NYT las vegas advertising at the bottom http://phys.org/news/2012-10-candidates-web-ads-investing-heavily.html
There are several stories in the media about how social media can be used to predict elections. Between facebook and twitter, there is a lot of free information floating around. According to Quantcast almost 70% of Americans that are over the age of 18 visit facebook and 44% of the same population is on twitter. This is a huge proportion of the United States. This is more people that any poll could reach. This could be a huge asset to researchers because of the vast subject pool and the freedom (as in free of cost) of information.
In 2010 a group of researchers looked into using twitter to predict a German election. Their results were published in a paper called "Predicting elections with Twitter: What 140 Characters Reveal about Political Sentiment"
The researchers used an equation that compared favorable and unfavorable mentions of each candidate. They only tested their equation with high profile elections. In the end they predicted that it could be possible to use the data from twitter to accurately predict a winner in an election.
In the two years since this research Twitter has many more users. There have also been many critiques of their findings.
The MIT Technology Review posted an article detailing the many ways in which twitter can not predict an election (as well as box office revenues and the stock market).
A Cornell student was published for his article "I Wanted to Predict Elections with Twitter and all I got was this Lousy Paper" after he tried to recreate the German study and it failed. In it twitter predicts with the same accuracy as chance.
The Atlantic ran a detailed critique with the subtitle "Twitter may provide useful clues for detecting a spike in influenza, but its users are too different from the population at large for counting votes before they're cast." In it the argue that the population of twitter is a niche of the American population and therefore it can not represent us all.
I believe that there is a lot of potential use for social media in polling. I would prefer pollsters to track my social media than to call or stop by. I feel that it is far less invasive.
However, there is still a lot of work to be done before social media could be used for this.