My view of copyleft and thoughts about copyleft-next

[zefr0x at tuta.io]

I'm a software developer who embraced the FLOSS and copyleft philosophies since my beginnings, every software project that I started is licensed under either GPLv3 or AGPLv3 except for two libraries licensed under LGPLv3. I'm not a lawyer, but I have some experience in writing very simple policies and guidelines for my projects. However, something might be missing or unintentionally against my intent (will be always open to productive discussions).

I'm writing this to provide my vision of what should be the optimal goal of a copyleft license. Also, I have some thoughts and ideas of things that are in copyleft-next that I think should be discussed for modification, along with things that are not and should be discussed for addition.

This should help to develop an intent for this license, as I believe that copyleft-next should not be just a license, but a vision, mission, and goal. As of this, to clarify any thing that I will say, I will start by writing -as possible- about my personal intent.

---

My optimal world of copyleft for software, hardware, and knowledge of creative works in general is that all knowledge belongs to the whole humanity (and any other type of creatures), so no one has the right to restrict any type of knowledge for themself or for a group of people. This doesn't mean that you can't put your name such as the author, designer, inventor, or discoverer. You have the absolute right to be credited in the history. It just means that you don't own the knowledge for yourself. Everyone has the right to do anything with it including (but not limited to) copy, distribute, study, change and improve, but they don't have the right to own the knowledge or restrict it for themself.

There is no exception for any type of knowledge from this rule. However, we must define the barrier between knowledge and applications of knowledge. So, for example software is considered a type of knowledge, while hardware is an application of knowledge. For software, you are required to open the source-code, history of development, design, documentation, user manual, and etc. But for hardware you are required to open the design, design history, list of materials, manufacturing process, user manual, and etc. The word "knowledge" is very general and this is intended.

This might be off-topic, but this rule is applicable for digital media, books, research papers, digital art, and anything as long as it begin a knowledge. Some people might oppose this with the argument that it stimulate creativity, but I believe that creativity has two pillars: resources (off-topic) and knowledge (on-topic). For projects aimed to provide an application of knowledge there is not a problem, since they can be backed with resources from anyone interested in the application. But for projects aimed to provide knowledge by itself the passion of its developers will be the main driver, and I believe this was always the case with or without copyright. The problem is about finding a sustainable framework to provide resources for creative projects aimed for knowledge, and passionate creative development will always solve this problem. Additionally, the copyleft will ensure that who build applications of knowledge is the best one with the best resources, not the one who restricts the knowledge.

So the knowledge will be open, and the competitiveness will be mainly derived by resources and building a framework to obtain it. They might be mainly driven with passionate human resources, or anything else. I don't know what the optimal framework for any project, maybe it is by having both copyleft and copyright projects igniting each other? or is it both knowledge driven and application driven projects igniting each other? You can have resources with multiple methods, but the knowledge is one.

I see copyleft licenses as a method or abstraction to run this optimal world on top of the copyright and intellectual property systems adopted in the current world. So what I want to see is that by using the copyleft-next license for your project or contributing to copyleft-next licensed projects you are committing to those ideas of an optimal copyleft world no matter what is written in the license, since it is just an abstraction to fit into the current legal system. So I would like to see copyleft-next as the new license with the exact previous intention of having this optimal world of open knowledge in the current legal system.

---

Now lets go throw the latest draft of copyleft-next:

"""
9. Copyleft Sunset
The conditions in sections 4 through 8 no longer apply once fifteen
years have elapsed from the date of My first distribution of My Work
under this License.
"""

This is currently being discussed in another thread, and I believe that it oppose my vision of copyleft, since it make a restriction earlier than the copyright legal system and not just fitting into it.

"""
11. Later License Versions
The Copyleft-Next Project may release new versions of copyleft-next,
designated by a distinguishing version number ("Later Versions").
Unless I explicitly remove the option of distributing Covered Works
under Later Versions, You may distribute Covered Code under any Later
Version.
"""

There must be restrictions on "Later Versions". Can they go away from the copyleft concept? or are they required to just make copyleft stronger? Can they left any restrictions from previous version?
The license can change, but there must be an explicitly defined vision of the Copyleft-Next Project that will never change to maintain trust.

"""
15. Definitions
"Source Code" means the preferred form of a work for making
modifications to it.

[...]

CCS means: (i) the Source Code form of the Covered Work; (ii) all
scripts, instructions, know-how, and similar information that are
reasonably necessary for a skilled developer to generate such Object
Code from the Source Code provided under (i); and (iii) a list clearly
identifying all Separate Works (other than those provided in
compliance with (ii)) that were specifically used in building and (if
applicable) installing the Covered Work (for example, a specified
proprietary compiler including its version number). OCS must be
machine-readable and, to the extent that its licensing is not governed
by section 2 of this License, must be available under terms satisfying
the Free Software Foundation's Free Software Definition
(https://www.gnu.org/philosophy/free-sw.en.html, version 1.135).

