One principle, expressed as two specific practices
Separation of privilege holds that access to something consequential should depend on satisfying more than one independent condition, rather than any single one being sufficient on its own. Two more specific practices covered elsewhere are both this same principle, applied in different ways: split knowledge divides a secret so no one person holds the whole thing, and dual control requires two separate parties to both act before a consequential operation proceeds. The principle is the umbrella; the practices are specific instantiations of it, not separate ideas that happen to resemble each other.
The underlying logic is identical across every instance: a single point of compromise — one person, one credential, one satisfied condition — shouldn't be sufficient by itself to grant access to something that genuinely matters. Naming this as one general principle, rather than a loose collection of specific unrelated practices, makes it available as a design tool beyond the two examples already covered.
Using the principle to design something new
Recognizing separation of privilege as a general principle makes it easier to ask the right question when protecting a new, consequential capability that doesn't map neatly onto split knowledge or dual control specifically: what are the independent conditions that could be required here, and is any single one of them, on its own, actually sufficient today. If the honest answer is yes — one thing alone is enough — that's the specific gap the principle is naming, whatever the details of the actual mechanism eventually turn out to look like.