What license should work done by Amybo be placed under?

RobJ

I was wondering what licenses people think are best for the work done by Amybo?

As far as hardware goes, I think the CERN-OHL-P or the MIT license may be good options. CC0 1.0 or CC BY 4.0 are also worth taking a look at, depending on how permissive you want the license to be.

Martin

@Gerrit & @danwchan - do I recall your discussing this some time back, or was that licensing text & experimental data..?

Gerrit

I suggested that we use CC-BY or CC-BY-SA for experimental data and text, but I don't think we've landed on which license to use for hardware yet. I would suggest that we use a license specifically created for hardware, and the CERN OHL (Open Hardware License) does seem to be best practice. There are three variants:

CERN-OHL-P-2.0 is a permissive license
CERN-OHL-W-2.0 is a weakly-reciprocal license
CERN-OHL-S-2.0 is a strongly-reciprocal license

I think deciding between those options will depend on the hardware we develop. I'm not against using a reciprocal license if we want to ensure that development stays in the open.

Martin

I think reciprocity (or share-alike in Creative Commons terms) is indeed the big question. While it means that what we develop remains open, it may prevent what we develop from being used commercially. We need to determine whether that is what we really want. At some stage manufacturers are going to need investment in m3-scale bioreactors:

If they're not permitted any secret sauce, how are they going to finance scale-up?
If they get finance - will they ever contribute anything back?

danwchan

I've heard that data is best licensed cc0 . A blog post by the repository data dryad explains that it reduces legal burden. That being said, the paradigm that they are commenting is that the initial publishers of the data have created some sort of derivative work (such as a journal article) which would be the entity with some sort of legal attribution (eg cc-by). So I suppose we have to answer some questions about how we want the data to be used in order to determine the best license.

Martin

Would everyone be happy with dual licensing, with a standard license of:

Hardware: CERN-OHL-W-2.0
Software: LGPL-3.0
Text and data: CC-BY-4.0

And it being made clear that commercial entities requiring more permissive licensing can do so by:

Entering a contract to pay reasonable royalties to AMYBO, or
Regular updates to AMYBO on how people are benefiting from the IP if such IP is used exclusively in LDCs

Martin

Martin We need to raise the tide that floats all boats, as such I was keen on extremely permissive licenses that would let big commercial ships float on our tide too. There is however the argument that such big commercial ships won't contribute back. @Gerrit proposed the compromise of dual licensing.

I'm not saying that the three licenses here are necessarily the best - if not, please do let us know which are better and why.

I get the CC 0 argument, but it's not a license, it's effectively waiving all rights. Until @Gerrit explained, I was assuming a permissive hardware license would give us attribution like CC BY, but he suggested they are more like CC 0 than CC BY. I strongly dislike CC-BY-SA as it prevents commercial use & to me, AMYBO isn't about Open Source for Open Source's sake. We're using Open Source as a means to maximise uptake of sustainable nutritious protein, in order to do so we will need organisations to be able to develop their own IP so that they can get financing to procure things like large ($M) bioreactors.

Gerrit

Not sure we even need dual licensing if we go with weakly reciprocal licenses?

The LGPL is intended to allow use of libraries in applications that don’t necessarily distribute the source of the application. The LGPL has two requirements:

users must have access to the source code of the library
users can make use of modified versions of the library

I think the kind of license will really depend on the kind of thing we make! If we're making a software library, then yes, maybe LGPL is the right choice. If we're writing networked software, then maybe we should use AGPL. If we're creating a small hardware component, maybe it should be CERN-OHL-P-2.0 so that it can be integrated into anything and everything, even without our knowledge.

FWIW, CC-BY-SA does not prevent commercial use, it's CC-BY-NC-SA that does. But I get what you mean - when you're selling something you don't always want to share what you make as well.

Martin

Gerrit So do you think we should go strongly reciprocal with dual license?

With text & data I think attribution (BY) is enough, I think SA as well as NC will inhibit funding of commercial work hence uptake, hence reduce the global benefit. If hardware licenses don't have attribution options, I think more is required. I'd personally be happy if our work was suitably attributed and everyone could get on accelerating mass adoption. Of course financial and technical contributions would be massively welcome, but I'd rather someone got on and made money fixing the planet than didn't do so because our licensing put them off.

Other than data, I think the first thing we made that needs a license was @RobJ's threaded vial cap, a component of our electropioreactor mod, the second was @SandrineH's CO2 canister holder. While I think having a clear licensing strategy will help people decide whether they want to work with us, but knowing the license to apply to those two is most urgent, so we can publish them.

Much of what we use is currently Pioreactor based - some of the licenses they use:

Hardware: GPL-2.0 for the 20ml printable parts, CC-BY-4.0 for the dovetail components
Software: MIT
Text: CC-BY-SA for their privacy policy

If we sequenced a single cell protein genome, would we PatentLeft it?

Gerrit

@Martin I was thinking if we are going to dual license, we should use a strongly instead of weak reciprocal license. Otherwise we're not getting the best of both worlds. I don't think we need a dual license if we want to use weakly-reciprocal license. That said, I think some parts/pieces/projects would be better served by a permissive license. I had a look at the OSHWA Certified Projects List, and here are the current usage of the various CERN-OHL licenses:

CERN-OHL-1.2: 41 projects
CERN-OHL-P-2.0: 55 projects
CERN-OHL-W-2.0: 30 projects
CERN-OHL-S-2.0: 126 projects

My understanding is that OHL-1.2 is also permissive, so weakly reciprocal projects are in the minority. I think we may need to consider each project/part/component separately, and also consider how the main designer/creator want to license their work. Maybe we can agree on using CERN-OHL for hardware, and then leaving the specific variant of the license open to the designer?

I'm happy with using CC-BY for for text and data, yes.

I think a problem with PatentLeft is the cost of applying for the patents. I believe it's better to publish in places where patent examiners may find our work as prior art.

Martin

Gerrit I know that one-size-fits-all solutions are rarely ideal, but still hoped that we could have a standard AMYBO license that we use by default, so we don't need to think too much each time. We then only deviate from it if someone raises an objection or the inventor sees fit. We could then have a page at say AMYBO.org/license that people can link to next to the actual license they are using. That way hopefully nobody writes off any of our IP as too hard to license.

So, how about:

Unless otherwise specified AMYBO IP is licensed as:

Hardware: CERN-OHL-S-2.0
Software: AGPL-3.0
Text and data: CC-BY-4.0

But entities requiring more permissive licensing can do so by:

Entering a contract to pay reasonable royalties to AMYBO, or
Providing regular updates to AMYBO on how people are benefiting from the IP if such IP is used exclusively in LDCs

Then, for example, if you wanted to use the Strong Beerware license on your hardware you could say:
License: Strong Beerware or AMYBO.org/license

I'm not a lawyer but I assume that would mean that people using that IP could either do so by:

committing to purchasing the copyright holder a pint of beer, OR
reciprocating, OR
arranging to pay royalties, OR
arranging to use the hardware exclusively in LDCs and providing regular updates

Gerrit

@Martin AGPL was created for networked software. So are we assuming most AMYBO software will be web-based/networked? If so, then I'm happy with those as the defaults, yes.

Related: Yesterday @RobJ posted a link to a (NGI Zero supported) project that makes it really easy to keep track of licensing: https://reuse.software/

It's basically a tool that scans your project to make sure everything has the right (machine and human readable) license applied. For example, it can check that you applied the CC-BY-4.0 to the images, and that you have the AGPL-3.0 header in each source code file.