Concept

A social network that asks users one question per day. The goal is to create an AI-assisted online referendum platform that fosters collective dialogue. Users vote on politically divisive or thought-provoking questions, then submit answers. The cycle runs on a 2-day loop:

  • Day 1: vote for the question to ask
  • Day 2: answer the selected question

Over a year, the network can surface and explore 150 key questions shaped by global participation.

Features

  • Users vote for the most popular questions to ask
  • AI groups similar questions and consolidates votes
  • AI clusters answers into common themes or “answer groups”
  • Answers may belong to multiple groups

Existing Tools

Security & Authentication

To ensure 1 person = 1 vote, the system may use biometric authentication (voice, face, ear, nose, or skin image). Users choose their preferred method. Data is anonymized but fully verified. Email is optional; registration requires a biometric method and password.

Question Format

  • Questions limited to 300–500 characters
  • Option to submit 1–3 short questions per day
  • Questions can reappear over time with longer character limits if consensus isn’t reached

Answer Format

  • Text box answers with character limits
  • AI clusters responses into answer groups with shared arguments
  • The structure enables surfacing majority and minority perspectives

Facilitation

  • AI may suggest questions to guide useful conversation
  • Pre-seeded with 20 foundational questions in year one

Sample meta-questions:

  • What topics should be prioritized?
  • Should depth of reasoning affect vote weight?
  • How often should questions be revisited?
  • Should voting require participation in a debate?
  • How long should voting windows last?

Decision-Making

Every decision requires two votes:

  1. Whether to ask a question
  2. How many people need to agree for action to be taken

Consensus mechanisms should be resistant to manipulation, but all voting systems have potential exploits. The aim is transparency and accountability, not perfection.

Monetization

Funding through donations to maintain neutrality. The community can vote on monetization models.

Societies

  • Sub-communities (like Reddit) submit daily questions
  • Societies that reach consensus can share their decisions with others
  • Debate and voting may be restricted to members who meet specific criteria

Role of Language Models

LLMs support:

  • Summarizing diverse viewpoints
  • Identifying root disagreements
  • Highlighting logical fallacies
  • Suggesting alternative framings
  • Detecting semantic overlaps and misunderstandings

This structure turns a simple Q&A system into a scalable tool for collective reasoning.