. We're at the point where anyone can stand up a database in a VM with minimal effort. Cloud providers are spending billions on infrastructure to make sure the bits in that database are highly available with low latency. This is scarcely distinguishable from what a traditional colocation service does. The whole idea of "open source abuse" amounts to special pleading. No one is forced to use an open source license. MongoDB and its peers have access to lawyers who can explain the consequences of this or that licensing decision. Cloud providers offer database deployments under license terms that were offered to them. If Redis Labs or MongoDB decided tomorrow to take their database proprietary there wouldn't be much gnashing of teeth. The projects would fork. The proprietary forks would see more feature development and we'd all go on with our lives. You're seeing pushback for a few reasons: 1. The name "Commons Clause" appears to intentionally invite confusion with "Creative Commons." Plus the idea that you can jam a rider onto an arbitrary Open Source license and end up with something coherent is woefully misguided. The GPL explicitly blows up language like this. 2. On the heels of the Commons Clause, the SSPL was put up to the OSI as an open source license even though it obviously violates OSD 5,6 and 9. Along with Salil's rants on TechCrunch, this makes it look like a transparent attempt at undermining open source and the OSI. > In the case of Redis this is a bit a natural process because I'm very unwelcoming to a lot of things that users want because I feel they are not a fit. As an aside, I don't send you patches for this reason.
- kjeetgill 8y ago
  > The name "Commons Clause" appears to intentionally invite confusion with "Creative Commons." Plus the idea that you can jam a rider onto an arbitrary Open Source license and end up with something coherent is woefully misguided. Seriously. Trust takes ages to build and seconds to destroy. That and good will. I think Redis Lab's wouldn't have gotten quite the push back it did if it wasn't for the perceived shadyness.
- antt 8y ago
  Creative commons has an NC version. All the video game art I made in my younger days was NC/SA because the freedom to inspect all software you run does not mean you get to sell the software to other people.
Fufnrifnr 8y ago
Linux would not have been successful were it not for the licensing. Copyleft licenses are appealing to a lot of developers. People who are willing to donate lots of time writing free software don't want to feel they're being taken advantage of. Programmers are the life-blood of an open source project, I think developer preference is a huge factor. A BSD might have caught on in the same way eventually but it would have taken longer.
Avshalom 8y ago
>>bout BSD,MIT,... vs CopyLeft. We can say what we want, but if we just stick to facts, MIT-alike licenses provide more freedom for the final user. They provide more freedom for the next user but because they allow freedom to be taken away at any point in the chain bsd/mit licenses provide absolutely no guarantees of freedom for the final user.
ajross 8y ago
> we just stick to facts, MIT-alike licenses provide more freedom for the final user. That's not "fact", that's semantics. You're playing games with the definition of the word "freedom". The FSF has been outrageously clear about exactly what it means by that term, and what freedoms specifically it's trying to protect. Argue with those points. Don't fling around "facts" like this, it's just bad form.
cyphar 8y ago
> but if we just stick to facts, MIT-alike licenses provide more freedom for the final user. It depends what you think a user is. Is a user an end-user who has no intention of creating a proprietary fork? Or is it a developer who intends to create a proprietary fork? "If we just stick to the facts", copyleft provides equal freedom to both classes. Neither can create proprietary forks. For permissive licenses, the latter class is given an additional freedom (creation of proprietary forks), but the former class gains no practical freedom. However, what if the latter class gives the former class their proprietary fork? Then the former class has less freedom (and there is a history of this happening with Windows, OS X, and Android using BSD code without providing it to the end-user). This has always been the argument made by the GPL. Yes, proprietary software developers lose the freedom to reduce the freedom of other users. Instead it guarantees equal freedoms to all classes of users. I get it, you don't agree with the GPL. That's fine. But clearly the "user" being protected the GPL is the end-user. That was always the point (and that protection comes in a form that allows developers to have the same freedom and nothing extra).

pauldix 8y ago

Yeah, I'd be interested in an analysis of open source projects that are copyleft licensed and successful with the "what if" question posed about if they were MIT or similar. I'm definitely not the person to put forward an educated opinion on those.

zeotroph 8y ago

> Linux kernel [..] BSD licensed. It would be another BSD with not many commercial players contributing to. BSD-ish licenses mean free labor for companies, but are problematic for cooperation. The easiest way is always to "defect": Take the sources and never contribute back - it might help a competitor. The Copyleft licenses work around that and make cooperation between of adversaries possible.

