What happens when a false positive is trained as good, then I change my mind and train as spam

[toms (Tom Simpson)]

If I train a false positive (good mail in spam folder) as “not spam,” I presume a whitelist entry is created, correct?
Then what happens when I later decide that I not longer want these messages and (re)train it as Spam? Is the whitelist item removed? Have I created a conflicting entry and SpamSieve gets confused? Do I need to manually remove the whitelist entry?

[Michael_Tsai]

@toms

If I train a false positive (good mail in spam folder) as “not spam,” I presume a whitelist entry is created, correct?

Yes.

@toms

Then what happens when I later decide that I not longer want these messages and (re)train it as Spam?

You should be training based on whether a message is spam or not, not whether you want to see the message. So that shouldn’t change later. Non-spam messages that you don’t want to see are usually easily filtered by regular Mail rules below the SpamSieve rule.

@toms

Is the whitelist item removed? Have I created a conflicting entry and SpamSieve gets confused?

If you train the same message again, the whitelist rule will be disabled and the corpus will be updated to treat the message content as spammy. If you train a different message from the same address as spam, the whitelist rule will be disabled and the corpus will get messed up with conflicting information.

@toms

Do I need to manually remove the whitelist entry?

No, it is usually better to leave the disabled rule in place.

[toms (Tom Simpson)]

Thank you, that is helpful.
An additional question about the way things work (looking at the whitelist creates more questions than answers!)…
I noticed a huge volume of entries for:
from(address) equal to: (lots of digits and characters)@groups.facebook.com
I don’t think we have trained these as good, and they are good- they arrive in the Inbox and we trash or file them On My Mac as desired. Does SpamSieve automatically create whitelist entries when it predicts that a message is not spam? Will this large volume of entries (500+) impact performance to a noticeable degree?

(on a lighter note- how is it that a company which creates Mac OS software has a name that is reminiscent of a PC-DOS command line entry? hahaha)

[Michael_Tsai]

@toms

Does SpamSieve automatically create whitelist entries when it predicts that a message is not spam?

Yes.

@toms

Will this large volume of entries (500+) impact performance to a noticeable degree?

It’s won’t impact performance. If you really have a lot of unneeded rules (100K) you could delete them to reduce SpamSieve’s RAM usage.

@toms

how is it that a company which creates Mac OS software has a name that is reminiscent of a PC-DOS command line entry?

I never really thought about it that way. The term “C-Command” comes from Linguistics. I was thinking that the Command key is a Mac-specific thing, and the C relates to the software being written in the Objective-C programming language. (That was in 2002; now most of our new development is in Swift.)