From: Eugen Leitl (eugen@leitl.org)
Date: Thu Apr 11 2002 - 03:01:27 MDT
On Wed, 10 Apr 2002, Robert J. Bradbury wrote:
> In contrast, the support for the MAPS Realtime Black Hole List
> and Relay Spam Stopper (which were free before Aug. 1, 2001)
> could be "built into" the "sendmail" step which would allow
> it to reject the messages as they were received.
I'm pretty much against this, for the following reasons. The transport
layer should be content-agnostic. As soon as you try to make the transport
layer play the role of filters, it introduces inefficiencies, and -- much
more importantly -- breaks transport in many interesting and unanticipated
ways. Fixing brokenness introduced by initial attempts to fix it is hard.
It's okay to introduce friction/throttling at the transport layer, though,
which is what hashcash does. It's an attempt on the spam side to invest
CPU resources at message delivery, making bulk email prohibitively
expensive. (A mix of hashcash and cryptographically authenticated
whitelisting alone could kill spam quite easily).
I personally feel my current needs to be adequately addresed by
SpamAssassin. I pore through the spam folder once a day or two, and this
allows me a highly efficient way of dealing with it (usually I don't even
have to look at message body, as headers and subject usually say it all).
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:13:24 MST