Well that is one method I have seen a program that is accurate to about 90%
of the time. Even when they come out and make it more difficult they adjust
the program parameters a bit and they get it right. I also know of the sites
using people but it is in efficient since spammers send on the order of
millions of emails every night. 

Bttom line it is a losing battle unless the email system redesigned. 

Sean

-----Original Message-----
From: squeak-dev-bounces at lists.squeakfoundation.org
[mailto:squeak-dev-bounces at lists.squeakfoundation.org] On Behalf Of oxe
Sent: Thursday, September 01, 2005 7:11 PM
To: squeak-dev at lists.squeakfoundation.org
Subject: Re: Wiki Vandalism again

>>there may be newer better methods, but
>>the type-in-the-letter hacks i know of are somewhat
>>labour- and resource-intensive on the spammers, so i think it 
>>might be worth a shot.
> 
> 
> The method used to defeat that technique has been to have the bot pass the
> type-in-the-letter image through to a pron site. Users at the pron site
type
> in the keyword in order to see the next image. Then the bot takes the
> results back to complete the challenge at the first site.
> 
> Insidiously clever, co-opting humans to do the jobs which are too
difficult
> for the computer!

yes! ingenious!

that's the method i know of as well.

i think it's still fairly labour-intensive for the spammers tho
and they tend to use it on bigger fish such as getting yahoo or hotmail
accounts.

it seems like the image could be dickied to discourage that.

i chatted with some folks at yahoo once about like
putting the yahoo logo in the image,
but they were reluctant to do that because
then the Y! logo would be seem to be associated w/ pr0n sites.

anyhow.

if it's easy to implement it might still
reduce wiki spam a bit.

orion