Byte Language Models: Scaling, Emergent Abstractions, and Information Allocation | Hacker News Reader