Re: e-postage stamps, was Welcome to the new(ish) ASRG list

On 17 Mar 2013, at 1:51, Franck Martin wrote:

> So once again who are these Many ISPs that outright block on no rDNS?

My employer isn't exactly an ISP, but we provide outsourced email for 
SMB's and reject mail from IP's with no rDNS. No exceptions. Appeals 
from senders or customers are very rare, and none has required careful 
thought in my 4+ years there. When last I looked at the numbers, there 
was no doubt that the garbage being rejected pre-DATA was enough to 
justify the policy and the possible false positives were statistically 
equivalent to zero. Of course there are degrees of "no rDNS" that get 
treated differently, but that should be obvious. Possibly transient DNS 
failures are distinguishable from NXDOMAIN replies, and perfectly match 
the design justification for SMTP 4xx replies.

Beyond ISPs and ISP-like entities, the long tail of corporate mail 
servers is rich with systems rejecting mail offered by IP's that yield 
NXDOMAIN to a PTR query, and they often have no useful route of appeal 
for deliverability problems. Config rejecting mail from nameless IP's 
(i.e. Postfix's  reject_unknown_reverse_client_hostname, Sendmail's 
require_rdns, etc.) has become part of the canon of anti-spam advice for 
good reasons (it blocks a respectable pile of spam, it's cheap, and it 
has very low FPs) and many mail systems run on autopilot after the 
deployment of such advice until they noticeably break. A sender using 
nameless IP's may well get not be blocked or dropped by half of the big 
freemailers and 3/4 of the largest consumer access providers, but below 
that scale run into a brick wall.