Mickey Kaus has a theory that robo-polls might actually be more accurate than polls that use actual humans.

Who won Pennsylvania? Final conventional Quinnipiac Poll (conducted by human beings): 48-42 Specter lead. Final Survey USA robo-poll (conducted by machines): Specter 48, Toomey 48. Actual result: Specter 51, Toomey 49. Advantage: Robo-poll! ...

Survey USA's robo-poll also did well predicting the results of the Georgia, Oklahoma, and Ohio Democratic presidential primaries and the Illinois Dem Senate primary--though they were off in picking the winner of this year's Iowa caucuses. ...

Explanation: Why might robotic polling (which uses a recorded announcer's voice) be more accurate? Kf's Mystery Pollster--who is not associated with Survey USA but is toying with the idea of using robo-polls--gives three reasons:1) They can poll more people more cheaply, creating a larger pool from which to pick only "likely" voters; 2) On some questions, people are less likely to lie to please a machine. In particular, they'll often tell a live human of course they're going to vote (when they're not) because they think voting is what's expected of them; 3) Robo-polling duplicates the impersonal and anonymous nature of voting itself. Often that results in fewer inhibitions about choosing--and a lower "undecided" vote. ... Meanwhile, the key presumed defect of robo-polls--they have a ower response rate because it's easy to hang up on a recording--may not have much of a biasing effect. ...