Joky 8y ago
LLVM is an example that this black-and-white view of the world does not generalize.
- antt 8y ago
  LLVM only exists because companies can't steal from gcc. The second gcc disappears LLVM will be forked a dozen times with no one seeing any source code again.
  - Joky 8y ago
    Can you clarify what is this claim based on? I have been working on LLVM for years and this is absolutely not the dynamic in LLVM...
- robocat 8y ago
  Context for anyone who wants to understand what parent is referencing: http://llvm.org/foundation/relicensing/ http://llvm.org/foundation/relicensing/ https://news.ycombinator.com/item?id=18699228 https://news.ycombinator.com/item?id=18699228
  - Joky 8y ago
    Actually, I was trying to address "The Copyleft licenses work around that and make cooperation between of adversaries possible" to show that a more "BSD-like" license like LLVM have many "adversaries" collaborating: researchers, enthusiasts, and otherwise competing companies. This is a rebuttal to "It would be another BSD with not many commercial players contributing to. BSD-ish licenses mean free labor for companies, but are problematic for cooperation". https://www.wired.com/2013/07/apple-google-llvm/ https://www.wired.com/2013/07/apple-google-llvm/

nickpsecurity 8y ago

One of reasons Im ditching traditional FOSS is Ive found via all the lobbying and lawsuits that we need money to protect software freedom. Lots of it. I rarely see this discussed. I wrote more here: https://lobste.rs/s/0h98nl/it_s_not_okay_pretend_your_software_is_open#c_zim2ks https://lobste.rs/s/0h98nl/it_s_not_okay_pretend_your_softwa... Follow up: https://lobste.rs/s/kbcjnx/open_source_confronts_its_midlife_crisis#c_yokqtn https://lobste.rs/s/kbcjnx/open_source_confronts_its_midlife... Whats your thoughts on using more paid, shared-source stuff to generate money to protect software freedom and fund development of FOSS infrastructure?

blackorzar 8y ago

I would like to release a software with Commons Clause, but I don't entirely like the "Commons" name. Would it be accurate to call it "Apache 2 + Do-not-resell Clause"?

wmf 8y ago
Saying "Apache 2 + anything" is disrespectful of the Apache brand because you're adding additional restrictions that don't exist in the original; likewise with "GPL + restriction".
- cyphar 8y ago
  Though with GPL + restriction, the license explicitly states that such restrictions can be removed. So it's less of an issue.

zzzcpan 8y ago

> MIT-alike licenses provide more freedom for the final user There is this freedom for end users to modify and run software, sort of anti lock in freedom. And there is this corporate freedom allowing corporations to restrict freedoms of end users and lock them in. The second one is what MIT-alike enables. It doesn't make sure end users keep any freedom at all. So the choice is to either restrict corporations or let corporations restrict users. Letting corporations restrict users also lets them capture all profits and make sure original creators won't be able to compete with them. No matter how you look at it, the idea of unrestricted use for corporations will die. Startups, small companies, independent creators are either going to switch to restrictive licenses or won't be able to sustain development and abandon it. Things could have been different if tech companies weren't allowed to get so big.

antt 8y ago
You are getting very close to the only idea I think makes sense. A license that is only to be granted to human beings. You can use, distribute, copy, modify and sell my software to your hearts content as long as you are doing it on the behalf of a single human being. If you are doing it for a legal entity of any description, you need to start paying license fees.

