Best Practices for MWP 2010

[jmparsons]

I have been using MWP for longer than I can remember. I have found it useful primarily to identify quickly what is spam so I can delete it from the server before it hits my mail client. Because of my ongoing fear of false positives, I have been reluctant to use auto-delete for more than a few very obvious cases, like Cryllic or Hebrew language mail.

Now, with the increased volume of spam (for me, I'd say it's around 500/day), I'd like to have MWP handle more of the deletion automatically, but with the lowest risk of deleting good mail. I know one obvious approach--be liberal with whitelisting good senders--but there are many times when a first-time sender could possibly be flagged as spam and sent to the recycle bin without my ever seeing it.

What are other users doing to accomplish this and minimize the workload?

Joe Parsons

[anniebrion]

I have several filters that auto-delete for example Viagra, Casino, Banks I don't have account with, emails that have my domain in the from: & To: addresses etc...

[Sidewinder]

Rather than risking the auto delete you may want to just ratchet up the Filter to -200, set the override to mark for delete. and then use the "Group" with a set up that is to your liking but includes the Status Column. That will put all messages from that filter into a bucket by itself so that it is easy to scan but does not require manual intervention unless you find on that you need to reclassify. Your can turn off the display of the groups and column headers to pick up more grid space.

[stan_qaz]

I'd never trust the auto-delete myself, been using MW for years but auto anything just isn't for me.

I'd say the best thing to do is look at the example filters and see just how many of them you could use, on my box I get 80% of my mail into either a good or spam group on evaluation. The remaining 20% is split into likely bad, likely good and a couple groups that need checking carefully to make the decision.

As was suggested I group by Status so the messages all fall together for quick checking. My filters are preceded by a "A-", "B-", "C-" to get them to sort into the iffy/good/bad order I mentioned above to make looking at the iffy ones faster.

If you really want to do auto deletes save it for filters you have tested and are 100% sure of as a first option, a second option is setting the auto-delete by spam score to a high number and enabling that. With the combination of tools the auto by weight should be pretty good (better than any v6 option) once you get your Bayesian trained up, a few filters and your lists filled.