Associació Decidim - Reunió Comité de Coordinació

Time: 17:00 CET

Attendees:  Ali, Txema, David, Antonio, Pauline

Guests: Nil, Carol

Agenda:

1/ Proposed Jigsaw-Google AI Tooling

2/ Salary scales

3/ Local Proposal Chapter Switzerland

4/ Elections Coordination Committee to fill 3 vacancies

5/ Decidim Fest

6/ To initiate the process of drafting the Internal Regulations

---

1/ Proposed Jigsaw-Google AI Tooling

We have to discuss the proposal that Jigsaw-Google has made to us and make a decision.  Here is all the information we have from Jigsaw:

To recap, we're developing an open source library, leveraging state of the art AI models to reduce the cost, time and complexity of extracting meaningful insights from online deliberations, allowing them to be deployed more widely and at greater scale. This tooling intends to improve the sensemaking of outputs from deliberations including better summarization and categorization. For your reference, we have put together a brief overview for you that summarizes this work and our overall goals. Please see the attached document for the background information.

Regarding your specific questions about the models and license, we can connect and support any model in Model Garden on Vertex AI, which includes Gemini 1.5 Pro and Gemma. Although some are not yet directly supported in the library, we can add additional models with low effort. The license that we are using is the Apache License version 2, which is the Google standard. If there are any challenges or concerns regarding the Apache license, please let us know, and we can discuss it further.

Additionally, we currently offer a Node JS library to integrate the summarization and categorization features. We understand that your tech stack is Ruby, so if this may be an integration blocker, please let us know, and we may have a few suggestions that can be explored.

As we mentioned, since we know you have other milestones on your current roadmap, we'd be open to supporting you financially to help fund resources to support an integration and potentially support your research as well. If this might be interesting to you, we'd love to see how we can define a collaboration plan to integrate and test these new features with you to understand how they might improve deliberative processes.

Is this enough information for you to share this potential integration collaboration with your governance team? Please let us know if there might be any additional details that we can provide. If this information is sufficient, it would be great to know when you might be able to get feedback on this opportunity.

PDF Attachment: Participation Sensemaking Overview - October '24.pdf

Discussion notes

Carol: Do we want to use AI? Yes, go ahead and experiment.

We will always have to warn the participant that the content is AI-generated. We have to think of concrete use cases that add value. I would rely on criteria from industry experts (recommendations in slides 55 and later): https://faculty.washington.edu/ebender/papers/Bender-GRAILE-2023.pdf 

Ali: can't think of a use case with respect to abstracts because of the nature of the proposals. If it is to leverage funding, translation and moderation bots would fit us, it would be a matter of seeing if we are in a position to negotiate. Also the possibility of a process open to the community.

David: Doubts about the info: They only have one library in Node.JS Have we thought about how to integrate it with Ruby?  If we are going to rely on a proprietary intermediary (Google's Vertex) it doesn't make sense. He sees possible use cases like process summarization. 

Antonio: We need to think about the costs and benefits of moving forward with this proposal. Consider the value that AI can add in Decidim. Then there is the issue of money. We can make a counter-proposal to Google, asking for money. 

One cost to consider is the vision and image of Decidim, by collaborating with Google. A possible way out would be to use the enemy's dough against them.
Take the debate to the community.

Txema: He doesn't see it either way. There are two questions:
- Are we interested in experimenting with AI? Yes, but we should do it ourselves, knowing the technology.
- What is the energy efficiency of this model? We are feeding the beast.
Make a specific model for citizen participation.
It is interesting to start experimenting, but first we have to think about what AI model we want.

Nil: Define the red lines and try to negotiate, let's see where your limits are.
For us the priority is: Avoid vendor lock-in and think about specific use cases: Accessibility, Translations and spam.

Antonio: Let's try to tighten the rope, asking for maximums and see where its limit is.

Carol: We can go down the line of inquiry. They think they're the best looking in the class. 

Points to negotiate:

• Vendor lock-in: We want to make something interoperable and open.

