Synthetic Philosophy
Problem Solver

The Maximally Coherent Theory (MCT)

Derived from Louis Vervoort's Problem Solving in Philosophy, the MCT-hypothesis states that answers to philosophical (and complex) problems can only be given with reasonable certainty if they are coherent with, or derived within, the maximally coherent theory. It is the theory that coherently answers the maximum number of well-posed questions in the problem-space.

Hardening Against Hallucinations via Agent-Based Reasoning

Large Language Models (LLMs) often hallucinate when asked to solve complex problems in a single pass. To harden the problem-solving cycle against these errors, we deploy Agent-Based Multi-Round Reasoning.

• Cartographer Agent: Maps the problem space without proposing solutions, acting as a structural constraint.
• Theorist Agent: Proposes competing hypotheses (T1, T2...) based on the cartography.
• Critic Agent: Tests each theory for coherence against known facts and the problem map.
• Synthesizer Agent: Selects the MCT and formulates the final answer.

This mimics human deliberative reasoning. By breaking the task into specialized, adversarial roles, the system self-corrects. Evidence suggests that multi-agent debate improves factual accuracy and reasoning and aligns with the hypothetico-deductive method necessary for scientific discovery.