Hector Santos <winserver(_dot_)support(_at_)winserver(_dot_)com> writes:
Second, we need to focus on the most prevalent fact we have:
By far, by industry measurements, 60-80% of all transactions are
non-SMTP compliant.
Therefore, using any form of mathematics, physics and/or logic, it
should go without saying that a great benefit is theorically achievable
at the SMTP level simply by enforcing SMTP compliancy.
Unfortunately, the 60-80% of SMTP transactions that aren't SMTP-compliant
and the 80% of transactions that are spam are not quite the same 80%, or
this would be a lot easier than it is. Not that I disagree with you, in
general, but that's why the notion of "let's tighten SMTP down to only
what's exactly compliant" has been such a hard sell. Like all other spam
filtering solutions, it has false positive problems, and in fact has more
false positive problems than some of the other spam filtering techniques.
We have a lot of good ideas out there, but there all suffer with the
same realization:
"What to do about the legacy and/or non-compliant transaction?"
Yup. That and the question of whether, given the degree of restrictions
some of these protocols put on how SMTP can be used, whether what's left
after the required slashing and burning of features is really usefully
SMTP any more.
--
Russ Allbery (rra(_at_)stanford(_dot_)edu)
<http://www.eyrie.org/~eagle/>