Sybil

This explores the problem of finding multiple distinct nodes in a
distributed environment. Malicious entities can attempt to forge
their identity or take on multiple identities for the purposes of
malice and/or espionage.

I must admit that hard as I tried, the proofs escape my ability to
comprehend them. Thus I simply accept the proofs. The results show
that for direct validation (i.e. no chain-of-trust validation), an
entity must simultaneously validate all identities it is presented.
For indirect validation, all entities in the system must perform
validations concurrently. Thus in a distributed system, the resource
requirements required to strongly maintain identity verification is
not practical.

However, these strong claims and conditions are based on the
assumption that there are no external communication links between
entities. For circle-of-trust systems like PGP, external
communication mechanisms are needed and part of the norm. However,
this result means that P2P systems will have difficulty verifying and
thwarting potential cheaters.

In terms of load balancing, there doesn't seem to be much to be gained
by wanting to forge multiple identities for taking on extra load.
However, malicious entities may forge multiple identities in order to
disrupt or break the P2P system. By accepting a large enough load,
the malicious entity can exit the system and cause loss-of-data damage
that is not recoverable. Alternatively, a malicious node may forge
multiple identities in order to gain more than their fair share of
resources.
Received on Thu Nov 17 2005 - 10:14:07 EST