The world has many "wicked problems," which are difficult, confusing, and where it's not clear what a solution would even look like.

Many of those problems are high stakes, and time sensitive.

We need to develop a science and art of solving wicked problems. The bottleneck is that feedback loops are slow and noisy. Researchers and entrepreneurs tackling ambiguous problems work on long timescales, making up processes as they go, getting lucky in subtle ways. It's hard to tell which skills were most helpful for dealing with a given situation.

The Feedbackloop Rationality project aims to discover, distill and train skills that help with generalized "confusing-problem solving". Along the way, we develop feedback-loops that bridge the gap between what we can easily measure, and what ultimately matters, when facing confusing problems.

We teach our current state-of-the-art, while being clear about the state of the evidence.

We currently focus on "metastrategy" - the art of efficiently generating and evaluating plans, choosing directions that seem most promising, and noting what evidence would change your mind.