Observation
How does reality become observable?
Reality exists independently of our models. Every decision begins with imperfect observations. Before collecting data, we must first determine what deserves to be measured.
Representation
How should reality be represented?
Observations alone are not knowledge. They must be encoded into representations — variables, geometry, latent spaces. Every representation emphasizes certain properties while hiding others.
Patterns
Which structures survive noise?
Reality is noisy. Not every fluctuation contains information. Statistical inference and machine learning attempt to reveal persistent regularities. Patterns are the first indication that a system exists.
Mechanisms
What generates these patterns?
Patterns alone are insufficient. The objective is to understand the mechanisms producing them — interactions, constraints, feedback loops, dynamics, hidden processes. Understanding begins when we explain observations instead of merely describing them.
Learning
How are these mechanisms learned?
Every observation updates our understanding. Models are continuously confronted with reality. Representations evolve. Connections strengthen or disappear. Learning is the progressive refinement of hypotheses.
Knowledge
How does information reduce uncertainty?
Knowledge does not emerge from isolated observations. It emerges when evidence reinforces relationships. The objective is not collecting more information — it is constructing an increasingly coherent map of the system.
Decision
How does knowledge improve decisions?
Knowledge becomes valuable when it informs action — forecasts, scenarios, optimization, decision support. The objective is not replacing human judgement. It is reducing uncertainty before acting.
Understanding precedes implementation.