bhickey 8y ago

> cloud providers are exploiting open source system software This keeps popping up from database maintainers and unless addressed we're just going to keep talking past each other. I'd like to challenge the notion that cloud providers are somehow competing with the likes of MongoDB, et al. We're at the point where anyone can stand up a database in a VM with minimal effort. Cloud providers are spending billions on infrastructure to make sure the bits in that database are highly available with low latency. This is scarcely distinguishable from what a traditional colocation service does. The whole idea of "open source abuse" amounts to special pleading. No one is forced to use an open source license. MongoDB and its peers have access to lawyers who can explain the consequences of this or that licensing decision. Cloud providers offer database deployments under license terms that were offered to them. If Redis Labs or MongoDB decided tomorrow to take their database proprietary there wouldn't be much gnashing of teeth. The projects would fork. The proprietary forks would see more feature development and we'd all go on with our lives. You're seeing pushback for a few reasons: 1. The name "Commons Clause" appears to intentionally invite confusion with "Creative Commons." Plus the idea that you can jam a rider onto an arbitrary Open Source license and end up with something coherent is woefully misguided. The GPL explicitly blows up language like this. 2. On the heels of the Commons Clause, the SSPL was put up to the OSI as an open source license even though it obviously violates OSD 5,6 and 9. Along with Salil's rants on TechCrunch, this makes it look like a transparent attempt at undermining open source and the OSI. > In the case of Redis this is a bit a natural process because I'm very unwelcoming to a lot of things that users want because I feel they are not a fit. As an aside, I don't send you patches for this reason.

kjeetgill 8y ago
> The name "Commons Clause" appears to intentionally invite confusion with "Creative Commons." Plus the idea that you can jam a rider onto an arbitrary Open Source license and end up with something coherent is woefully misguided. Seriously. Trust takes ages to build and seconds to destroy. That and good will. I think Redis Lab's wouldn't have gotten quite the push back it did if it wasn't for the perceived shadyness.
antt 8y ago
Creative commons has an NC version. All the video game art I made in my younger days was NC/SA because the freedom to inspect all software you run does not mean you get to sell the software to other people.

Fufnrifnr 8y ago

Linux would not have been successful were it not for the licensing. Copyleft licenses are appealing to a lot of developers. People who are willing to donate lots of time writing free software don't want to feel they're being taken advantage of. Programmers are the life-blood of an open source project, I think developer preference is a huge factor. A BSD might have caught on in the same way eventually but it would have taken longer.

Avshalom 8y ago

>>bout BSD,MIT,... vs CopyLeft. We can say what we want, but if we just stick to facts, MIT-alike licenses provide more freedom for the final user. They provide more freedom for the next user but because they allow freedom to be taken away at any point in the chain bsd/mit licenses provide absolutely no guarantees of freedom for the final user.

ajross 8y ago

> we just stick to facts, MIT-alike licenses provide more freedom for the final user. That's not "fact", that's semantics. You're playing games with the definition of the word "freedom". The FSF has been outrageously clear about exactly what it means by that term, and what freedoms specifically it's trying to protect. Argue with those points. Don't fling around "facts" like this, it's just bad form.

cyphar 8y ago

> but if we just stick to facts, MIT-alike licenses provide more freedom for the final user. It depends what you think a user is. Is a user an end-user who has no intention of creating a proprietary fork? Or is it a developer who intends to create a proprietary fork? "If we just stick to the facts", copyleft provides equal freedom to both classes. Neither can create proprietary forks. For permissive licenses, the latter class is given an additional freedom (creation of proprietary forks), but the former class gains no practical freedom. However, what if the latter class gives the former class their proprietary fork? Then the former class has less freedom (and there is a history of this happening with Windows, OS X, and Android using BSD code without providing it to the end-user). This has always been the argument made by the GPL. Yes, proprietary software developers lose the freedom to reduce the freedom of other users. Instead it guarantees equal freedoms to all classes of users. I get it, you don't agree with the GPL. That's fine. But clearly the "user" being protected the GPL is the end-user. That was always the point (and that protection comes in a form that allows developers to have the same freedom and nothing extra).