openai/gpt-chat-latest → A (4:1)

Side A introduces substantive validation by enforcing vote ratios of 1–100 in the DSL parser, browser POST handler, and reducer, preventing invalid zero-weight edges and unbounded ratios while adding focused regression and integration tests. Side B primarily performs a sweeping DSL syntax migration (moving vote explanations before comparisons) and updates documentation, fixtures, UI formatting, and tests to match, with comparatively little new functional behavior beyond the syntax change.