The CAUBE.AU Spam Survey
As part of the CAUBE.AU mission to measure the extent Australia's contribution to the spam problem, we have been deliberately placing email addresses in locations where they would be collected by spammers. To our knowledge, this is the only effort in existence attempting to quantify Australia's share of the problem.
The spam surveys only attempt to identify spam from entities unrelated to the recipient. That is, acquaintance spam is not measured at all. At the end of a survey period a human goes through all the suspected spams caught, manually removes anything that is not spam, and manually identifies whether the remaining spams are Australian in origin. A spam that is forged to appear to come from Australia is not counted as Australian.
The 1999 Survey
The 1999 survey began in February 1999 and ended in January 2000. Addresses were seeded to three places where spammers are known to harvest addresses:
- USENET News
- Each address was used as the "From" address of a single USENET news article. Addresses seeded into USENET news were seeded early in the survey.
- Web Site "mailto" addresses
- The addresses were made available on a web page that would not normally be discovered by a person browsing the web, but would be discovered by programs designed to harvest email addresses from web pages.
- Internet Contact databases
- One address was seeded into each of the AUNIC and InterNIC network contact databases.
The addresses used included both Australian (".com.au" and ".org.au") addresses and American (".com") addresses.
We also attempted to include some 3rd party addresses in the survey, but there was considerably difficulty in determining which of the messages received at these addresses was genuinely spam, so we were unable to use the results from these addresses.
The Results of the 1999 Survey
Outlying Points
The email address that received the most spam appeared as the first email address on a web page with multiple email addresses on it. That address received 136 spams, compared to 27 for the next most spammed address. It appears that somebody may have subscribed that address some pornographic mailing lists hoping to affect the results of the survey – the address in question received 119 of its suspected spams from the same pornographic newsletter site, and none of the other addresses received spams from that site. Since there appears to have been a deliberate attempt to compromise the survey in this case, we have removed this address from the sample.
The Numbers
Of the spam received at the included survey addresses, 16% was clearly Australian in origin. Australian spammers represented approximately 10% of the spammers caught in the survey, but none of the Australian spammers caught used multiple addresses – which is common elsewhere – so the real figure may be higher than this.
The effectiveness of an email address exposure for attracting spam is almost identical for posting a single message to USENET as it is for posting the address to a single web page. While this survey did not attempt to measure the effects of repeatedly using the same address for posts to USENET, anecdotal evidence suggests that repeated posts to USENET result in proportionally more spam.
Since some of the addresses used to post to USENET were posted in groups in the Australian newsgroup heirarchy, it is worth measuring the ratios for spam sent to just the addresses seeded via web pages, which could be expected to show more neutral results. When we checked the figures for these addresses in isolation, we were surprised to find that Australian spam accounted for an incredible 18% of all spam received at those addresses.
If we consider only American (".com") addresses posted to a web page, Australian spam still accounted for 11% of all spam received.
As various states in the United States ban spam, there is a strong movement to contract the spam out to other countries. Australia is unfortunately the primary choice of destination for spammers taking this approach, as is shown by the note shown below, which was in a spam caught in the CAUBE.AU spam survey on 10th April 1999. This spam claimed to be from gh5@nla.gov.au, and was relayed through the National Library of Australia without permission (this was not counted as one of the Australian spams, as it was clearly forged to appear to be from Australia). The real sender in this case appears to have been in Las Vegas.
NEW - AUSTRALIAN BULK SERVE9 (sic) GUARANTEED NOT TO BE CANCELLED. NO SETUP FEE. CALL FOR MORE INFORMATION.
Curiously, while spam has a reputation internationally as being the medium of pornographers, con artists, and pyramid scheme participants, not a single one of the Australian spams caught in the survey fit this profile. The most prolific Australian spammers included an automobile sales web site, an online investment magazine and a real estate investment company. Other spammers included computer retailers, a tourism promoter and a Melbourne radio station.