However, this sounds like a "reverse bayesian spam filter" that instead of filtering out spam is filtering out "most important email" and learning over time ?
It shouldn't be too hard adjusting SpamBayes or similar filters to do this, should it ? Anyone know about any solution for Outlook ?