SPAM is a truly egregious problem. This article does some analysis of popular spam filtering software. It is no surprise to me that signature-based programs tend to have slightly higher false positive rates and that Bayesian filters work best when trained properly, and have low or zero false positive rates.
• 1 min read