Author: P.Benny (3 Feb 05 10:07pm)
I'd like to point out that "advanced munging" (http://www.projecthoneypot.org/how_to_avoid_spambots_2.php) of the email address in the href attribute could also contain random CR/LF characters (in their raw form or encoded as ...). These are ignored by browsers when found in a URI , while at least some harvesters may be confused by them.
This doesn't bother any recent browser (though Lynx doesn't like it).
If displaying the email address in HTML, ­'s can be inserted in the address and won't render visually (except Netscape 4.x-) as long as the email address isn't likely to break on several lines.
However, this isn't cut & paste friendly.
Also, feeling a bit inspired by the project, I have moved my site's email addresses to a single contact page generated by a script which appends the client's IP address and current time to the email's alias part, hoping that some spam will arrive to such generated addresses. Then the harvester's own IP and request time would be revealed (they can be matched against server logs to verify authenticity). Advantages could include blocking a harvester's specific alias for example.
If anyone has comments regarding whether this is hopeless or not, I'd be interested.
 RFC-2396 http://www.ietf.org/rfc/rfc2396.txt p.33 Recommendations for Delimiting URI in Context
Post Edited (4 Feb 05 1:11am)