And it's a pretty narrow fraud, based on the article. It appears that he might not have been the individual in the photo, but maintains that he was treated the same way, and was playing an "iconic" role. The Army admits that it happened, that at least one person was tortured using specifically electric shock...nobody disputes that this stuff happened. So basically all you've got is a technicality.

Score another one for the Nitpicking Squad.

I'm afraid I found your last paragraph, the argument for bias, to be a little contrived. It might make sense if you assume that events are random and conform to a Gaussian distribution, but the real world's not like that. If most of the news out of Iraq is bad, then most of the articles about it will be bad news articles, and so statistically a constant error rate will produce more errors in bad news articles than good news articles. In fact, the evidence of bias would consist of an equal number of errors in the two categories despite their asymmetry; in other words, the "errors" are actually false information deliberately injected to provide spin. If the filter's linear, asymmetrical input produces asymmetrical output.