One of the greatest movements in the 1990s was patterns (design patterns, architectural patterns, stereotypes, archetypes). Pioneers like the Gang of Four (Erich Gamma, Richard Helm, Ralph Johnson, and John Vlissides), Martin Fowler, and Douglas Schmidt wrote books and wikis to capture patterns.
The truth of a pattern was obvious. The explicit and precise specification in human language enables immediate affirmation, because the reader recognizes it as being not novel. The reader has seen it many times before, so the pattern communicates no new knowledge. This is a key characteristic. It is important to say a true thing that is obvious to everyone when expressed explicitly, especially when it has never been said explicitly before.
The utility is that a pattern formalizes a term of art. It establishes a word or phrase as referring to a specification that details its meaning. Knowing the name of the pattern, the listener can immediately reify a mental model, and know that everyone hearing the same name will share that mental model. That is powerful.
Through patterns, coders effectively changed natural language within the domain of software development to be "English as code". When communicating with English in other contexts, there is a high degree of imprecision. We have to account for ambiguity with more or less generous interpretations. However, with patterns we can be more confident in what we say and what we hear from others who speak in those terms.
I feel like in our journey toward Humanity as Code to enable AI functions that can subsume human toil, we should be specifying patterns as formally as possible across all domains. Perhaps AI models can be doing exactly this on our behalf by distilling our writings into pattern specifications, which can then be verified and cleaned up. This would form a reliable repository of truth, which can then be used for training.