The Great Spam Experiment

Cadzow conducted an experiment to track spam volumes over a period of time.

To do this, on 16/04/2004 we created 7 new mailboxes which are used only for this experiment. They were placed on the internet thus:

1. On our public website as a clickable hyperlink;

2. On our public website as a clickable hyperlink (all received emails will send a read receipt and download inline graphics);

3. On our public website as unclickable text;

4. Used as reply address in newsgroup posting;

5. Placed in body of newsgroup posting; and

6. Placed in body of newsgroup posting obscured as “(alias) at cadzow dot com dot au”; and

7. Used to subscribe to an IT-related website.

The winner of the first hit was the mailbox posted as a newsgroup post reply-to address which received a virus (W32/[email protected]) after about 6 hours.

The runner-ups were the clickable and unclickable links on the website which both received the same message on 19/04/2004.

And, just for fun, we also tracked spam messages sent to non-existent mailboxes at our domain name. This is an increasingly popular way for spammers to get their message out because it doesn't even involve knowing any real addresses. They simply send emails to random names at domains. Domains are publicly known, so fetching a list of domains is quite easy. This can explain why people who set up a domain name start receiving spam almost immediately if the domain hosting provider is collecting “stray” mail.

The experiment ended in June 2011. Spam arriving in inboxes reduced to virtually nothing due to the aggressive filtering and black-listing employed by ISPs and other service providers.

Results

Yearly Totals

As of the end of 2007, this is a 189% year-on-year increase from 2004. However, since the spam only started to arrive in bulk in 2006, it makes more sense to compare 2007 with 2006, in which case it is a 35% increase. (See also Spam/Phishing — What's Happening Lately? (2007) and Spam/Phishing — What's Happening Lately? (2009).)

† Extrapolated to full year.

Chart!

The right-hand scale (red) is all spam received, including stray mail.
The left-hand scale (blue) is spam received only by named mailboxes.
The black line is the linear trend.

Conclusions

• Putting an email on a website will attract a steady stream of spam. Whether the email is clickable or not seems to make only a small difference.

• Posting to newsgroups invites a quick burst of spam and then it peters away.

• The worst attractor of spam is a stray mail collector on a domain name.

• Spam delivered to Inboxes has diminished to a trickle due to ISP and hosting-provider spam filtering, blacklists and so on.