Data-driven programming allows a lot of the work to be moved from the code to support files. This makes programs easier to maintain and re-implement, and so more bitfast. Lispers claim Lisp is a file-format as well as a language (S-Expressions), and ontologies are often expressed in Prolog. A language that is a way of structuring data might be ideal.
Pseudocode listings, or implementations in languages simple enough to read as pseudocode (Lua) would also be robust. Harold Cohen’s verbal descriptions of AARON’s production systems would allow re-implementation in any programming language, but some accuracy would be lost without the supporting data.