Show pragmatism: "I use function decorators for cross-cutting concerns like caching and auth. Class decorators for registration patterns. Metaclasses almost never — in 10 years I have needed them twice."
Decorators are functions that take a function/class and return a modified version. Function decorators wrap individual functions (logging, caching, auth checks). Class decorators modify or wrap entire classes (adding methods, registering plugins). Metaclasses control class creation itself — rarely needed, used for ORMs or validation frameworks. Strong candidates use functools.wraps to preserve metadata, know the @property decorator, and can explain when a simple decorator beats a metaclass (almost always).
Tests Python depth. Candidates who cannot write a decorator from scratch may lack understanding of closures and first-class functions. Those who reach for metaclasses when a decorator suffices are over-engineering.