Some have questioned why I put the Compensate part upward on the right side, instead of downward, as is usually considered.
Well, this may be obvious to those in the know, but: Compensating control weakness at a higher level simply does not work..!
This, because of some very basic principles:
1. Any ‘control’ at some level, will have to be implemented at at least one lower level or not exist at all except for some ink on paper (OK, ‘ink’ on ‘paper’).
2. ‘Implementation’ of some deficient control at any level by ‘compensating’ it at a higher level, will lead to an implemenation at the level of the deficiency or lower, or will not be implemented.
3. The lower the implementation level, the stronger. The higher, the weaker. ‘Compensating’ at a higher level requires more controls there to be about as strong, and hence more at the same/lower levels as implementations otherwise the same strength may not be achieved.
4. ‘Compensating’ at a higher level doesn’t fit in the design at that level or would be there already, the deficiency would not be ‘compensated’ by pointing at its rationale. Adding to the design, obviates that the design was deficient or is overcomplete now – the resulting implementation will be flawed by design.
5. Occam-like efficiency thus requires implementation of compensating controls at the same or lower levels.
[Paris, La Défense, for pictoral reasons]