Spampot test resultsPosted by Roger Keays, 14 June 2008, 1:45 AM |
You read a lot online about how spammers are 'harvesting the Internet' by collecting email addresses off the web. People have come up with all sorts of methods to obfuscate their email address. So I thought I would give them a bit of a test out. I created some dummy email accounts and posted them on this site each using a different method of obfuscation. The variations used plain old mailto links, tinyurl redirects, escaping with unicode, inserting invisible tags in the address with display:none and using images. The addresses have now been online for several months so I thought I'd release the results of this monumental experiment. Here they are:
- Plain old mailto link: 2 spams
That's where I stopped. Spammers don't harvest email addresses off the web. Writing crawlers to do it is too hard and uses too many resources. How many pages is it worth parsing to find just one email address? My guess is that their most productive sources are mailing lists, newsgroups and databases traded amongst themselves.
Mythbusted. Totally.
| << Neal Gafter on Java 7 Language Proposals | Back to Blog | Escaping XML in XML in JSF >> |
Comment posted by: Kauri on 06/10/2008 6:57:07 AM
Your experiment is quite interesting. Conclusion reasonable. However, my suspicion is, they may just crawl the "social networking" sites, forums and their likes to reduce their load... so we may be little careful there.
