We provide several captcha options that vary in ease of use vs protection.

I would base any decision only on real feedback from real users. I do not know of any success rate statistics based on typical users. While inconvenient I think most real people can solve any of the captchas, though with Re-Capthca some things may be easier to solve than others within the same Captcha because they are using OCR to transcribe books, the user can click to get a new challenge if they can't solve it.

Success rate based only our own or my own site traffic would be meaningless and more likely a high failure rate just indicates a lot of bot traffic.

Best,

Joe

All I know is that I've looked into possibly integrating alternate CAPTCHA choices in mojoPortal, but each time I find one that looks interesting (image based, video based, etc.), they always turn out to be easier to compromise than expected. The latest news I've seen on this was how Stanford University researchers broke the video-based NuCAPTCHA last week. These same researchers were unable to break ReCAPTCHA last fall, so I'm sticking with that for now.

Jamie