Re: New proposal from Ronald J. Fitzherbert

[Kent Crispin <kent@songbird.com>]

Bryant Durrell allegedly said:
> 
> 
> As a side note, I think you're absolutely right on the main point.  
> Email is a lousy way to determine first come first served.
> 

OK, OK, in the face of blistering resistance I withdraw my email
proposal.  How about this:

A perl program, listening on a privileged port, that accepts up to
2048 bytes of data on a single connection, writes the data to a file,
appends a timestamp, signs the file and timestamp with pgp, and sends
the signed file back to the client.  This gives an unforgable
timestamp from a neutral third party.  The client registries would
send a file with the domain name and customer identifying information,
signed with the registries private key.  The registry could use this
certificate to prove priority.  The timestamp server would be 
stateless. 

Bet I can do that perl script quicker than Perry and Michael can get
their database stuff going.

Just joking guys!  Remember:  "Kent doesn't hate databases".

The point is that there is an embarassment of solutions to this
problem -- enough so that I don't think it is necessary to specify one 
that applies to all TLDs.  Any reasonable solution that is specified 
in the TLDs charter, or in an RFC, would be fine.  In fact, I don't 
care if an inefficient human-intensive approach is used -- perhaps 
that would make the TLDs that used it less efficient, and thus 
not as able to provide competitive prices.  But that's a problem for 
the registries, and it doesn't impact DNS or the correct functioning 
of the net at all.  Thus, it really doesn't need to be specified by 
the IAHC/IETF/ISOC.

To repeat, contrary to Bob Allisats ex cathedra pronouncements, there
is no big technical obstacle to implementing shared registries. 

-- 
Kent Crispin				"No reason to get excited",
kent@songbird.com,kc@llnl.gov		the thief he kindly spoke...
PGP fingerprint:   5A 16 DA 04 31 33 40 1E  87 DA 29 02 97 A3 46 2F