John,

Thanks for reading it and for the questions.

On (2) first, because I think my wording misled you. "Parent" and "child"
in the draft are not DNS hierarchy. They are just the two roles: the
established domain that confirms, and the domain being confirmed. They can
be any two names.
https://www.google.com/url?q=http://example.com&source=gmail&ust=1793079784897000&sa=E
and example.foo, or
https://www.google.com/url?q=http://example.com&source=gmail&ust=1793079784897000&sa=E
and the IDN look-alike, or two completely different names under different
TLDs. No registry or registrar is involved, so it works across registry
operators. The only thing needed is that both operators can publish records
in their own zones.

So yes, those cases are exactly the ones I'm trying to cover. Using
"parent" and "child" in a DNS document was a bad choice on my side. In -01
I'll rename them to "confirming domain" and "affiliated domain", and say at
the start that the two can be anywhere in the DNS, with no delegation or
hierarchy between them.

On (1), you are right that DAC does not tell you a look-alike is an
impersonation, and I don't think any DNS mechanism can, that is a judgement
about the name and the content. What it does do for that case:

- if a look-alike claims to be affiliated with the real organization, the
claim fails, because only the real organization can confirm it. The
impersonator can register a look-alike of the confirming domain too and
confirm itself from there, but then all it has shown is a link to another
unknown domain. A confirmed domain gets no more standing than the domain
confirming it (Section 8.3), so a chain of look-alikes is worth nothing.
- the real organization can get a report when someone claims to be
affiliated with it and it hasn't confirmed (the f= tag, reason
"unconfirmed-token"), so it learns that someone is claiming to be it.

I'll make both points explicit in -01 so the applicability is clearer.

Dimitri

On Mon, Oct 26, 2026 05:25 AM, John C Klensin <[email protected]> wrote:

> Dimitri,
>
> Since I haven't seen other responses to your posted I-D or note to
> the DNSOP list, some questions:
>
> (1) The description in your abstract and your Introduction introduce
> two problems, the second of which is "Legitimate look-alikes are
> reported as impersonations".   It seems to me that the opposite
> problem is also important, i.e., impersonation look-alikes are
> mistaken for legitimate names.  Am I correct in understanding that
> you are ignoring that case, even in your Sections 1.4 and 1.5?
>
> (2) The most important, or at least very widely reported, cases of
> look-alike names involve names at the same level in the DNS
> hierarchy, not parent-child relationships.  For example, one might
> have questions about whether example.com and exámple.com
> <http://xn--exmple-qta.com> belonged to
> the same organization.  That is not a parent-child relationship.  One
> might be able to answer the question with a variation on DAC and/or
> one on [I-D.ietf-regext-epp-same-entity], at least assuming the TLD
> registry is willing to take on such responsibility and, if different
> registrars were used, even that it has access to sufficient
> information.  At least some experience suggests that many registries
> are unable or unwilling to do that.    The even more common case
> would involve a comparison between example.com and example.foo, where
> "foo" is some other TLD, supported by a different registry operator
> than the one for .COM or the one between example.com and example.
> Am I correct in understanding that you do not address those cases
> either?
>
> If the answer is that those cases are no addressed, my personal
> (possibly not very useful, depending on what others in DNSOP have to
> say) is that this proposal is probably not worth the effort except
> for some very specialized cases... or that the draft needs a rework
> to make the useful applicability much more clear.
>
> good luck with it.
>    john
>
>
> --On Wednesday, October 21, 2026 15:49 -0700 [email protected]
> wrote:
>
> > Internet-Draft draft-manelidis-dnsop-domain-affiliation-00.txt is
> > now available.
> >
> >    Title:   Domain Affiliation Confirmation (DAC)
> >    Author:  Dimitri Manelidis
> >    Name:    draft-manelidis-dnsop-domain-affiliation-00.txt
> >    Pages:   46
> >    Dates:   2026-10-21
> >
> > Abstract:
> >
> >    Security systems assess domain names one at a time.  A new
> > domain of    an established organization therefore starts with no
> > reputation, and    a legitimate domain that resembles another
> > organization's name can be    reported as an impersonation.  This
> > document defines Domain    Affiliation Confirmation (DAC), by which
> > the operator of a parent    domain confirms that a child domain is
> > operated by it or under its    authority.  The child publishes a
> > DNS record naming the parent and a    random token; the parent
> > confirms that token, in the DNS or over    HTTPS, by naming the
> > child.  Verification takes at most three DNS    queries.  The
> > parent publishes no list of its domains, and an    affiliation is
> > ended by deleting one record.  The document also    states what a
> > confirmation means to relying parties such as takedown    services
> > and reputation systems.
> >
> > The IETF datatracker status page for this Internet-Draft is:
> > https://datatracker.ietf.org/doc/draft-manelidis-dnsop-domain-affil
> > iation/
> >
> > There is also an HTML version available at:
> > https://www.ietf.org/archive/id/draft-manelidis-dnsop-domain-affili
> > ation-00.html
> >
> > Internet-Drafts are also available by rsync at:
> > rsync.ietf.org::internet-drafts
> >
> >
> > _______________________________________________
> > I-D-Announce mailing list -- [email protected]
> > To unsubscribe send an email to [email protected]
>
>
>
_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to