One popular technique to counter spam email from filling one's inbox is to require would-be correspondents to "authenticate" themselves prior to the delivery of an email message. For example, the first time yo send email to a particular recipient, that recipient's system sends back an email to you saying "Please click this link within 24 hours." If you click the link, then the system reasonably assumes you're a human being who read the message and followed the directions. The email you sent is now delivered to the recipient, and all future emails from you to that recipient will be sent without any additional intervention. 0spam is a purveyor of this kind of technique.
I personally think this is an invasive solution to the problem, and feel that it causes more problems than it solves. For example, I can't count the number of times someone has contacted me by email, to which I reply, only to have my reply delayed until I jump through this authentication hoop. I used to jump through the hoops, thinking "Hey, this person deserves a response to their question." But I've just dealt with this for the last time: I will no longer authenticate myself to any of these anti-spam systems when I'm merely trying to reply to a message sent first by the person who uses such a mechanism.
Here's a clue: if you use such an anti-spam system, you should probably add my email address to your whitelist before you send me an email, if you expect me to reply.
Indeed, if you expect a response, then you should lower the barriers for people who would respond to you.
The same holds true for blog comments. Captchas, ignoring the accessibility issues, are a hurdle that visitors shouldn't have to jump over to respond to you.
I wouldn't mind authenticating myself to a centralized database before sending email if that would help. Things like the Sender Policy Framework seem like a good idea. But I certainly don't want to have to authenticate myself separately for everyone I send email to.
Amen, brotha!