• Research: would there be money for that? Can we have an agency?

• How much money for development? 

• Use cases: Translations, Accessibility and Spam.

• Management fee

Agreements:

• Let's respond to Jigsaw 

• We will draw up a strategic document on our common position

2/ Salary scales 

We postpone the discussion of this point due to lack of time.

At OT we have been working on a proposal for salary scales. The objective is to begin to order the different labor categories we have, as well as to be able to offer workers a horizon of growth. The idea is that the salaries of each worker follow a logic that is shared and known by all.

You can read the proposal here: Proposta d'estructura salarial.docx 

3/ Local Proposal Chapter Switzerland

We postpone the discussion of this point due to lack of time.

The people of Switzerland have made us the proposal to create a Local Chapter in Switzerland. 

You can see here the proposal in Metadecidim: https://meta.decidim.org/assemblies/coordination-committee/f/1799/proposals/17859 

Here is the full text of the proposal: https://octreegva.notion.site/Proposal-for-the-creation-of-a-local-chapter-for-Switzerland-V0-2-12212ef1253c806bbf67c72cf373cd0a?pvs=4 

4/ Elections Coordination Committee to fill 3 vacancies

• Dates were adjusted

• An AMA meeting will be held on November 12 at 4:00 p.m. CET to receive questions about joining the committee.

5/ Decidim Fest

We postpone the discussion of this point due to lack of time.

Evaluation of the Committee.

6/ To initiate the process of drafting Internal Regulation

Scope

Work previously done: https://docs.google.com/document/d/1678VypCOUD7dC6nLu_5i8n4U5KCwddpu_EQ750-yS5U/edit

1. Scope of application

2. Rights and duties of the members

   • Member definition

   • Membership fees

3. Registration and deregistration from the Associaiton

4. Coordination Committee

   • Election process

5. Other Committees (?)

6. Decision-making process

   • Political statements

7. Technical Office

   • Workers rights

   • Executive direction

8. Consultative bodies

9. Technological governance

10. Partnership policy

11. Internationalization and Local chapters

12. Conflict resolution and sanctions

    • Code of conduct

13. System of incompatibilities

Ali Proposal:

Model a process on the Citizens Assembly methodology, but open to the community:

• Stage 1: Talks facilitated by experts in both Decidim and DGP governance. Module conferences.

• Stage 2: Proposals. In order for the community itself to discuss the topics of the text, a stage of proposals is opened. Proposals are discussed, merge, etc. and the resulting proposals form sections of the text.

• Stage 3: Drafting committee. A committee formed by the community collects the proposals in a draft. 

• Stage 4: The process described by Nil above but shorter. That is, a presentation of the text and goes through at least one round of comments.

Logic behind this:

• Working on the text alone can leave us with blank areas not contemplated. The intention of including conferences and proposals is to add to the process moments in which 1) through the conferences the community has the opportunity to get rid of the blindness of the decidim workshop, seeing good practices in other projects and using the insight that the community of decidim scholars has gathered in their research on the project 2) using the proposals module allows issues that are perhaps more general concerns or topics and that may not be ready to be grounded in a text (e.g. chapters in other countries) to have their own deliberation processes.

• Having other modes of participation, beyond working directly on the text, makes the process more accessible. The text is a high barrier. Ideally we want more participation.

• Using insights from the community of academics who have studied decidim is not only a smart way to use that knowledge, but also allows us to involve that community in the process, and at the same time to appreciate the value of the knowledge we have gained.

About the link with People Powered

The People Powered grant is about making public policy in a participatory way. We initially presented the idea of getting a government that wanted to release a digital public good and make it public policy. After studying the issue, I am proposing that this be the PPA project. 

• A vital part of a public policy of this kind is to ensure that the

Agreements and tasks

• Nil: Open the process in metadecidim

• Nil: Open Element room and convene first Drafting Committee meeting

• We merge the designs: Nil Write the new process design 

• Open a discussion component to collect proposals