Traditional dist sys often borrow concepts from the real world & hence "unique identities" seems like a natural thing to do coz we're used to using identities to divvy up work & attribute faults/punishment. Nakamoto consensus is brilliant because it leapfrogs this mental block.https://twitter.com/NickSzabo4/status/1099343107114524674 …
-
AFAIK the idea of identity has such a strong appeal that it’s omnipresent in both regular software designs & network designs. Process ID in Linux, Android application ID. MAC address blacklist/whitelist, IP subnets based on address groups, etc.
General pattern is the limited aspect one relies upon to be unique (e.g. per-process memory and CPU limits) are centrally controlled (e.g. by the local OS). Can't rely on them to be unique in other ways (an app can spawn many processes, a person can have many IP addrs, etc.)
