argue

Multi-persona structured debate orchestrator. Two or more personas research independently via /dogpile, consult domain colleagues via /ask, and argue toward a nuanced synthesis — not a winner.

Why Not Just Ask One Model?

Single-model answers are confident but flat. Real understanding of hard problems (Collatz Conjecture, P vs NP, consciousness, policy trade-offs) requires:

• Multiple domain lenses on the same question
• Adversarial pressure — each side stress-tests the other's reasoning
• Research depth — each persona does independent /dogpile research
• Social graph — personas can /ask colleagues for expert opinions
• Concession tracking — the synthesis captures what was conceded and why
• BDI-weighted reasoning — persona bridge weights influence argument style

Architecture