RE: Is predictable spam filtering a vulnerability?
> -----Original Message-----
> From: PSE-L@xxxxxxxxxxxxxxxxxxxxx [mailto:PSE-L@xxxxxxxxxxxxxxxxxxxxx]
> Many sites employ SpamAssassin and the like to simply FLAG
> messages and pass them along to the intended recipient, who can then
> employ their own filter process within their email client
This is what I do. Spam is tagged by a statistical filter, then tagged
messages are filtered into a "Junk Mail" folder by the user's email client.
In a corporate environment, where silently dropping mail from a customer is
totally unacceptable, this is a good compromise. The user can skim their
junk mail folder now and then and pick out anything that looks like it's
important. (I do this about once a day; only takes a few seconds. A
non-spam message in a folder full of spam tends to be surprisingly obvious.)
> Of course, what do I know? Up till now, I assumed
> intelligent folk could
> manage to send a reply to a listserv without also sending an
> unnecessary
> carbon to the original message poster, and if not, at least courteous
> people would pay attention to the sigline making such a request...
If I did this earlier, I'm sorry. I correspond with a lot of people who
prefer to get carbon copies of list replies, especially on moderated lists.
I'm also not in the habit of reading signatures because they tend to be a
waste of time. After seeing several dozen with bogus disclaimers and the
like in them you lose interest...
> John Fitzgibbon wrote:
> >Archiving the dropped mail *and* terminating with a 5xx would be a much
> >better approach.
To me that seems *totally* broken. A 5xx response means you didn't deliver
the mail, and the failure was permanent. Terminating with a 5xx and then
delivering the mail somewhere isn't kosher; in fact, it's the worst of both
worlds. You've still accepted the spam, *and* you've potentially created a
DSN.