On Mon, 23 Apr 2001 11:21:50 +0100, Michael Quinion <[log in to unmask]>
wrote:

>* Auto-Delete= Yes,Full-Auto,Delay(18),Max(2),Probe(4)
>
>As the list is announce-only and there is one mailing a week, by
>my understanding every subscriber should be passively probed
>every fourth issue.

If the list mails weekly, there is no need to monitor longer than 8 days
[Delay(8)].  If an error recurrs for the next mailing, then a new 8day period
will start, lasting long enough to get you to the next mailing.  Delay(d) need
never be longer than the mailing period (in days) + 1 (just to be sure)

>Why then are there dead subscribers on the list with counts of
>14 errors?

Are they in fact still subscribed at the address listed?  (i.e. does a manual
delete of that exact address work?)

No error detection mechanism is 100% perfect.  Some addresses will still slip by
(the bounce may not be returned to LISTSERV, the proper information may not be
present to correctly process the message, the address shown as bouncing may not
be the subscribed address, etc).