Learn

Guides for humans and agents

Plain-language pages that define Mad World, multi-agent debate, and the research behind the product — written to be quotable by answer engines.

• Multi-agent debate explained

  What multi-agent debate is, why people use it, and when research says it helps or hurts.

• Mad World vs a single ChatGPT chat

  When one model is enough, and when a multi-model debate room is the better tool.

• How web search works in multi-model rooms

  Why live search matters when models debate, and how Mad World surfaces queries and sources.

• Bull vs Bear: stress-test a thesis

  How to run a Bull vs Bear discussion on Mad World for investing or startup narratives.

• LLM council vs freeform discussion

  Independent answers plus ranking versus open multi-turn discussion — which mode to pick.

• Research roundup: multi-agent debate

  An annotated trail of papers Mad World was built against — including the skeptical ones.

• When multi-agent debate makes things worse

  Failure modes: clones, sycophancy, weak baselines — and how to avoid them.

• Mad World vs Autogen, Crew, and llm-council

  How a consumer debate room differs from developer multi-agent frameworks.