admin Yellowikis wrote:
It's a neat idea, but there are several serious problems with it,
particularly so for a wiki environment.
Most seriously, it requires client cooperation -- a spambot can simply
send the expected number of bytes along with the request. It can only
therefore be effective against humans doing manual cut-n-paste of spam
text... or real genuine humans cutting and pasting text.
It requires JavaScript, so either it cuts out submissions from humans
with JavaScript disabled or unavailable or it must accept submissions
without the number, making it trivially defeatable by humans...
For wikis, we send out the entire current page text and take back the
entire text with modifications. Determining number of changed characters
is not trivial; think again of cut-n-paste, rearrangement of sections,
deletion of sections, etc. This makes it much harder to get meaningful
information out of such a comparison.
-- brion vibber (brion @
pobox.com)