We are proposing to replace an information-rich PSL resource, with DNS
lookups that currently have no explicit information about organization
boundaries.    The good news is that we can get pretty close to
PSL-equivalent results with the existing DNS and some heuristics.   But
certainty is always preferable to heuristics, which is why we need a flag
for registrar entities (organization boundary below) and a flag for
organization domains (organization boundary above).  DMARC has always been
about empowering domain owners and evaluators to work together to improve
trust.    Using the tree walk instead of the PSL puts those participants
back in control of their relationship.
In the near term, the PSL will have an information advantage, especially
for private registries.   Consequently, existing implementations will
continue to consult it.   But for the long term, registrars and their
registrants are the best source for information about organization
boundaries, and they will become the official source if we give them the
tools to do so.

On the other hand, to build a complete solution around the PSL, we would
have to document all of the information flows between the PSL entity and
the registrar entities, define the integrity assurance processes for that
communication, and eliminate the ambiguity caused by having more than one
list.   It seems clear that we do not want to undertake that effort.  We
also have the nagging fear that at some point the PSL will become
unreliable or unavailable for our purposes.

So we define a new algorithm that eliminates the PSL using the
best-available heuristics, treat the old algorithm as valid but deprecated,
and let the industry pick the long-term winner.

Doug
_______________________________________________
dmarc mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmarc

Reply via email to