The thing which concerns me with this discussion is the inference that such matters ought to be dealt with in a black and white manner and that such an approach is philosophically sound.
In the (typically training) environment, this is a reasonable construct and is useful for the training function.
However, in the real world, life is all about statistics and backing the horse most likely to run first. Doesn't mean that one always is on the money but, overall, one is likely to end up in front.
SOPs are based on this philosophy with the goal of having a win most (in the case of aviation, this means to a very high probability) of the time.
We can easily come up with f'instances of where a non-SOP approach could have (would have ?) saved the day. The Concorde crash is a most suitable such example.
However, SOP cannot be predicated on a very low probability exception to a generally observed "rule".
There are risks associated with aviation, just like with life in general, and, on occasion, a few of us are going to be caught out by a risk coming in on long odds. It is tragic when the result is catastrophic but that is the reality of the certification basis for aircraft ... it doesn't revolve around guarantees, only low risk probabilities.
Therefore, SOPs are based on the general rule(s) rather than the exception. On the infrequent occasions where something comes in out of left field, the outcome may be problematic. Sometimes the result is catastrophic (eg Concorde, O'Hare) .. sometimes the crew can pull off a miracle (eg Sioux City).