An user's point of view on the privacy issue

[Vittorio Bertola <vb@bertola.eu.org>]

Hello,

I am a newbie of this group and of the IETF WGs in general (please
pardon me for anything inappropriate I might unvoluntarily do).
However, I have been discussing DNS privacy issues extensively in the
last years, so please allow me to give my point of view on the ongoing
privacy discussion.

Not addressing the privacy issue in the base protocol would likely
imply that the service would often be deployed in real life without
any means to achieve privacy protection. Unfortunately, the present
lack of privacy protection in the WHOIS system is plainly illegal in
many countries, and I don't think it's reasonable to think that this
situation can go on for long without actual lawsuits starting to
happen, both towards ccTLD and gTLD registries and registrars. 

In fact, as others have already pointed out, many registries
(especially European ccTLDs) have already started to allow opting out
from WHOIS under certain conditions or for certain types of data, or
even, have already been sued on this. Personally, I think that the
present situation where gTLD registrants are required to make all
their data public won't last long.

Thus, any new protocol being created in this field should be able to
support the ability to mark data as private - otherwise in the end it
might be useless or even damaging. If this protocol doesn't implement
any simple and standard way to specify reasonable privacy directives
together with data, it is likely that many registrars and registries
will be soon forced, by law, lawsuits, or public opinion pressure, to
add their own (non-standard and non-interoperable) ones.

The protocol must allow customers to specify privacy conditions with
the highest possible granularity, because it must be able to support
policies that will be very different one from the other and will vary
often (much more often than the protocol itself) according to
non-technical decisions. No privacy policy should be hard-wired in the
protocol (and this includes the policy of "no privacy is possible"
that would result from the lack of privacy specification tools in the
base protocol).

I must also point out that, according for example to the European law,
it is the customer, nor the registrar nor the registry nor any policy
or standard making body, that decides what should be published and
what should not. The registrar or registry are not allowed to alter
the customer's indications on privacy. At most, the registrar/registry
may refuse to supply the service if the customer does not accept to
distribute data that are strictly necessary for the service to work.
(It seems to me very doubtful that publishing my name and e-mail to
the whole world is strictly necessary for my name servers to work. But
this is a policy and legal discussion anyway, and is out of this
list's scope.)

So, the minimum level of granularity that the protocol should support
to be applicable in real life is the ability to mark each field of
each domain name registration form as private or public, singularly
for each (domain, field) couple.

The EU law also states that the owner of the data has the right to
verify and update the data and retire the consensus to the
distribution at any time. So the protocol should allow for updates not
only of the data but of the privacy indications too.

Theoretically, a registrar could ask separate approvals to the
customer for different uses of the same data. In this case, a
mechanism with more levels of privacy would be necessary. However,
this is an option for the registrar, not a requirement, so this could
be left to extensions. Similarly, a specific approval is required to
export data outside of the European Union, so a mechanism to specify a
list of countries to which data can(not) be exported could be of use,
but this problem can be easily avoided by the registrar by asking for
such consensus, so this could be left as a possible extension too.

Thus, summarizing, I support the idea that a mechanism to specify (at
least) whether each single field of each single domain name is meant
to be public or private should be added to the base protocol, and its
implementation should be mandatory.
-- 
vb.                  [Vittorio Bertola - vb [at] bertola.eu.org]<---
-------------------> http://bertola.eu.org/ <-----------------------