What to do with backscatter?

Hello all,

My small home network is reasonably free from spam thanks partly to a careful Internet posting policy, and mainly to the wonderful SA which catches about 98% of the spam I do get.

One thing that does plague me however is a periodic rash of Non Delivery Receipt messages (I've just had one now - about 10-15 or so). These score anywhere between 1.2 and 11.1 but mainly around the 3.7 mark (below my spam threshold of 5.0). They all hit the ANY_BOUNCE_MESSAGE rule.

I always wonder what to do with these. Should I: a) Put them into my spam folder for bayes training with sa-learn; or b) Increase the ANY_BOUNCE_MESSAGE score from 0.1 to something likely to bite; or
c) just live with it?

What is the best thing to do with stuff? I don't want to mess up my bayes, but I would like to keep this out of my inbox...

Thanks in advance.

AD