What we were trying to accomplish was the auto-deletion of bad addresses
and general error handling without a lot of time-consuming human bounce
administration.  However, recently it seems aol.com, yahoo.com,
hotmail.com, themail.com, and others have started rejecting bounces
outright, causing large numbers of good subscriber addresses to be deleted
from our lists.  We then tried to disable probing all-together and simply
handle the bounces ourselves, however this is not possible due to the fact
that we use Mail-Merge for our distributions, (see post from Ben Parker).
Now we are simply trying to find a method of dealing with probes so that
good subscribers are not deleted or told that their email accounts are not
receiving mail, when they obviously are, and we are not overwhelmed with
the administrative tasks including fielding probe failures.  I am simply
trying to find a happy medium where our subscriber list is protected and we
don’t receive 1000+ error messages a day.  I agree whole-heartedly with
your statement:

Any procedure that you successfully implement today may not satisfy
tomorrow -- "it" tends to be a moving target.

Probing worked great for us until last month, when, suddenly, the rules
seemed to change.  Now we can’t trust it and I’m not sure what to do next.
I have studied the documentation thoroughly, however I still don’t quite
understand how probing works and the subtle art of probe tuning to find our
happy medium.