Dialogue · Experiments

AI generated dialog research

Project documents start here · methodology · system prompts
Start here
Methodology
System prompts
  • Actor the performer's stable posture; thinking-block rules; closing the scene
  • Coordinator the LLM coordinator's dumb-pipe discipline (scheme 2)
  • Monolithic the single-LLM dialogue generator (scheme 3 — the null hypothesis)
  • Auditor the auditor's posture and what to refuse to do
Study 001 — three schemes, same scene

Same characters (Brian Yuan + Shane Lauft), same scene (a Sunday meeting at Red Rock Coffee about a Palo Alto listing and a secondary-share opportunity), same model (claude-sonnet-4-6). Three architectures. See the three schemes for the design and the study README for what's being measured.

Study 002 — Claude vs DeepSeek, same scheme

Same scheme (python), same scene, same characters. Different LLM. Tests whether the framework's discipline transfers across models. See the study README for what's being measured.

After the post-mortem · framework revised
Pilot experiments