Manufacturing balance.

Sadly No brought this to my attention. Apparently there was a consevabot called smarterchild, a computer program that responds coverstationally with AIM users. Well apparently part of the program samples the frequency of opinions that it runs across in conversations and apparently comes out on the side of the more popular opinion.
Here is the story.

Computer savvy and interested in presidential politics, 13-year-old Erin was upset to learn that America Online's Instant Message (AIM) robot, "Smarter Child" favored Democrat John Kerry in this year's election. Erin, whose last name is being omitted in this article, is a fan of President Bush.

[...]

But when Erin told the robot that "George Bush is awesome," she was shocked by the response. "No way. George W. Bush is way uncool," the reply stated. She asked, "Do you like George W. Bush?" and the program replied, "I'm a Kerry supporter myself."

SmarterChild the name of the program has a webpage up to explain that they have cleansed the the robot of it's treasonous ways. It turns out there is a reason that this 13 year old bush supporter, who doubtless wear with honor his jump wings as a member of the 82th fighting keyboarders had to put up with the trenchant AIMbot.

A recent story in the press misrepresented how SmarterChild was programmed. SmarterChild was never programmed to oppose George W. Bush.

SmarterChild has been programmed to form opinions based on the aggregate opinions of its users. If a certain number of users express a favorable opinion on a subject (e.g., "I like ham," "I think ham is good," "Ham is wonderful!" etc.) and the number of favorable opinions significantly outweigh the number of unfavorable opinions on the subject ("ham is lousy," "I don't like ham," etc.) then SmarterChild takes on the opinion of the majority (in this case, SmarterChild would tell you that it liked ham).