Captchas solve the wrong problem. Abuse is what you want to prevent, and so you need to detect abuse, not humanness. Rate-limiting, reputation check, web of trust etc all are valid techniques.

Bots can be nice. Humans can be bad. Focus on the right problem.