Key Takeaways from the MozFest session

Let’s use this topic to provide feedback and add any comments or thoughts to the summary we at OKFN made of the first session to start rethinking the Open Definition.

Full version in the blog:

Key Takeaways here:

  • Diversity – It was recognised that the original Open Definition process was mostly carried out by people with a fairly high-level profile, but little diversified.

  • Governance – It is necessary to design a new governance model for the Open Definition to seek an even greater consensus than before. There is a need to actively and radically include people from other origins, races, genders, classes, etc., and in a way that everyone feels a constituent part of the process.

  • Misuse – One of the biggest problems when it comes to open content today is the misuse and abuse of the word open, used to describe technologies and contexts that actually do not satisfy any of the criteria defined by the Open Definition. Participants mentioned the need for mechanisms for reporting misuses, or how the definition could have a more supervisory/validating role.

  • Ethics – There was a discussion about the term “for any purpose” which, according to the current definition, is an essential part of what makes content open. Some arguments revolved around the concepts of “responsible use” (like in Responsible AI Licences), or “do no evil”.

  • Universality – There were also debates about the universality of the concept. Some argued that there should be a single generalised definition, while others pointed out the need to make the definition always dynamic and context-related.

  • Language – Many pointed out that the Open Definition should abandon jargon, legal, economic and technical vocabularies to adopt a more accessible and easy-to-understand language, especially for those who are not familiar with the concept.


I haven’t looked into the complex question of “AI” in this space, but I think the push for usage restrictions in copyright licenses is tactically misguided. It’s as though people saw that there was a copyright hack in copyleft that made copyright do things differently from what was assumed before and so they assume we can hack anything at all into copyright.

We need to deal with responsible-use questions, but I think they need to be at the level of direct regulations and enforcement. We get nuclear arms treaties by working directly on treaties, not by copyright enforcement on atomic science.

In practice, usage restrictions put into copyright licenses are likely to undermine the success of openness and create all sorts of “open for uses that I approve of” sorts of gatekeeping which really can never lead to actual openness and which create permission-chains that get unwieldy. There’s little chance of getting the restrictions just right to draw clear lines about usage. And licenses are not adaptable in the way that laws are.

We can have narrower definitions within an overall one. If we don’t have a concrete universal foundation, we will be unable to deal with the misuse issue.

Creative Commons has a general lay-reader summary of the legal licenses. The OD could work like that with the formal OD being clear and precise (and as little technical jargon as possible) and have a more readable summary as well.

Hi @wolftune for me the question: do open and responsible copyright licenses make sense is a fundamental one, on par with previous debates on issues like free/open licenses, etc.

And while the issue is being raised, I haven’t seen attempts to address it, through some collective discussion (there has not yet been any event, any public discussion between stewards of different licenses). Seems to me that the Open Definition and its participatory governance body (the Council, or some new form) would be in a great position to organize such a conversation.

I for one would prefer not to immediately dismiss the new licenses as no open. Especially that they already exist, so that they’re not hypothetical

(By the way, we did some research on how these licenses are used on HUggingFace for licensing AI models and datasets, I hope it’s an interesting data point for this conversation:

@AlekTarkowski the link to the HuggingFace research is missing or broken?

It would be great for this to start with a survey of open data and open definitions in the wild.

On open data: Who is claiming to be open data? What challenges have they faced? What legal approaches do they use? Are there obvious gaps of data that should be open but isn’t for some reason?

On open definitions: what other ones are there? How many are actively maintained/supported? What are the most successful ones?

That’s something it’d be great for OKFN to bring some organizational resources to bear on, so that these other questions are not asked in a theoretical vacuum.

I might have forgotten to paste it.
Here goes: