reCaptcha if you don't know if a captcha service that provides the user with two words, one of which is for authentication while the other is used to aid in the transcription of old books and other literature.
Lately, however, I've found that most of the provided captchas are extremely easy... it will be one very clearly written word and one absurdly written word that likely contains a number or a foreign symbol. It's just very apparent which word is the actual captcha and which word the computer is unsure of. I suppose it's running out of words that it's unsure of, but I find that a bit hard to believe. I would have imagined it would have implemented a system that filters out obvious words based on how easily they're answered and how rarely they're skipped.
Has anyone else noticed this?