The Regularizing-Reducing Model

Link post

The linked document is my (last-minute) submission for the AI alignment prize. Feedback is appreciated, even though it cannot change my submission entry.