Severity answers one question: how badly is this hurting the person in front of it? That is not the same question as priority, which also weighs effort, revenue, and whatever else is already in the sprint. Conflating the two produces a queue where nobody can tell whether a 1 means harmless or merely unimportant.
Five points is the usual compromise. Three collapse into "bad" and "not bad" and stop discriminating almost immediately. Ten invite an argument about the difference between a 6 and a 7 that nobody wins. Five gives you a clear floor, a clear ceiling, and a middle that means "this is slowing me down".
The scale only works if the person choosing it can read it. Labels written from the reporter's side — "I can't finish what I came to do" rather than "P1, major functional defect" — get a usable number out of someone who has never seen your taxonomy and never will.
In BugLoop the scale runs 1 Cosmetic, 2 Minor, 3 Moderate, 4 Major, 5 Blocking. It is optional by default, because a required field is a reason to abandon a form, and when it is set it prefixes the issue title as "[Bug · S4] …".
Why it matters
An inbox sorted by arrival time treats a typo and a broken checkout as the same event. A severity the reporter chose in one click is the cheapest signal you will ever get for telling them apart.