A metallurgist's doubts about self-replicating probes