> Question is, how is it done? How do you protect
> a site from harvesting?
There are many tricks. The most common one is to use javascript to hide links
and html code. You can also make a dynamic downloader detector, by detecting a
specific threshold (number of pages downloaded per minute)

Some of them are documented here:
<http://www.httrack.com/html/abuse.html#WEBMASTERS>

But the more you protect a site, the less it will be accessible to prople (and
especially people with accessibilitylimits)