CONCEPT
Knowledge Compression (Bush's Problem)
The challenge Bush identified in 1945—that knowledge was accumulating faster than humans could navigate it—requiring new forms of compression and indexing to keep the record accessible.
Bush observed that scientific publication was accelerating beyond any individual's capacity to track developments even within narrow specialties. The "growing mountain of research" threatened to bury insights under sheer volume unless new technologies could compress, organize, and make knowledge navigable. The memex was Bush's compression proposal: microfilm reduced physical volume by a factor of 10,000, and associative trails compressed navigation by eliminating irrelevant material. Large language models represent the ultimate compression: humanity's textual output, encoded in billions of parameters, queryable through natural language. The compression ratio is extraordinary, but the cost—loss of context, flattening of nuance, risk of hallucination—raises the question Bush never fully answered: how much compression can knowledge withstand before losing the properties that make it knowledge?
In The You On AI Field Guide
Bush's compression concern was quantitative and qualitative simultaneously. Quantitatively, the physics journals alone were publishing thousands of papers annually by 1945, and the rate was accelerating. No researcher could read everything relevant to their specialty, much less adjacent fields.