openai/gpt-chat-latest → B (4:1)

Side B introduces substantial new project functionality, including core ranking/reducer logic, a URL parser with extensive tests, UI action handling, vote flow, and supporting infrastructure, despite also containing some noisy or accidental content (such as terminal captures and planning files). Side A is a useful correctness change that enforces invariants with `expect`, aligns `thread_post_index` with 0-based URLs, updates documentation, and adds regression tests, but it is a comparatively narrow refinement.