ietf
[Top] [All Lists]

Re: SMTP Minimum Retry Period - Proposal To Modify Mx

2004-01-12 14:05:40
From: Nathaniel Borenstein <nsb(_at_)guppylake(_dot_)com>

...
I also have to say that I fear your approach would help the larger ISPs 
use spam as an excuse to kill off smaller ISP's...  

How so?  Exactly what is my approach?  Please note what I've said too
many times:
  - I don't currently use a public blacklist and have never used one
      for non-trivial mail.
  - I'm flogging spam defenses that compete with blacklists.

                                                 and I question the 
fundamental legitimacy of blocking all of an ISP's customers before 
there's a fair due process to establish the ISP's culpability. 

"Fair due process" and "free speech" and even "legitimacy" are none
of your concern unless you own the mailbox that would *RECEIVE* the
blocked mail.  No one has any right to send anyone any mail.  We have
only privileges granted by targets of our mail.  If our targets are
foolish and hire ISPs with long histories of both permitting a lot of
outgoing spam and blocking a lot of incoming legitimate mail (see
recent complaints about RR's false positives), then that's just tough
and perhaps we should convince our correspondents to switch ISPs or
find new correspondents.

Good or bad spam filtering is merely a part of the rest of good or bad
SMTP or any other ISP service.  It makes no more sense to condemn the
HTTP protocol because many web pages are junk than it does to condemn
blacklists because some blacklists are junk or used badly.

If you think blacklists are bad because they can be run by fools, then
you also must hate any network authentication and authorization
mechanism.  What's the difference between Kerberos and a mail blacklist?
Both are responsible for summary denial of services.

I fear there are bad reasons for the disdain for blacklists:
 - they are effective against spam from spam friendly ISPs.
 - some of us work for spam friendly ISPs and let the interests of
    our employers color our thinking.
 - some of us are lazy and hire ISPs have been spam friendly.
 - some of us feel we have a devine right to send any mail to anyone
    and are deeply offended by any contrary suggestion, not to mention
    an effective mechanism.


                                                            "Caring 
enough about spam" is an awfully slippery concept on which to base a 
blacklist.

I am offended by your implication that I suggested any such thing. 
I only pointed out that using spam-friendly ISPs has consequences.
(You evidently know about XO's reputation, which I think has improved
lately.)  The only major blacklist that does anything remotely like
your implication is SPEWS, which "escalates" in order to get the
attention of ISPs.  If I did use a blacklist, it wouldn't be SPEWS but
that would be only one reason among serveral.


...
that is not blacklist, then why can't a blacklist be run properly?

Good point.  That's why I favor giving users access to their spam pool 
when they suspect problems, and using challenge/response in certain 
(carefully defined) situations.  A good filtering mechanism is not 
nearly as black and white as a blacklist.

The last part of that is simply wrong.  Every filtering mechanism is
exactly as black and white as a blacklist.  Whether or not an SMTP
server keeps good logs has nothing to do with whether it decides to
reject messages using blacklists of IP addresses or domain names or
anything else.  If your correspondents use software that consults any
blacklist but doesn't keep good logs, then the fault lies first with
your correspondents for using bad software, second with you for having
foolish correspondents, and not at all with the blacklist.

Yes, I realize that I'm implying that to keep good logs you need to
act on a blacklist (if you use one) at the end of the DATA command
instead of before the HELO.


Any fool
can set up a blacklist.  That many fools have and other fools have
used them does not show that blacklists are bad any more than the ease
of setting up an IP network shows TCP is the spawn of the devil.

I will confess that my personal experience makes it very hard for me to 
be rational on the subject of blacklists, as I fear that any concession 
to them will only encourage the creation of destructive blacklists by 
"fools".  In general I prefer a solution that any fool can implement, 
because one surely will.  

Then you'd better give up on the Internet.  As with much of the net,
the information in and functioning of any spam system is at least
somewhat "administrative" and subject to the whims of any fools
administrating it.  The buyer must beware, not only of hiring a spam
friendly ISP, but contracting with a foolish spam filter.  The greater
fool is often the buyer of services offered by lesser fools.


Vernon Schryver    vjs(_at_)rhyolite(_dot_)com



<Prev in Thread] Current Thread [Next in Thread>