Flagship paper · Original research
Transformational Intelligence
Learning to Change the Game
Invent a better problem-solving setup
Identify removable restrictions in representations, information channels, coordination and intermediate objectives. Construct a different formulation and test whether it better serves the authorized purpose.
When to reach for this paper. Start here when better search inside the inherited setup cannot remove the binding restriction.
Scope and evidence
The paper gives exact illustrative, model-conditional results and an evaluation program. A standard unrestricted planner matches its intervention policy; it does not establish a new causal-discovery algorithm or an LLM breakthrough.
This original manuscript is preserved unchanged. Hosting and practical guidance do not independently reproduce its studies or establish a current operational or commercial result.
Read the complete paper
Search, zoom and change pages with your browser’s PDF controls. On a phone or if the preview is blank, open the full paper or download the PDF.
Original edition and integrity
October 2026 · 24 pages · 305,040 bytes. The dated PDF URL preserves this edition; this reading page is the stable entry point.
SHA-256: d7170e8217e3bf6aa1b84868c1c76c5994c6995c5297ed092d7e72a6430c664f