I wonder, at a deeper level, if the political divide in the US is fueled by the "bubble" that google allows: many years ago, people read newspapers covering all points of view, but google news lets you select which sources you draw news from (allowing you to select only fox news and drudge report, for example)
That said, there's unquestionably room for improvement.
I do expect that bubbles have got less regional.
What would help is more meta data available to the reader. So, instead of measuring an unmeasurable "truthiness" (because really, what is truth, in the philosophical sense...) how about a Chrome plug-in that pulls down meta data from other sites that have written about this outlet.
So, you read NYTimes, and your little plug-i pops up some links to http://en.wikipedia.org/wiki/Jayson_Blair or some such similar thing. You can read up on that and see that there have been issue sin the past, but for the most part, NYTimes remains reputable.
I guess it would need some sort of Hacker News-like inaccuracy tracker. Post bad stories, more up-votes, more likely to be detected as relevant meta info on the TLD it comes from.
A front page story like "Pentagon Lifts Ban on Women in Combat" is probably highly accurate, breaking news less so. Op-Ed, might as well be a random blog.
Still, BBC, Al Jazera, NPR are where I go. That, and hyper local sites.
What I would love to see is a method for tracking or indexing reputation of a site by some automatic method. As it is, I can kind of infer how reliable a site is based on the language I see, and the types of headlines, and whether others have told me it's reputable. I would love to see that automated somehow. News aggregators try to solve this via upvotes, downvotes, or flags, but there are plenty of reasons to upvote that have nothing to do with reliability per se, and trying to use multiple upvoting buttons seems silly, because at the end of the day rank is a one-dimensional property, so you'll have to combine "funny", "reliable", "interesting" into some single number.
It's a "hard problem" but since the value of information depends so heavily on its accuracy, solving it would appear to be very lucrative and valuable to a society in which the quantity of available information grows by leaps and bounds each year. Google seems to have solved "relevance" and I would love to see someone solve reliability.