Machine politics

«Chat bots» or «social bots» disguise their identity as algorithm-controlled automatisms. They pretend to be real people with real opinions and information. Which means can they discover effectively? How should public language be evaluated, which adapts in style and information content to the bot language?