10:34:39	 From Keith J : you mentioned that always need to look at the valid flag, I can understand why it might be invalid initially but you mentioned going from valid to invalid and back to valid... what are some examples/reasons that data would be flagged invalid after having valid data?
10:35:19	 From Davy Baker : Is there a use case where a different architecture is better ?
10:35:24	 From BobF : With a formal background in Mathematics (+ CS), using a Data-orientated model as opposed to more common application function/object-orientated models, could mathematical ‘Formal Methods’ (Predicate/Proposition semantics) be applied in a similar but entirely different fashion? 

Comment : I’m thinking along the lines of increasing overall robustness of application design. Further, there appears to be a significant shift towards Data-orientated models in AI/ML/DL applications, understandably!
10:39:46	 From Keith J : got it thx
10:43:44	 From BobF : No prob ... thanks!
10:44:15	 From Gonzalo : I could not understand well why the data used must be atomic readable/writable. A model data point has many parameters anyway
10:53:04	 From René Andrés Ayoroa : Thank you, John. I enjoyed your book.
10:53:13	 From Caio J. B. V. Guimaraes : Thank you very much, John! I'll read your book and study more!
10:53:21	 From Lyden Smith : Thank you John!
10:53:33	 From Caio J. B. V. Guimaraes : Remotely is the best way kkk
10:53:46	 From Raul Pando : Thanks John