In traditional LR parsing, the lexer must decide what token to return before the parser sees it. But for context-sensitive syntax, the correct tokenization depends on context that only the parser knows. The lexer can't see it—so Ruby's lexer resorts to manual lex state management, guessing context through hand-written code. This scatters language rules across implementation details, makes bugs hard to reproduce, and complicates every syntax change. PSLR(1): Pseudo-Scannerless Minimal LR(1), which I've implemented in Lrama, inverts this relationship: the parser tells the lexer which tokens are valid at each point. In this talk, I'll show how PSLR(1) bridges the information gap between parser and lexer, and explore how it could liberate Ruby's parser from the Lexer Hack curse.