CypherGraph

Discussion Graph Traversal

The primary mission for this project is to make it possible for people to share information science actions. To make it easy for them to publicly describe, demonstrate, and encode mechanisms for curating information to resist disinformation and toxic infectious memes.

Graph traversal becomes a super-important element in those mechanisms.

Consider a message m, which is a node in a discussion graph D. There are various cases where I would want to incorporate or avoid that message.

Incorporate

• Root Nodes m, a vector of all the nodes m that make up the first level of discussion. To be included they would need to satisfy an inclusion criteria.
  • The node itself is desirable.
  • A descendant node is sufficiently interesting that it merits inclusion of this node to anchor the branch.
• Child Nodes c, a collection of nodes that have a parent, but are not root nodes.
  • Node Itself: The node itself is desirable.
  • Descendant Node: A descendant node is sufficiently interesting that it merits inclusion of this node to anchor the branch.
  • Ancestor Node: An ancestor node becomes worth of inclusion only with this added context.

Traversing:

If I start from a node m and want to decide whether to include its parent branch, under the "ancestor node" rule above, I might consider the following:

If node m is coded with the "supporting data" tag, and the parent node is mildly interesting but a citation is needed, m might justify inclusion of both nodes.

If node m is coded with the "counterpoint" tag, and the parent node is compelling but in need of challenge, m might justify the inclusion of both nodes (and perhaps suggests the inclusion of sibling and niece/nephew nodes).

GPT Thoughts on Reactions

Positive Tags

1. Insightful: For comments that offer a fresh perspective or deep insight.
   1. light bulb, 1257, U+1F4A1, 💡
   2. 69, U+1F92F, 🤯, exploding head
2. Cited Source: For comments that provide reputable sources to back their statements.
   1. graduation cap, 1188, U+1F393, 🎓
3. Well-Structured: For arguments that are logically structured and easy to follow.
   1. 865, U+1F9F1, 🧱, brick
   2. 864, U+1F3D7, 🏗, building construction
4. Open-Ended: For comments that invite further discussion or pose thought-provoking questions.
5. Empathetic: Recognizing comments that show a deep understanding or empathy towards another viewpoint, even if they don't agree.
   1. people hugging, 547, U+1FAC2, 🫂
6. Balanced View: For comments that consider multiple perspectives before drawing a conclusion.
7. Fact-Based: Comments that stick strictly to factual information.
   1. 339, U+1F575, 🕵, detective
8. Constructive Feedback: Positive suggestions or feedback on a particular idea or argument.
   1. 346, U+1F477, 👷, construction worker

Negative Tags

1. Fallacy Used: For comments that employ logical fallacies in their arguments.
2. Ad Hominem: For comments that attack a person's character instead of their argument.
3. Unsubstantiated Claim: Claims made without any credible source or evidence.
4. Over-Generalization: Making a broad statement without sufficient evidence.
5. Anecdotal Evidence: Relying heavily on personal experiences instead of broader evidence.
6. Emotionally Charged: Comments that seem to be driven more by emotion than reasoned analysis.
   1. 103, U+1F621, 😡, enraged face
7. Off-Topic: Drifting from the main topic of discussion.

Neutral/Informative Tags

1. Seeking Clarification: When a user asks for more information or clarity on a point.
2. Background Info: Providing context or background information relevant to the discussion.
3. Experience-Based: Sharing a personal experience related to the topic.
4. Hypothesis: Proposing a theory or idea that hasn't been substantiated yet but adds to the discussion.
   1. thought balloon, 167, U+1F4AD, 💭
   2. 35, U+1F914, 🤔, thinking face

Considerations

• Guidelines: Provide users with guidelines on how to use the tags appropriately. This can prevent misuse or over-tagging.
• Feedback Loop: Allow users to see how their comments are tagged by others. This can be a learning tool for them to understand how they can improve their contribution to the discussions.
• Avoid Gaming: Just like with upvotes/downvotes, there's a potential for users to "game" the system. Implementing a system where only a limited number of tags can be applied per user or per comment might help.