I've actually seen this type of feature employed on a financial site I use here in the US call Mint.com. Any time I've tried to delete an account I manually have to type 'DELETE' and click OK in order to proceed. It's not so much a CAPTCHA as it is an acknowledgement that the person has actually read the message. I agree that this method provides all the fair warning needed. If they still manage to perform some irreversible action after having to type in a specific phrase then they deserve what is coming. Logging their name into the Notes Agent Log tagging them as the offender would be a good option at that point :)
I've actually seen this type of feature employed on a financial site I use here in the US call Mint.com. Any time I've tried to delete an account I manually have to type 'DELETE' and click OK in order to proceed. It's not so much a CAPTCHA as it is an acknowledgement that the person has actually read the message. I agree that this method provides all the fair warning needed. If they still manage to perform some irreversible action after having to type in a specific phrase then they deserve what is coming. Logging their name into the Notes Agent Log tagging them as the offender would be a good option at that point :)