Data and Experience

Data often records observations or facts about particular cases. In the Experience Capitalization lens, experience becomes practically useful when repeated observations contribute to models that change future recognition and action. Normative models can also contain requirements established by applicable governing sources rather than inferred from observations.

Data Can Provide Evidence About Reality

Transactions, measurements, events, logs, and cases provide traces of what happened. Documents may preserve such evidence, but they may also contain prior models, rules, explanations, or normative constraints. Both kinds of material matter because reusable models may need contact with observed reality, with scientific models supported by evidence, with structures deliberately defined through engineering or design, and with requirements established by applicable laws, standards, contracts, specifications, or other governing sources.

But a collection of observations does not automatically specify what matters, which relationships are stable, or which action follows under a given condition.

Experience Compresses Observation

Human experience can be understood as a form of compression. A person sees many instances and gradually learns a pattern. The resulting internal model may represent that a certain condition changes the expected result, that two events appear related, or that a particular state calls for a different action.

The model is the reusable part. For descriptive or predictive structure, individual observations may provide evidence from which it was formed. For normative structure, the relevant requirement may instead be established by an applicable governing source such as a law, mandatory standard, contract, or specification.

Written Sources Combine Data and Prior Models

Civilization has accumulated raw evidence, descriptions of models derived from earlier evidence, and normative or designed structures established through authority and engineering. Research papers, standards, technical documentation, and textbooks can therefore contribute different kinds of model content and evidence.

LLMs can help recover those structures from text and relate them across sources.

Operational Value

Experience Capitalization uses data and other observations as evidence where appropriate, while released reusable models become the durable asset. The distinction matters because storing more observations and producing a supported operational model are different tasks.

The model represents how relevant facts fit together and how they affect action, giving future people and systems explicit structure they can apply.