Wait... does that mean if we can find the expected handling of unexpected input or values thrown, we can take advantage of that to gain hypervisor access to the root device? Or be able to write values directly into the memory of the system? Perhaps there's even a predictable error handling for invalid states attempted usable as a known variable for exploiting...