Re: [nc-impwhois] Melbourne IT WHOIS implementation comments

[Thomas Roessler <roessler-mobile@does-not-exist.net>]

On 2003-01-18 09:52:29 -0800, Rick Wesson wrote:

>> Plausibility checkers will always have a certain error margin
>> due to outdated information, data entry errors, programming
>> errors, and the like.  They can't be used as the sole source
>> for judgements on the accuracy of address information.

> please clarify, you want to ensure that a human is always in the
> loop?

Automatically verifying e-mail addresses is pretty much
straight-forward, so you don't need a human in the loop in this case
-- at least as long as the recipient's system produces failure
messages which permit a distinction between a permanent and a
transient error (i.e., no such address vs. disk full).

With telephone numbers, I'm skeptical about your claim (we discussed
that in Shanghai) that it's in fact possible to build a global,
reliable and up-to-date database-backed plausibility checker (which
won't go down to the subscriber level, anyways).  In this case, it
might be a good idea to test-dial numbers, and it would be polite to
have a human on the phone on the caller's side (just in case someone
picks up) -- note that this doesn't mean that you dial manually.

With postal mail, I don't think you can do much reliably without
actually sending a letter -- that's what the 15 day period debate is
all about, after all.


I suppose the place where automated plausibility checkers best fit
into the system is in checking updated information provided by the
registrant in response to an inquiry, so you can catch people moving
from Tinseltown to Gotham City, and demand additional evidence from
them.  (I.e., flag those changes which need human attention.)


The place where they do not fit into the mechanism is in immediately
cancelling domain names based on information which has been accepted
as an update after an accuracy complaint (or which has been
confirmed by the registrant in response to such a complaint, and
accepted by the registrar).  

If you want to put a general system in place which defines accuracy
in terms of addressing information being flagged as plausible by
some automated system, then the checking needs to happen at the time
of data entry, and not ex post.

Regards,
-- 
Thomas Roessler				<roessler@does-not-exist.org>