Workgroup Proposal
SingularityNET Ambassadors program Archive Workgroup Proposal
Last updated
SingularityNET Ambassadors program Archive Workgroup Proposal
Last updated
Proposal addressed to the SingularityNET Ambassadors program
Andre Diamond - Technical, Treasury Guild
Miro - Treasury Guild
Stephen Whitenstall - Technical, QADAO, SingularityNET Ambassador
Vanessa Cardui - Archivist, Facilitators Collective
Setting up an Archive Workgroup
The SingularityNET Ambassadors programme has been running since May 2022 and is building up a wealth of resources.
It is important to preserve these resources and records, and make them discoverable and usable. This will build a large corpus of data with which the community can develop AI tools and experiment with Large Language Models (LLMs).
This workgroup will archive information about the activities of the SingularityNET Ambassadors programme, and use AI tools as an aid to discover and analyse its corpus.
It is needed because the record of what happens in SingularityNET’s work is part of history; and like any community, we need our history, and it is important that we all have a voice in deciding what forms the record and is preserved. Building a decentralised, transparent and accessible record that we can draw on helps avoid constantly “reinventing the wheel”, and helps us build on what has gone ore.
There are precedents for community-led, collaborative and open-source archival processes, which it is important for us to draw on - but archival processes which integrate the use of AI tooling are groundbreaking, and it is appropriate that SingularityNET should have a work-group which leads on this.
Set up a regular, fortnightly working meeting
Set the terms for participation - GitHub membership, willingness to learn open-source tools.
Offer training-by-doing in using AI and LLM tooling, GitHub, and other relevant tools, for archiving. In other words, our meetings will offer a practice-based, “how to” approach, in order to support onboarding and increase participation. This could, for example, involve working together during a meeting to practically archive something, with input from all attendees and discussion of issues that arise.
Develop documentation standards. Our meetings will discuss what kind of standard information should be collected from a SingularityNET event in order to add it to the archive.
Work on archiving SingularityNet’s considerable backlog of material (this work will be done outside meetings by the work-group team).
GitBook maintenance - particularly the Archive GitBook and the Ambassadors GitBook. The aim is that the Ambassadors GitBook will become a source of truth for what the Ambassadors do, so that it can be used to support onboarding and marketing/publicity.
LLM Development - interfaces with Large Language Models (LLMs) will be developed, explored and tested. With an emphasis on using open source solutions.
We propose to:
track tasks using GitHub project boards integrated with Dework
record task points (hours) on Dework where appropriate
Additionally, pioneer the idea of costing some work elements via a cost-per-task rather than an hourly rate, and integrate this with Dework. This is important because some types of work do not lend themselves to the “piecework” approach of a cost per hour, and a cost for the whole task will better support and value the person who does the work. Also, for increased accessibility - some workers (due to neuroatypicalities or other reasons) prefer a price per task; so exploring how we can do this, integrate it with Dework, and offer it as an option for how to assess and pay for work, could be a useful addition to SingularityNet’s processes.
Which existing Guilds and workgroups would be the closest collaborators for achieving the proposed goals?
Everyone: In a sense, we will be collaborating with the whole of SingularityNet, since our aim is to archive everyone’s documentation. We will be advising all work-groups and guilds on approaches to documenting themselves, and we hope to work with each of them to develop adaptable, flexible processes which allow each work-group to document itself in the way that suits it best.
Specific groups:
Because much of our backlog work will involve archiving Ambassadors’ Guild material, we will collaborate closely with them.
Because an accessible archive will help onboard newcomers and allow them to find out about SingularityNet’s work, we may collaborate closely with Marketing.
Because archiving is a process, we envisage collaborating wit Process Guild
Because archiving has strong connections with governance (it supports transparency and accountability), we envisage collaborating closely with any Governance workgroup that emerges.
What budget would the proposal require?
Budgets will be estimated per Quarter (3 month period) for 3 people (at the outset).
The initial workgroup period is between April 2023 and June 2023.
April 2023 - Preparatory work
May 2023 - Workgroup setup
June 2023 - Workgroup active
Costs will be tracked per top level activity / task - tracked by GitHub issue linked to Dework tasks in the Dework Archive space
Costs and proposal progress will be reported to the Treasury Guild and Incubation workgroup ?
Archival backlog:
Can we suggest a cost-per-meeting to document the backlog?
Ambassador lead will cover some of the cost
Budget for going forward:
Two 2-hour meetings per month, for 3 months = 6 meetings, or 12 hrs, for 3 people = 36 person-hours
Bounties - Do we plan to incentivise others to attend meetings? Even just minimally, to track contribution?
Plus ongoing archiving of new meetings - how many new meetings will we cover in 3 months? Are we planning to do all meetings, or just Ambassadors’ Guild? And will it be the same rate as a “backlog” meeting?
Creating guidance notes on what info a meeting should collect and give us (and maybe running a meeting to share this and develop it with wider input).
Ambassador lead will cover some of the cost
Budget for development work?
Proof-of-concept level of cost - it’s not extensive
Ambassador lead will cover some of the cost
There is a large backlog of documentation and archiving of previous meetings to be cleared, as well as ongoing work; so there is a minor risk associated with finding the right balance between catching up, and archiving new material. We will mitigate this both by gradually onboarding new people, and via a regularly updated work plan to prioritise what is most significant.
The workgroup is quite specialised, so initially it will depend on the skills of a few key people; so there is a risk of working more slowly than planned if any of them are unavailable due to holidays etc. We will mitigate this by using our meetings to focus on developing new people’s skills.
As a result of this proposal, SingularityNET will have the beginnings of an ongoing archive - our own history, told and archived and managed by our community.
This has a practical use - it will give us a transparent, searchable record of SingularityNET’s activities to help everyone keep track of decisions and work.
It also has a community-building use, in that holding and managing our own history, using the AI and open-source processes that we believe in, helps to build a sense of who we are.
The proposal will also lead us to develop our practice in using and applying LLMs in innovative ways, and will offer a case study of using LLMs and testing their effectiveness on a corpus of text.
Lastly, the proposal will enable interesting development work, such as connecting our corpus to open-source tools like HuggingFace and Langchain.