6. What is Routing in Backend? How Requests Find Their Way Home The speaker distinguishes between HTTP methods expressing the "what" (intent) and routing expressing the "where" (resource), yet states the server uses the combination of both as a "unique key" to map to a handler. Are there architectural or semantic edge cases where this strict "what" vs. "where" separation becomes ambiguous or breaks down in practice? The use of path parameters for "semantic expression" (e.g., /users/123 ) is contrasted with query parameters for "metadata" or "user-defined values" (e.g., ?page=2 ). Beyond the examples given, what are the deeper implications or trade-offs (e.g., caching, security, URL length limits, developer experience) when deciding whether a piece of data belongs in the path versus a query string, especially in less clear-cut scenarios? Nested routes are presented as a means to "express different semantic meanings" and "nest different types of information." At what level of nesting does this practice become counterproductive, potentially leading to overly verbose URLs, reduced readability, or increased complexity for API consumers, despite the semantic clarity it aims to provide? Route versioning is described as a "stable and complete workflow" for introducing "breaking changes." What are the potential challenges or complexities that could arise if client teams fail to migrate within the provided deprecation window, or if the "breaking changes" are more subtle than just field name alterations (e.g., changes in business logic or error handling)? The colon ID convention for dynamic parameters is noted as an "industry-wide practice." What are the long-term consequences, both for maintainability and interoperability, if an organization chooses to deviate significantly from such widely accepted REST API routing conventions? The primary benefit of a catch-all route is stated as providing a "user-friendly message" for non-existent endpoints. Beyond user experience, what are the security or performance implications of explicitly routing all unhandled requests to a single handler, as opposed to relying on a default server 404 response? The speaker highlights that REST APIs aim to provide a "human-readable construct to routing." How does this goal of human readability balance against the increasing complexity of modern applications, where routes might need to convey highly specific, machine-oriented instructions or support complex filtering and sorting logic?