One of the better uses I’ve seen involved using this perspective to turn what was effectively a maximum likelihood fit into a full Gaussian model to make the predicted probabilities more meaningful.

Not that it really matters much how the perspective is used, what’s important is that it’s there.