>I'm glad this topic was brought up... I was just trying to understand this
>myself.
>
>Jacob Haller wrote:
>
>>  It may help to think about this in terms of a discussion mailing
>>  list.  Say you've got a reasonably lively discussion mailing list
>>  that gets about 50 messages a day (with some really busy days and
>>  some days where nobody posts much).  The question is, how can you
>>  tell when an address is really bad?
>>
>>  One method is to wait four days (or whatever) and see if the address
>>  is still bouncing.  This is the Delay() method.  (Note that with this
>>  method you're not counting bounces or trying to match them up with
>>  the original messages in any way; you're looking on day 4 to see if
>>  there are any bounces for the address, and if they are, the address
>>  fails the test.)
>
>So with Delay(), it doesn't matter how many bounces occurred -- if, in
>Bill's original example, more than one bounce (the initial bounce) has
>occurred by the end of day 4, the address is eliminated.  Is that correct?

I think so, but I want to be precise.  The answer is:  Yes, if by "by
the end of day four" you mean "sometime during day four".  No, if by
that phrase you mean "sometime over the course of days two through
four".

>For an announcement (or any type for that matter) mailing list with a fairly
>static subscriber base, in order to have LISTSERV handle bounces with
>minimal mercy, would I want a shorter Delay setting?  Say one posting is
>distributed per week.

Minimal mercy would be Delay(0),Max(1); with that setting subscribers
are removed immediately after a single bounce.  With this setting no
daily monitoring report is generated; if that's undesirable you can
use Delay(1),Max(1), which still deletes subscribers after a single
bounce but does generate a daily monitoring report.

If you're not looking for an "immediate response"-type setting what
Delay() setting you'd use would depend basically on how often the
newsletter goes out.  For those kind of situations what would
probably work best would be to set Delay() to slightly more than an
integral multiple of the number of days between mailings and Max() to
one more than the multiple you used when calculating the delay.  For
instance for a monthly mailing you might use:

  Delay(33),Max(2)
    (two bounces in a little more than a month indicates two consecutive
     bouncing newsletters)
  Delay(65),Max(3)
    (three bounces in a little more than two months indicates three consecutive
     bouncing newsletters)

etc.  Some experimenting may be required to determine what works best
for you.  For a decently-sized monthly newsletter I'd be tempted to
be pretty hard-nosed as you're likely to see a decent number of
addresses go bad from one newsletter to the next, and it may not be
desirable to wait two or three months for them to get removed.

>In another scenario, say I have Delay(2).  If a subscriber's mail bounces
>once every other day, will they ever be deleted?

I had to look at some daily monitoring reports for some mailing lists
I run to make sure that I have this right.  The answer is no, the
subscriber will never be deleted.  Delay(2) means that LISTSERV is
looking for bounces on Day 1 and Day 2, so if bounces are always
seperated by a day the address won't be removed.  (This may conflict
with things I said earlier.  If so I apologize for my earlier error.)

Thanks,
--
Jacob Haller, Technical Support
L-Soft international, Inc
http://www.lsoft.com/