Decentralized reputation system. #13

Open
opened 2025-11-28 17:20:30 +00:00 by warren · 2 comments
Collaborator

Can we create some kind of decentralized reputation system?

A major possible issue with anonymous profiles is ban evasion. If any profile can join any conversation context, banning a profile does nothing to prevent re-entry through a different one.

Example use-case: members of vulnerable minorities would likely wish to prevent interactions with harmful actors, ideally through community-based efforts to identify trustworthy parties.

Ideally, we’d have some reputation system, ideally managed in a decentralized fashion and which does not necessarily identify users using real identity. External parties unrelated to an exchange between two users should ideally not be able to extract data that they may abuse.

Can we create some kind of decentralized reputation system? A major possible issue with anonymous profiles is ban evasion. If any profile can join any conversation context, banning a profile does nothing to prevent re-entry through a different one. Example use-case: members of vulnerable minorities would likely wish to prevent interactions with harmful actors, ideally through community-based efforts to identify trustworthy parties. Ideally, we’d have some reputation system, ideally managed in a decentralized fashion and which does not necessarily identify users using real identity. External parties unrelated to an exchange between two users should ideally not be able to extract data that they may abuse.
Author
Collaborator

Ideally, trust and reputation should be slow to gain and quick to lose; this should make ban evasion through creating multiple profiles rather costly to a bad actor.

I propose a local graph-based web of trust approach.

A profile may have a public/private key pair. Profiles serve as nodes in a graph. Profiles may vouch for other profiles, signing such vouchers with their private key, forming edges in a graph.

A “trust score” may be calculated for a newly-encountered profile based on graph distances.

Stumbling blocks: how does decentralized graph search work? Can we minimize exposing additional information about the graph beyond distance?

How does discovery of new profiles work anyway?

Ideally, trust and reputation should be slow to gain and quick to lose; this should make ban evasion through creating multiple profiles rather costly to a bad actor. I propose a local graph-based web of trust approach. A profile may have a public/private key pair. Profiles serve as nodes in a graph. Profiles may vouch for other profiles, signing such vouchers with their private key, forming edges in a graph. A “trust score” may be calculated for a newly-encountered profile based on graph distances. Stumbling blocks: how does decentralized graph search work? Can we minimize exposing additional information about the graph beyond distance? How does discovery of new profiles work anyway?
Author
Collaborator

I believe we should emphasize social methods for this over technological ones. Prefer analogies to real-world social trust and reputation systems over huge technical approaches.

Perhaps avoid mechanisms like proof of stake/burn/work (though a little bit of PoW might be useful to avoid creation of profiles on truly huge scales)

I believe we should emphasize social methods for this over technological ones. Prefer analogies to real-world social trust and reputation systems over huge technical approaches. Perhaps avoid mechanisms like proof of stake/burn/work (though a little bit of PoW might be useful to avoid creation of profiles on truly huge scales)
Sign in to join this conversation.
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
nemalixi/chat-app#13
No description provided.