I'm curious if a little random number generation would do the trick or if the same robots that can read the Captcha's could figure out this, too.  I guess I don't know enough about how the bad guys are able to defeat image Captchas on websites...but if you've got a .NET site, creating some simpl code tied to your comment or blog post submission button that generates random numbers and checks to see if the end-user can do the simple math would be an easy way of stopping bots from comment spamming.

This has always been my thought...and something I could whip up in 5 minutes with Visual Studio and C#.

int Rand1         //Random #1 (1-20)
int Rand2         //Random #2 (1-20)
int RandSign     //Random math sign (1-2).  If 1, sign is plus, if 2, sign is minus.

Display
Random #1    Random math sign    Random #2
Type Answer here: ______
Submit Button

when the button's clicked:
if (Convert.ToInt32(this.txtRand1.Text) + Convert.ToInt32(this.txtRand2.Text) != Rand1 + Rand2) ||
(Convert.ToInt32(this.txtRand1.Text) - Convert.ToInt32(this.txtRand2.Text) != Rand1 - Rand2)
{
      //Failed to match, probably robot, do not let them in
}
else
{
     //They did good math, probably human, let'em in.
}

Are these Captcha defeating robots able to screen scrape or something and that's why something like this doesn't seem like it's implemented anywhere on the web?  Any thoughts on why this wouldn't work are welcome.

-Corby-