The same intent, drifting apart across platforms
An organization running workloads across multiple cloud providers or a mix of cloud and self-managed infrastructure has to express the same underlying policy intent separately for each environment's own native access-control system, which rarely work identically to one another. Even careful, good-faith implementation on each platform can end up enforcing subtly different rules, purely because each platform's own mechanism has different defaults, different granularity, or different edge-case behavior.
This isn't usually a mistake on any one platform's implementation; it's a structural consequence of policy intent having to be translated separately, multiple times, into environments that weren't designed to behave identically to each other.
Portability as the target, consistency as the actual test
Policy portability — expressing policy in a way that's decoupled from any single platform's specific mechanism — aims at solving this at the source, rather than accepting drift and trying to reconcile it after the fact. But portability alone isn't sufficient; the real test is enforcement point consistency: whether the same logical request actually receives the same answer no matter which platform happens to handle it, verified deliberately rather than assumed from the fact that each platform claims to implement the same underlying rules.