I've been getting a bunch of FNs lately that are managing to avoid my Bayes DB. Invariably, they ALL seem to hit on BOGUS_MIME_VERSION (which I don't know whether is standard, but I implemented it locally and would recommend it in the distro if it's not there already), and it seems like most of them are ALSO hitting on DKIM_VALID. All of them are _just_ shy of the 5-point threshold.
John or others, might you perhaps sandbox a meta on BOGUS_MIME_VERSION && DKIM_VALID to see if that is worth something against the masscheck? In general, I guess BOGUS_MIME_VERSION might be enough to just be a poison pill, in which case a meta might not be necessary... but I'm curious if it performs even better with meta. Thanks! --- Amir
