Both of these bayesian spam filtering packages do a great job. i use SpamSieve on my PowerBook G4, and use POPFile on my PC.
i have to say that the process of training POPFile is a lot less of a hassle than training SpamSieve. in Entourage X you have to run an AppleScript in the Scripts menu to tell SpamSieve if it is a good or spam message. you train POPFile via a web page, a much nicer way to handle this IMO. having the entire UI for POPFile web based makes editing settings a lot easier. i really like the "magnets" feature of POPFile, wich allows you to setup whitelists and blacklists in a snap. i use magnets for all the listprocs i am on, so they always go to the correct bucket.
if i were still using my PowerBook G4 for my email full time i would install POPFile on it. i think i have had POPFile installed for 3 weeks on my PC, and its just hit 90% accuracy, and i am pretty happy with that.
i think more mail applications need to incorporate bayesian filtering. Apple's Mail now has bayesian filtering, it would be nice to see others do the same.
