[DMM] Stephen Farrell's Discuss on draft-ietf-dmm-4283mnids-04: (with DISCUSS and COMMENT)

["Stephen Farrell" <[e-mail]>]

Stephen Farrell has entered the following ballot position for
draft-ietf-dmm-4283mnids-04: Discuss

When responding, please keep the subject line intact and reply to all
email addresses included in the To and CC lines. (Feel free to cut this
introductory paragraph, however.)


Please refer to https://www.ietf.org/iesg/statement/discuss-criteria.html
for more information about IESG DISCUSS and COMMENT positions.


The document, along with other ballot positions, can be found here:
https://datatracker.ietf.org/doc/draft-ietf-dmm-4283mnids/



----------------------------------------------------------------------
DISCUSS:
----------------------------------------------------------------------


I don't consider that merely mentioning that there are some
privacy issues (maybe) is nearly sufficient here.  Instead I
would argue that any of these identifier types that could have
privacy implications need to be specifically justified or else
dropped. By specifically justified, I mean that there ought be
an argument (and a fairly holistic one) that the Internet is
better, and not worse, if we define a codepoint that allows
MIPv6 (and later, other protocols) to use that identifier.  I
do accept that my position is perhaps innovative, in terms of
IETF processes, so I'll split the discuss into two parts, one
process oriented and mostly for the IESG, and the second
relating to the content of the draft.

(1) For the IESG: is it ok that we introduce (codepoints for)
a slew of new long-term stable privacy-sensitive identifiers
just because they might someday be needed, or do we need to
have specific justification for defining such things? I would
argue the latter, but that may need us to validate that there
is IETF consensus for that somehow, and perhaps in the
meantime hold on to this draft. Part of my reasoning is that
once we define such codepoints (e.g. for IMSIs) then that
inevitably means that other protocols, and not just MIPv6,
will do the same eventually, so accepting this draft basically
means accepting that we end up commonly and perhaps
carelessly, passing such highly-sensitive information about on
the Internet in many protocols and in many contexts.  My
argument here I think does adhere to various of our BCPs that
do argue for security and privacy, but I do also accept that
this may be novel and to some extent goes against another of
our generally accepted ideas which is that we benefit from
folks documenting things even if those things are sub-optimal
in various ways. So I'd argue this is a real case for an IESG
discussion - I know what I think, but what do the rest of you
think?

(2) For the authors: to the extent you are willing to, and
want to get ahead of the discussion on point (1) above, can
you in fact provide an argument, for each of the identifiers
here that have privacy-sensitivity, that the Internet is better
overall if we define these codepoints knowing that if we do
define a way to represent e.g. an IMSI in MIPv6 that is likely
to be copied elsewhere? Note for the authors: I think it's
entirely fine for you to do nothing pending the discussion of
point (1) above, if that's your preference.


----------------------------------------------------------------------
COMMENT:
----------------------------------------------------------------------


While I'm entirely sympathetic to Mirja's discuss point, I
don't think that a statement here can really constrain how
these identifiers, once they are defined, are used in other
protocols. While there is a chance that some IESG sometime
might say "hold on, RFCxxxx (this doc) says you SHOULD encrypt
if <that> identifier is used" the chances that a future IESG
notice this isn't that high, but it's also even more likely
that the designers of future protocols will successfully argue
that since not all identifiers are privacy sensitive, their
specific protocol need not adhere to the SHOULD. In the end, I
think that should or SHOULD will be ineffective.