Judge and Economist Richard Posner has an analysis of media bias in yesterday's New York Times.
An economist friend sent me that link, and below is my reply. I am impressed by how much jargon and verbiage I've been using in my replies to him recently. I think I've been properly socialized into Economics...
the theory sounds largely consistent with the recent economic papers on media bias, such as alesina and mullainatahn, groseclose, or by our fellow gsb student william minozzi. that's always been my view of things. the new york times appears perfectly unbiased to its audience, as does fox news. it's a simple case of horizontal product
william has the interesting result that regardless of the owner's bias, the profit motive is strong enough that it becomes irrelevent, and instead the bias is targetted to the audience's opinion as posner suggests. william's theory also makes the simple point that bias is not so bad so long as there is at least some competition.
someone who wanted the truth could just consume two newspapers. watch both cnn and foxnews. etc...
one problem with posner's theory is that the cost trend is not so unambiguous. network effects makes the cost of entry go up not down. hence, most towns have fewer newspapers than they used to, most now only have one. (someone has an empirical paper on testing teh bias using this fact recently). similarly, radio is now dominated by a few conglomerates like clearchannel, etc...
also, he is a little two cheerleadery for the blogosphere. the hayek marketplace for ideas myth is nice, but you could cite groseclose on valence here.
his view is again the one of horizontal product differentiation in a hotelling sense in bias space, where perfect competition by the blogs leads to a condorcet jury winner effect of finding the median. (wow, that's a lot of jargon)
but, mainstream press has other dimensions (aside from bias) that the blogs can't replicate, such as reputation, and thus accountability, and money and a budget and training, to report on iraq for example, access, increasing returns to scale... and who's to say the median view is the truth.