Spam stat

I've (reluctantly, and for the time being) shifted over to using Apple's Mail.app exclusively--Eudora was too crashy, and Odysseus won't be ready for what I need anytime soon. (I considered trying Thunderbird, but I'm pretty sure it won't be significantly better for my needs than Mail.app.)

Decided to start sending Apple some bug reports and enhancement/feature requests, even though I doubt Mail.app is likely to get much attention from Apple anytime soon.

When you choose Mail > Provide Mail Feedback, the first thing that appears is a dialog box that lets you choose whether to send stats about your accounts and mailboxes to Apple. I was curious about what it was sending, so I scrolled through, and found the following interesting stats:

  • number of messages evaluated: 216254
  • number of messages evaluated as junk: 208900
  • number of messages manually marked as junk: 2653
  • number of messages manually marked as not junk: 737

I assume those are totals since I started using Mail.app in early June.

So I'm getting somewhere around 2300 messages a day, 97% of which are spam. (I think I was getting more like 4000 messages a day a few months back; not sure why the reduction.)

Fortunately, Mail.app's built-in spam filter appears to be even better than Eudora's (which in my experience is usually quite good but occasionally becomes less effective and has to be reset). (Um, but Mail.app's filter has had more false negatives in the past few days than it did before; I don't know whether that's a sign of it becoming less effective or just a temporary fluke.) If I'm interpreting their numbers right, there were 210816 pieces of actual spam; 1.3% of those were false negatives (spam that wasn't marked as spam), while less than 0.4% of the messages marked as spam were false positives.

The false-positive rate is actually probably slightly higher; there've probably been false positives that I didn't notice were non-spam and so I deleted them without manually marking them as non-spam. Still, I doubt there've been more than a couple hundred of those, and probably less. I would guess that the actual false-positive rate is still less than 0.5%.

Join the Conversation