Jeremy Hylton : weblog : 2003-10-20

40% Spam

Monday, October 20, 2003, 1:05 p.m.

This morning I had a chance to measure the amount of spam I get -- about 40% of incoming mail. I was away for the weekend, so a lot of email collected on the POP servers; that gave me that chance to count spam from a large batch of messages. I fetched 1618 messages; Spambayes marked 628 as spam and 71 as unsure.

The percentage of spam may be lowering during the week. It's hard to measure since I collect a little email at a time, so I don't know how much ham comes in. Since the test period was over a weekend, I didn't get much work-related email. On the other hand, python-dev was busy -- more than 100 messages.

I get python-list delivered by email. That means I get a lot of ham traffic that I don't read closely. If half of my inbound email comes from python-list, then I'm getting more like 85% spam.

The number of unsure messages was unusually high. I received about 15 uncaught bounce notifications from Mailman; I've never trained on Mailman bounces as ham or spam, so they always come up unsure. I also received a lot of duplicates of new spams. New spams sometimes show up as unsure; usually, you get several spams at a time.