DUL (was Re: RBL report..)
I've just sent another, long, message about mail acceptance,
blacklisting, and this whole flamewar. Please read that message
first; it explains the context of this mail, and without it you might
misinterpret this one.
This message is about my opinion of the DUL, which I support and use.
In fact my software will not usually accept mail from dynamic dialups
anyway - even those not on the DUL.
It does seem that some people do find it beneficial to send mail
direct from their dialups (static or dynamic). I don't understand why
they think this is a good idea, and I think it has a number of
technical problems. However, I don't think that it's reasonable to
effectively forbid people from doing this solely for those reasons,
provided they're willing to accept the consequences - which will
include excessive retransmissions over their modem, long connect
times, and/or extended delays to the delivery of mail.
*But*, there is a definite problem with people using _dynamically
assigned_ dialup. This is because a dynamic dialup address cannot
effectively be blacklisted, and mail sent direct from such an address
cannot be monitored or controlled by the connectivity provider. Since
much of the net's current spam-fighting infrastructure is based on
blacklists of IP addresses and proactivity by ISPs, this is a big
That mail direct from dynamic dialups is a problem is recognised
throughout the community. Not only did Paul Vixie, the author of
BIND, and other leading lights of the Internet, decide to host,
support, etc, the DUL. Many ISPs prevent you from doing direct SMTP
by having their routers block outgoing SMTP or transparently redirect
it to their own mailservers. I think that this is going to become
much more common. Use of the DUL is becoming more common too - for
example, Cambridge University no longer accept DUL mail. Sites that
use DUL blocking report that it has very low false-positive rates -
some claim even lower than the MAPS RBL.
Now, I agree that for those people who want to do direct SMTP from
dynamic addresses it is inconvenient for them to have to change, but I
don't think this inconvenience is very great. Furthermore, the number
of people inconvenienced in this way is very low, and all the people
who are doing this are technically competent and have quite reasonable
alternative ways of having their mail delivered.
(IMO doing direct SMTP from a dialup accidentally or `by default'
almost certainly reflects a bug in the software or documentation or a
mistake by the user.)
It's clear, though, that the project will have to come to a common
decision about this. It's not just about what the project's
mailservers will accept. As I said in my other mail, since we all
need to communicate with each other, either every developer must be
forbidden from using the DUL, or every developer must either not send
mail direct from their dynamic dialup, or must be prepared to send it
differently if there is a problem.
Until a common decision can be arrived (if only by vigorous ranting
here until one side feels they can't win), this issue will keep
raising its head. We can't punt on it.
If we decide that developers are allowed to reject DUL mail then the
listmanagers should be allowed to do so too on the central systems.