The whole point of designing a 'thing' is that it should behave predicatably.
It seems to me the point of designing a thing is that it should behave predictably when one wants it to behave predictably.
If there are times when keeping things predictable is not necessary, then a good design might save time and money by making things simpler, as long as no harm is done when the circuit behavior is unpredictable.
That appeared to be the nature of the question posed to us here.
Like Dippy, and most other good technical people I know, I personally like my electronic devices to operate reliably under all circumstances. (In fact, I'm well known and laughed at for heavily over-building just about all of my projects. Hi-Fi amp power supplies with 200,000 uF caps? Just about right. What can I say?)
But sometimes that's just not necessary or desired, and depending on the nature of the project, it may be entirely detrimental.
Good engineering practice boils down to understanding the desired result, and getting it simply, reliably and inexpensively, while ensuring that safety and other peripheral requirements have been attended to as well.
As with many of the other questions posed in the forum, our lack of knowledge of the specific intent of the questioner, the amount and type of their expertise and the exact details of the project, lend to our initially providing a variety of different answers, all appropriate and correct from different points of view and with different basic assumptions, none of which may match those of the questioner.
So we stumble along as we work our way through these threads, learning more and more about the exact nature of each project and its individual requirements, providing better (more appropriate for the user's goals) advice with each post.
I think we do a pretty good job of it by the time all is said and done.