[SD-3] MVPR Architecture: Generic #3
Labels
No Label
discussion
draft
No Milestone
No project
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: DGF/dao-governance-framework#3
Loading…
Reference in New Issue
No description provided.
Delete Branch "%!s(<nil>)"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Goals:
Context:
This system has enough complexity that there are many possible starting points for an approach to implementation. We suggest that a high level architecture diagram can help navigate this decision landscape.
Diagram:
Sequence.org diagram
Figma
Documents:
DeSciPubDAOArchit22July19.pptx
MVPR Parameters
Main Contract
Forum Contract
Mindmap of Parameters:
calcaterra2018-On-chain-Governance-of-Decentralized-Autonomous-Organizations_withMarginNotes.pdf
Layers of smart contracts
changed title from MVPR Architecture to {+[SD-2] +}MVPR Architecture
changed title from [SD-{-2-}] MVPR Architecture to [SD-{+3+}] MVPR Architecture
changed the description
Aaron: Question, do reviews get a score like articles do?
Jonathan: No, because we want them to be double-blind
Article data structure:
Title/Address
Contributors, weighted by contributions
AREP - Derived from citations and reviews
Citations away - Weighted list
Article data - Artifact - Content and/or Hash and/or External Reference
In this context, we agree to assume all operational data will be on-chain. Off-chain, non-operational shall include things like article content.
Question, will some posts live on-chain and some not?
To what extent do we want or need to encode our own data history vs relying on records provided by the underlying blockchain storage?
In regards to science publishing, I limit the referrers and references (q5,6) by requiring annotated bibliographies that will be actively reviewed and weighted more heavily than context citations. I also putting a cap on the amount of reputation allocated to work cited so that citation mills will have diminishing returns. Hard coding the parameters works, but improving practices within the system is preferable.
changed title from [SD-3] MVPR Architecture to [SD-3] MVPR Architecture{+: Generic+}
mentioned in issue #4
Technology Stack Overview
How will
Watch
be implemented? Maybe something along the lines of https://github.com/Neufund/smart-contract-watchWe would just need a node to spin up, read current blockchain state, and then standby to monitor for changes and issue notifications.
Ok so the base protocol should probably support implementations adding custom parameters that map to the base parameters
c1...
,q1...
What is the process by which implementers will enumerate their domain-specific considerations?
Different groups have different norms, expectations, aesthetics around how to express notions of truth. E.g. composability. How can ideas be connected with one another; how precisely can the relationships be expressed? In what way are claims context-dependent, and in what ways can they be abstracted across contexts?
Enable each DAO to mint reputation corresponding to each category of reputation that its members are able to identify.
This means any given post should be able to stake a claim for 0 or more denominations of reputation. Other members of the DAO should then have an opportunity to affirm or refute these claims. The resulting graph can be evaluated with weights given to each reviewer's own reputation.
Weight of claims can change over time. We should identify the process/mechanism for each way they can change. For example a comment gaining reputation because other members find it valuable.
Diagram distinguishing
Proposal
andProposalReview
. ReusingEdit
andComment
as generic elements.Diagram focusing on AREP and RREP
Diagram focused on GREP
I might have misheard the problem. Reviews will also get scores. In fact Reviews are the heart of MVPR. Anyone can make a post, however its real value should be based on how well those posts survive validation processes. Also I think we also talked about identity regarding the topic.
Regarding double-blind reviews, this is where we also need to rope in different identities to prevent collusion etc. Having ones reputation at stake on the paper being reviewed is another form of accountability that would deter collusion.
In the whitepaper I wrote, I suggested seasonal governance where the scoring of reviews and articles will oscillate like a predator prey model. Sometimes the ecosystem needs to be more innovative, sometimes it needs to focus on validation depending on the "health" of the ecosystem - ie. if the ratio of negative citations, peer review, and replication studies raises above a threshold, the parameters will shift (tbd by group) to value replication papers more.
Sequence diagram for interactions among smart contracts and internal/external user interfaces
Sequencediagram.org diagram
Current expectation is that Proposals and ProposalReviews will both utilize GREP. This is a distinction between the Work Product process and the Proposal process; in the Work Product process, AREP and RREP are separate.
Science Peer-Review Typology as a real-world example to help with review abstractions as it relates to different identities.
Sequence diagram for interactions among smart contracts only; specifically voting and forum.
Sequencediagram.org diagram
Updated sequencediagram.org diagram
Updated diagram for possible validation pool and forum interactions
Sequence diagram
changed the description
changed the description
changed the description
changed the description
changed the description
changed the description
changed the description
Craig had mentioned problems with the long-term sustainability of the MVPR/Semada system if "old" reputational value doesn't decay with time. The q7 hyperparameter can address this problem, but new ideas are welcome.
Our current solution is to still maintain a history of reputation but to differentiate between active and passive reputation. Active reputation relates to reputation salary but its value will decay or have a strong cutoff over time converting to passive reputation. This means people can still see a person's history yet continue to be engaged in earning a reputation salary.
Example: A newcomer to a DAO brings in $100 to a mature MVPR/Semada-based DAO. Without the reputation decay factor, the $100 that low-reputation newcomer brings into the system would be largely distributed to the senior members of the DAO. By including a decay/cutoff transition from active to passive reputation, the total accumulated reputation can still elicit outside funding based on total reputation accumulated while providing a power distribution mechanism that is amenable to short-mid-long term engagement.
https://sciencepublishingdao.slack.com/archives/C03PF04MHL2/p1662744242137759
mentioned in issue #5
Preparation for DxD Admin grant proposal
Summary
The outcome of this work shall be a prototype of an MVPR DAO.
Under this model, an MVPR DAO is comprised of a group of experts in a given discipline. The DAO will enable these experts to do the following:
Architecture
This MVPR DAO prototype will consist of the following elements:
The smart contracts constitute the hard protocols of the DAO.
The off-chain server, client software, and user behavior constitute the soft protocols of the DAO.
Smart contracts
Off-chain server software
Web client
Technical Architecture
Authentication pathways
Application code delivery
Read
Write
Verify
Task-based, revenue-generating work
Expert stakes REP to register availability
Public submits work request with fee
Expert submits work evidence
Peers validate the work evidence
Rewards are distributed
Internal operations
Expert starts new DAO
Expert joins existing DAO
Expert submits governance post
A governance post can be considered one that is respected in some way by the operations of the Client/UI.
This is a broad class of posts.
Each type of governance post can be individually (or by category?) handled by the business contract for a given DAO.