# ArguCheck

> ArguCheck is a structured debate and argument mapping platform. Users post claims (debatable propositions), build pro/contra argument trees, rate truthfulness and relevance, and an evidence-based scoring algorithm surfaces the strongest reasoning.

## What ArguCheck Does

- Users create claims (statements that can be true or false)
- Each claim can have pro and contra sub-claims forming a recursive argument tree
- Community rates each claim's truthfulness on a -2 to +2 scale
- Community rates each argument's relevance on a 0-100% scale
- A Bayesian scoring algorithm combines truth ratings + argument influence into a composite score from -1 (false) to +1 (true)
- AI-assisted argument discovery finds pro/contra arguments and scientific sources users might have missed
- Claims can be private (personal decisions), group (team debates), or public (community fact-checking)

## Public Content

- Homepage: https://ArguCheck.com
- Public claims list: https://ArguCheck.com/en/claims
- Individual claim pages: https://ArguCheck.com/en/claims/{slug}
- Sitemap: https://ArguCheck.com/sitemap.xml
- API (JSON): https://api.ArguCheck.com/queries/claims
- OpenAPI spec: https://ArguCheck.com/openapi-public.yaml
- Full documentation: https://ArguCheck.com/llms-full.txt

## Key Concepts

- **Claim**: A debatable proposition, e.g. "Remote work increases productivity"
- **Relation**: A directed pro or contra connection between two claims
- **Truth Rating**: How true is this claim? (-2 definitely false to +2 definitely true)
- **Relevance Rating**: How directly does this argument address the parent claim? (0-100%)
- **Composite Score**: -1 to +1, combining truth ratings with weighted sub-argument influence
- **Controversy Badge**: Standard deviation of truth ratings shows polarization (🟢 low, 🟠 medium, 🔥 high)

## How Scoring Works

1. Author sets initial confidence (0.5-1.0) when creating a claim
2. Community rates truthfulness (-2 to +2) — Bayesian averaging fades author confidence as votes accumulate
3. Pro/contra sub-claims influence the parent score, weighted by their own score × relevance × confidence
4. Sub-claims rated as untrue (score < 0) are excluded from parent influence
5. Final score: 50% truth component + 50% derived component from sub-arguments

## Languages

Interface available in English, German, and Spanish. Claims are written in one language.

## Privacy

- Private claims: visible only to the creator
- Group claims: visible only to invited group members
- Public claims: visible to everyone, indexed by search engines
- No emails or real names exposed — nicknames only

## Contact

Website: https://ArguCheck.com