💡 Moderation Functionality

[ValMBO]

Area

Proposal Pillar

Is there new design needed?

Yes

###Objective

Ensure healthy, constructive, and inclusive discussions within the Proposal Discussion Forum by introducing a moderation system that promotes engagement while discouraging disruptive behavior.

Why?

• Encourages constructive dialogue by discouraging inappropriate behavior.

How?

Key Features:

1) Flagging System:

• Allow users to flag comments or posts for inappropriate content (e.g., spam, harassment, or off-topic discussions).
• (Nice to have) Add a simple menu (e.g., "Flag as inappropriate") to each post/comment.

2) Automated Warnings:

• Implement an automated system that sends gentle warnings to users flagged multiple times.
• If a post reaches a predefined number of flags (e.g., 3), automatic actions should be follow, and temporarily hide flagged content with a notification like "This post is under review due to community reports."

3) Provide Guidelines:
-Display forum guidelines to remind users of acceptable behavior.

NOTE: For the MVP, we aim to avoid introducing a moderator role and instead prioritize an automated system.

[ValMBO]

Hi @nebojsact - I have updated the ticket with the main requirements. I’ll leave it to you to add the two approaches we discussed during our call.

[nebojsact]

Hi, as we discuss, i like to add that we can try automate process using AI but currently i have no idea about cost of that kind of services, all other main points you already describe very well :)
What we didn't discuss but i just get it on my mind, what will be happened if some comment in this approach Flagged as inappropriate and we need to remove flag? Without some moderator role we will not be in able to do this... just topic for discussion.

[l-br1]

I think the code of conduct to be followed in the forum, is a great idea. For moderation, perhaps we just have an option in the comment dropdown to 'flag inappropriate comment', and for now that simply comes to us, that is probably enough. Then we can check them weekly and bring the decision to the working group on how to action

[ValMBO]

Great question! I just discussed that with @l-br1 and and he mentioned that we could get support from the GovTool Working Group. This way, they could take ownership of the moderator role and decide how to handle each flag.

[spannercode]

In any moderation system, it's crucial to ensure that content and comments are linked to legitimate identities. Sybils—fraudulent identities—can disrupt messaging and lead to unintended outcomes, especially in automated moderation. While manual moderation is an option, it often lacks scalability. Here are some of my first thoughts:

1. Weighted Flagging: Assign a weight to each user's flag based on their stake, as each user identity is associated with a stake key. This approach minimizes the impact of cheaply created sybils, as their flags would carry little to no weight.

2. Minimum Stake Requirement: Additionally, require each user to maintain a minimum stake of, for example, 1 ADA in their wallet. While this will not entirely prevent spam, it increases the cost for spammers to create sybils, thereby reducing spam incidents and lessening the burden on other moderators.

Since the Proposal Discussion Forum (PDF) operates off-chain, spam prevention measures must be enforced at the backend. Frontend checks using browser wallets alone will likely not be sufficient.

[ValMBO]

Hi @kickloop - Would you mind reviewing the solution proposed and add your comments?

[ValMBO]

Summary of the Meeting

User Interaction:

• Users can flag inappropriate/concerning comments by clicking a flag icon.

Escalation Process:

• If a comment is flagged by three different users (threshold: 3 flags), a notification will automatically be sent to the assigned reviewer by email. On that email It will appear 2 options with their own link: Deleting, or unflagging
• The flagged comment will then be escalated to the GovTool Working Group, which will collectively decide on the next steps:

1. Delete the Comment: If deemed inappropriate or harmful.
2. Unflag the Comment: If determined to be flagged in error.

Security Measures:
A single level of security will be implemented for individuals tasked with deleting or unflagging comments.
This security mechanism will have a form of hashing.

I am also creating a new ticket for Sandip's comment above.

@nebojsact @bosko-m @spannercode