Onaquest...
In the the mid 1980s I was able (privilaged) to attend several MIL-HDBK-5 comittee meetings as the sole USAF-ALC rep non-voting observer [NOT same as voting AFMC/ASD/AFWAL reps]. Coincidentally most of the voting members were also MIL-STD-1312 and MIL-STD-1515 (fastener) comittee members who discussed related work on-the-side.
What impressed me the most was the massive use of statistics to derive "the simple numbers on the tables". Also, all voting members were acutely aware that as test-sample populations increased, so did the overall confidence in the derived data. Problems with data accuracy were often evident as material fabricators (ALCOA, ALCAN, KAISER, etc) submitted "spec" qualification test data... and the data variations affected the statistically derived "simple numbers on the tables".
In "extreme materials", such as 7178, the allowables varied remarkably over time (usually downward).
In more stable/popular materials properties such as 2024, 7075, etc... A and the B basis allowables have "generally" migrated upward from "S" [spec] basis allowables... all because manufacturers learned the nuances of alloying, heat treating, forming, straining (sheet, plate, extrusion, forgings) etc... the material for optimum/stable mechanical/physical properties (and usually improved SCC & corrosion resistance... etc).
Regards, Wil Taylor