[vworld-tech] Ultimate MMO Platform

[J C Lawrence]

On Mon, 12 Apr 2004 10:52:01 +0100 
Crosbie Fitch <crosbie at cyberspaceengineers.org> wrote:

> Anyway, a private key based identity can be thought of as just a
> unique name (cos that's what it is), and that any time another node
> pops up with the same 'unique name', the node has to re-demonstrate
> its identity/authenticity via a more longwinded process. 

How would you determine that two distinct nodes have the same key etc
given the presence of NAT/PAT, load balancing proxy clusters, dynamic
DHCP address (re-)allocation[1], etc?

> Just as humans have to do when someone introduces an identical clone
> (but which only has a small, albeit recent, subset of shared
> memories).

So nodes exchange secondary shared secrets to establish localised
identity graphs, essentially attempting to build a SAME AS identity
relationship across time (which is a different aspect of identity)?

> Of course, there's another problem and that is: "How do you know that
> two identical nodes aren't actually just a single genuine node that
> is having its communications maliciously duplicated in order to have
> it invalidated?". 

Or behind load sharing proxies, or behind a NAT device that changed
address, or is a single host with multiple interfaces, or or or or.

> In order to detect duplicate identities one does also need to rely on
> addressability, i.e. that it is possible to rely on there being one
> and only one address per node, or rather that if there are multiple
> addresses or communication paths, that it is possible to determine
> whether these are equivalent or not.

I don't believe that it can be assumed.  The correct approach in this
space may well be the 80/20, with resilience/containment methods for the
20% it breaks/fails/is_compromised.

-- 
J C Lawrence
---------(*)                Satan, oscillate my metallic sonatas.
claw at kanga.nu               He lived as a devil, eh?
http://www.kanga.nu/~claw/  Evil is a name of a foeman, as I live.