On Sat, 2004-10-02 at 14:22, Dave Crocker wrote:
My concerns regarding a level rating system for a domain,
like one would used to rate a movie with 3 out of 4 stars,
implies comparison. This is bad with respect to the
information required to justify such a rating.
Currently the DNA proposal offers:
'A' for Strongly Recommended
'B' for Recommended
'C' for Unknown
'D' for Not Recommended
'E' for Strongly Not Recommended
This should be changed to something more along the lines:
'A' Good
'B' Good with pending complaints
'C' New (unknown)
'D' New with pending complaints
'E' Bad
boy oh boy, do i ever disagree!
"with pending complaints" gets into the details of internal
rating computations. that is very definitely NOT something that
this particular mechanism should do.
Any large domain may have interludes of abusive accounts. These
situations are often short lived and exist within the period of the
problem detected, and the problem accounts being disabled. This
mechanism seems a rather straight forward means to close the loop and
protect the recipient.
this mechanism should be cast strictly in terms of recommended
recipient action, rather than anything that requires deeper
analysis by the recipient.
The action for pending complaints may allow the use of temporary refusal
as a coping mechanism. The provider would also experience this change
in state within their SMTP logs. The provider then obtains the feedback
regarding these problematic accounts, clears the problem, and the domain
reverts back to a good state. This offers a clearer scenario with
respect to recipient actions than would an A or B rating.
consumer transaction approval systems currently only have a
yes/no choice. by going to a 5-value range, rather than 2, we
are already being adventuresome. But let us at least keep the
semantics of the 5-value range utterly straightforward and
consistent.
What is straight forward about what an A or B rating involves? What is
straight forward about the actions a recipient should do as a result of
one or the other? With respect to justifying such assertion, there is a
huge difference with respect to discovery you fail to consider.
It would be much easier to define the breakdown of Good/New/Bad with
pending complaints added as a warning of an unresolved issue. This
pending complaint state allows for temporary refusal as a means to
ensure problem accounts have been resolved.
Very-Good/Good/New/Bad/Very-Bad requires at least an understanding of
the percentages of accounts involved in these distinctions over a
specific time period before a recipient can make an intelligent choice
of action. Each reputation service would apply these terms
differently. What I am proposing is to eliminate these relative
ratings. Such ratings are meaningless without at least a statistical
definition and defined sampling period.
Currently these services offer a binary Good/Bad. The structure that I
have proposed offers Good/New/Bad with a warning added of pending
problems as a means to reflect the dynamic nature of these ratings.
Problems happen, but with such a mechanism, the speed in which these
problems are addressed will be increased and offers recipients a
sensible means of avoiding spam from normally good domains.
This avoids the very difficult task of justifying the rating.
saying that complaints are pending gets slips down the
justification slope.
The discovery process remains limited to the single domain as the
information required for the assertion remains limited to the single
domain. To say very-good or good implies a comparison of all domains.
This is where justification gets hard. There is no slipping with
respect to isolating information to a single domain, but this does
provide an extremely effective means to squelch transient problems.
There's no WG chair to invoke BCP 83: we can go ahead and
talk about most anything -- until we're chartered...
For reference, I am hoping that we conduct the mailing list
exactly as if it were a chartered activity. First of all, that
will train us all to have good habits and second of all it will
be more productive.
The term "basic accreditation rating" should read
"basic accreditation information" as rating and
accreditation seem at odds.
I agree that "basic accreditation rating" seems a poor
term.
I have been using 'rating' as a distinction from 'report' where
the former makes a direct recommendation and the latter simply
provides lots of source data. There is no religion to the term
'rating'.
How about 'basic accreditation recommendation'?
This term recommendation is a bit better. : )
-Doug