[TAG] The Spam Reduction Act :)

[Jimmy O'Regan]

Ben Okopnik wrote:

>I've built a "zero spambot bait" mechanism for email addresses. It's
>already implemented for the authors' emails (you can see the result in
>the bios); unless someone sees a problem with the idea, I'm going to
>extend it to any addresses exposed in the content as well.
>
>The way it works for authors is that I've made all the email addresses
>in the bio "source" files private (i.e., they don't show up anywhere in
>the published content.) Next, I set up a little script -
>$LG_ROOT/bin/emailpic for those who have updated their SVN repository
>lately - that converts that string to a tiny PNG image, ~200 bytes. I
>also updated Mike's Cheetah template for bio generation. Last, the only
>thing that's contained in the 'email:' field is the name of that file
>(if I knew Python, I'd just key it off the file name, but I don't.)
>
>The result is that instead of the entity hash that we've been using to
>hide the addresses - a technique whose usefulness decreases as time goes
>on - we now have
>
><strong>E-mail: </strong><img src="../gx/authors/email/okopnik.png"><br>
>
>(Let them 'bot *this*.)
>
>The way I see it working for content is to do the conversion for any
>rfc822 string, name the image with a per-article sequential number, and
>place the images in the misc/<author> directory, which is where the
>images for a given article normally go.
>
>Thoughts, anyone?
>  
>
The obvious thought here is that if, as Rick mentioned 
(http://linuxgazette.net/mailman/private/tag/2004-June/002597.html) 
sites like Hotmail are having to use the Captcha test 
(http://en.wikipedia.org/wiki/Captcha), it follows that spammer's bots 
must already use OCR.