Fable
The archive that boils old cases into rules
The South City archive has collected countless case files. Each file records who came, what they asked, which document they took, and whether it was ultimately completed. When the curator was young, he believed that as long as the case files were kept complete, later generations would naturally learn how to get things done.
As files piled up, the problem changed too. Every time a new copyist met a client, they had to leaf through dozens of old case files to guess what this type of client usually cared about. Clients grew impatient, and copyists often mistook incidental situations for general rules.
The curator first asked everyone to write summaries on the covers of the case files. The summaries were indeed shorter, but they were still stories case by case: the Zhang family was urgent last time, the Li family was wrong that time, the Wang family once changed its story. The system knew many fragments, but could not state stable facts.
A newcomer once received an out-of-town chamber of commerce. He dug up an old case file and found that the chamber had once required a paper stamp, so he insisted the client wait three days. Only later did he learn that was just a temporary requirement during last year's audit, and now the client wanted an electronic receipt.
The curator realized that past events cannot be directly treated as rules. Experiences need to be abstracted, validated, and updated before they can become reusable common sense for next time.
He had senior transcribers review case files weekly, distilling recurring facts, preferences, terminology, and constraints into a knowledge book, annotated with source, scope of applicability, and last verification time.
The knowledge ledger did not record every quarrel, only stable conclusions: which customers require two-person confirmation, which industries often ask about data retention, which forms must first obtain authorization. If a conclusion came from only a few events, it was marked low confidence.
Later, when the newcomer received the chamber of commerce again, he did not have to dig through the files from scratch. He first read the knowledge book, then reviewed the supporting case files when necessary. His answers were faster, and he less often mistook an incidental experience for a rule. The curator said: the maturity of memory is not memorizing everything, but distilling repeatedly verified experiences into knowledge that can guide the next action. Later, the knowledge book would also expire. If customer habits changed, old conclusions had to be downgraded; only if they remained stable after repeated verification were they promoted to common rules. Memory thus became both fast and not rigid.