What does it mean that a meta-schema is a schema about schemas?

You can build models of how your models work — this is the beginning of recursive self-improvement.

Example: You have a schema for evaluating job offers (weigh compensation, culture, growth). But you've never examined how you built that schema — whether it came from a single bad experience, parental advice, or systematic research. A meta-schema would let you ask: where did this model come from? What assumptions does it encode? When should I rebuild it from scratch? Without that meta-level view, you're running decision software you've never audited.

Try this: Pick one schema you use regularly — how you evaluate people, how you decide what to read, how you prioritize tasks. Write down the schema itself (the rules, heuristics, or criteria it contains). Then answer three meta-questions about it: (1) Where did this schema come from? (2) When was it last updated? (3) What would need to be true for me to discard it entirely? You've just built your first meta-schema.