Reasonable and unreasonable bulk mail

Gary Mills
Fri Mar 12 19:18:44 UTC 2010

I've been reviewing the DCC logs for messages that were rejected as
bulk mail and subsequently whitelisted by users here.  They fall into
some interesting categories.  Messages like customer notifications,
newsletters, and other messages from mailing lists were generally
detected as bulk mail by both Fuz1 and Fuz2 checksums.  One that was
in plain text was detected by all three checksums.  One that used
base64-encoding for both text and html portions was only detected by
the Fuz2 checksum.  These are all messages that users expect to be
classified as bulk mail simply because copies are sent to many people.

At the other extreme are empty messages that contain only a PDF, DOC,
or JPG attachment, as well as one-line messages in html format.  These
are all classified as bulk mail because of the Fuz2 checksum.  Users
send these all the time, and expect them to work because they are
inter-personal messages.  They don't expect them to be rejected, and
can't understand why they need to whitelist them.

In between are brief messages from web sites that contain passwords
or authentication URLs.  These too are often classified as bulk mail
because of the Fuz2 checksum.  Users expect them to work because they
are part of the procedure described by the web site.  They certainly
don't think of them as bulk mail.

I'd like to do something so that our DCC system would not reject
messages in these last two categories.  These are the ones that users
don't expect to be bulk mail.  I realize that I could whitelist
checksums of empty message to alleviate that problem, but unless this
process is automated, I'd have to be doing it regularly.  We also
advise people to add some text to empty messages, but that seems like
a peculiar solution.  Is there anything else I can do other than
disabling the Fuz2 checksum?

-Gary Mills-        -Unix Group-        -Computer and Network Services-

More information about the DCC mailing list

Contact by mail or use the form.