on the whole i'm pretty impressed with mail's spam filter. i'd say it probably weeds out about 2/3 of my daily crop of viagra and penis extension offers [although i still think the logical place to perform spam filtering is on the server, not after i've already downloaded it! - but that's another rant for another day.]
however, i've noticed that some spam that i get regularly from the same source [some employment agency bollox] is always missed by mail's junk filter, even tho' i mark it as spam every time. i thought the spam filter was meant to use some kind of bayesian system, whereby it learned and got better as it went along, so why is it not 'learning' that these particular emails are spam, even after i keep telling it?
does the spam filter only actually 'learn' while it's in training mode? or is it an ongoing process? after letting it get up to speed for a few weeks at the start, i switched from training mode to 'custom' so i could filter the junk into a specific folder. do i need to switch back to training mode again to let it 'learn' about these new emails?... and if so, what's the point then, of ever coming out of training mode, if the learning process isn't ongoing in other modes?
[so many questions - so little time. i don't think the inner mechanics of mail's junk filter is very well documented at all.]