Autolearn whitelisted emails

[Matt Kettler]

Kit Wong wrote:
> Hi All
> 
> Not sure if this is a feature that I need to switch on or something else.
> 
>  
> 
> Is it possible to autolearn whitelisted emails as ham?
> 
> This would ultimately help the accuracy of Bayes.
> 

This is in general not a good idea.

SpamAssassin intentionally doesn't consider its own whitelisting scores when 
doing autolearning.

Why? Because it's *very* common to make a small mistake in your whitelisting 
config, and if that gets abused and spammed, your whole bayes DB is now polluted 
with garbage learning. If you can't recover all the mis-learned messages and 
retrain them, you pretty much have to wipe out your bayes DB and start from scratch.