I'm a strong supporter of reCAPTCHA, since it helps to correct and confirm digitized texts from the Internet Archives. One of the things I do in my spare time is volunteer as a proofreader at Project Gutenberg / Distirbuted Proofreaders. (I'm the nerd who usually gets left rekeying entire indexes from ridiculously obscure books no one should ever have to read.)
We use the Internet Archives to store high resolution scans of texts that can be referred to in cases where the low-res proofing images are ambiguous.
P.S.--If you want to see the most difficult CAPTCHA images ever designed, try playing some of the games at the Live Search Club for a while. After a few games, you start to get interrupted by images that even I can't decipher, and i'm a P3 proofreader!