NSS means: (i) the Source Code form of the Covered Work; (ii) all
scripts, instructions, know-how, and similar information that are
reasonably necessary to use the Covered Work to substantially
replicate the Network Service, and (iii) a list clearly identifying
all Separate Works that are specifically used in or required for
providing the Network Service (for example, a specified web server
including its version number).
"""

Since I have the freedom to modify Source Code, and not just to generate Object Code or replicate a Network Service, I expect to have reasonable knowledge to modify it. This is by having development scripts, architecture/design documentation, development history, and a list of development tools. Without this type of knowledge included, even a skilled developer will need a huge work of reverse engineering of the design and the way that the project was developed to work on it properly. Additionally, I would expect documentation and specification of the language used to write the Source Code, even if the compiler was proprietary. This will covers the freedom to study the Source Code and maybe to create a compiler by yourself.

Also, with the freedom to run the software, there must be a list of runtime dependencies and know-how. It's not necessary to have documentation of how to do every specific thing using the software, but I will need to know where and how can I run it and if I need a special hardware to do so etc.

As for the lists of Separate Works, other than the name and the version, I would expect a URL where I can have access or a communication channel to request access.

---

Now I will write about things that I think should be included in the copyleft-next:

1. Learning and Reuse of Knowledge

There will be no problem if everything was just copyleft, but this is not the case. So to what extent can I learn from copyleft project and then use my knowledge to build a copyright one or another with a permissive license?

In my view, Source Code is not just code, but it's knowledge including algorithms, hacks, methods, design, and etc. So when you use this knowledge in your project to an extent out of Fair Use (not necessarily just coping portions of code), your work is a derivative. With the same concept, when learning from an API or a GUI to create a clone of my work with reverse engineering, your code is not derivative, but the overall interface and the design of the overall software is derivative and you must release it (the design or the interface) with copyleft license, proper credits, and notice. This should prevent things like creating an MIT licensed clean room Rust rewrite of a copyleft project and having any derivative design decision licensed under the MIT.

>From the same view, when dealing with LLMs (AI), it is just about learning and reuse. Yes, it is faster, but this is not an argument to treat it differently from humans' learning. It might be weird and humans often fear or oppose new things that they don't understand, but I will not blindly just ban AI training, since this will be loudly against the core concept of copyleft. So what I see as the best solution is to just consider the portions of AI models that learned from copyleft knowledge as derivative works (probably can't be enforced when mixing data with different licenses in one model, so we might restrict that?), and -to an extent- content generated with these models are also derivative works whether it was code, design, hacks, algorithms, methods, or etc when they are creative work. However, if this can't be enforced in the current legal system, I will go with just opening the knowledge without any restriction. Knowledge by default must be open when we can't legally enforce copyleft, otherwise we will just be hurting people who are building open LLMs (bad people will do it anyway). 

Just to clarify that, I'm not against training AI with anything just like humans do learn (knowledge should be open), but it is not a copyleft only world and most companies will try as they can to get as much data and then just restrict it. In my opinion the bad thing is not that they are training AI on my data, but it is about not releasing the models for the public. Additionally, private personal information should be anonymized, since even for humans it is not ethical to access these.

2. About Jurisdictions

I noticed that things might not be included in this license since it might be a common sense or not enforceable in USA or EU court. I believe that we should focus on making the copyleft-next as global as possible and as applicable as possible even in jurisdictions with few experiences in copyright cases or with different copyright concepts.

For example, to my knowledge I don't think that there is a clear definition of "derivative work" in the current draft.

However, I don't know how much work has been done here, since I'm not an expert of any legal system.

---

Summery and What is Expected:

- All knowledge belongs to the whole humanity, so no one has the right to restrict any type of knowledge.

-The Copyleft-Next Project should have a vision, mission, and goal to act as a definition of an optimal world of copyleft and how to achieve it. This will also help to trust the Later License Versions clause.

- The Copyleft Sunset clause is against the optimal world of copyleft.

- Definitions of CCS and NSS should be extended to include knowledge that are necessities or facilitation of the other freedoms, not just to provide the code and the knowledge to build it, but also the ability to run, modify, and study it.

- CCS is not just bare code, but a knowledge, and any knowledge of creative work done using it should also be under the same copyleft license.

- When learning and reusing to an extent, the produced content should be considered as derivative, not just by copying code, but also by creating a separate software with separate code but with a derivative design, algorithms, hacks, or etc. The code by itself will be Separate Work, but the design and any other knowledge of creative work will be derivative and under copyleft.

- The previous point includes AI training, so the model and anything it produces to be derivative and under copyleft.

- Make copyleft-next as applicable as possible in jurisdictions with different copyright systems.


P.S. Sorry for writing a lot in a single email. I think separate topics can be discussed in sub-threads :)

zefr0x