Distributed Direct Democracy — Local MVP
Prototype that runs entirely in your browser. Data is stored in localStorage. No server. No signup.
What should we do?
Write a concrete action. Clear, specific, constructive.
Merging uses local TF-IDF + cosine. Threshold 0.50. Tunable in settings.
Top Sentiments
Transparency
Every action is logged. You can export and inspect it.
one-human-one-account (soft) open algorithms local-first
How it works (local)
  1. We keep a corpus of all submissions. Build a TF-IDF vocabulary. Each submission becomes a sparse vector.
  2. Each sentiment has a centroid (mean vector). On submit, we compute cosine similarity to each centroid.
  3. If best similarity ≥ threshold, we merge; else we create a new sentiment.
  4. Score = (up - down) × log10(1 + members) × freshness_decay (half-life 72h).
  5. No network calls. Everything is auditable via the log + JSON export.