[Rspamd-Users] Autolearn rejected messages as spam

[Christoffer G. Thomsen]

On 05/12/2018 15.59, Ralph Seichter wrote:
> From rspamd's perspective, such a high score means that the message is
> almost certainly spam. What would be the point of "learning" something
> that is already known?

Maybe that particular message was rejected because it was sent from a
host listed in a bunch of RBLs, but next time it or a similar message is
sent maybe the sender IP won't have been picked up by RBLs yet. Then it
would be useful to have bayes recognize it as a spam.