How CNN’s Facebook Election Tool Spreads Misinformation
As election season nears, more and more people are discussing the candidates and their positions online. Facebook and CNN want to be at the center of that conversation – but they have a long way to go if they really want to add to the debate.
The two organizations announced Facebook-CNN Election Insights, a real-time tool that shows which candidate Americans are talking about online, sliced into a number of user-configurable demographics. The tool was architected by Mass Relevance under CNN’s guidance. Mass Relevance also provided a curated Twitter tracker during the recent Olympic games.
Following on the heels of election-oriented services from Twitter (The Twitter Political Index) and Amazon (the[Amazon Election Heat Map 2012), Facebook-CNN Election Insights may be a noble first step, but it lacks the essential data and context it needs to be truly useful.
For Viewers And Reporters
CNN sees this as a online resource both for viewers as well as its own reporters. Facebook-CNN Election Insights will be used during CNN broadcasts, including regular segments on CNN’s “The Situation Room with Wolf Blitzer” as well as in CNN.com and CNN Mobile campaign coverage, the news network explained. CNN’s daily “Political Gut Check” newsletter and the U.S. Politics on Facebook page will feature highlights from the tool.
“Facebook is naturally a place where friends engage in political discourse, and we’re pleased to announce that the Facebook-CNN Election Insights tool will offer an interactive, real-time glimpse into how and where this conversation is taking place across the country,” said Elliot Schrage, vice president of corporate communications and public policy at Facebook, in a statement.
The problem – or perhaps the benefit to CNN – is that the tool is sadly lacking in context. On the CNN site, for example, Election Insights shows that President Obama showed a spike in online conversations mentioning his name around Aug. 25 at 8 AM. But why? Were the mentions positive or negative? In what context was Obama’s name mentioned? None of this is explained, and it’s up to the reader, or the on-air CNN anchor, to explain.
In a sense, there’s even less information than the ubiquitous stock-market graphs so common on nightly business reports. Those, at least, display both the relative change as well as the absolute value. The CNN tool reports that 19% more people are talking about Romney “day over day”, and 19% fewer people are talking about Obama. It’s not even clear what that means, given that users can adjust the time periods from 7 days down to the last 12 hours.
A Shortage Of Context And Transparency
Facebook data is being used, of course, and it is being pulled and anonymized from Facebook’s entire U.S. member base. We still don’t know actual numbers, however, and any pundit worth his salt would appreciate knowing if Obama’s name is being mentioned 50 million times, or just 5,000.
Amber Quist, director of marketing for Mass Relevance, said that her company had “special access” to the Facebook data, but declined to comment on whether or not the company had access to its social graph API. She also wasn’t able to say whether Mass Relevance and the CNN tool taps into all of Facebook’s U.S. users, or just a subset. (Facebook representatives didn’t respond to requests for additional comment.) “It’s capturing the buzz coming across the Facebook platform,” Quist said.
Mass Relevance doesn’t provide the kind of detailed sentiment analysis that Radian6 and other competing companies offer. Mass Relevance’s “poor man sentiment analysis,” Quist said, can provide only positive or negative assessments. More to the point, “it’s not incorporated into this experience at all,” she said.
The Election As A Big Data Problem
If there’s one big data problem that users will care about this fall, it’s the election.
A ”Facebook election” could radically rewrite democracy with a simple request: in late October, a site message could simply ask the site’s users to “like” either candidate. That poll would most likely be the most comprehensive poll any agency would issue, outside of the election, especially if Facebook was able to provide demographic analysis of which users preferred which candidate. And if Facebook asked its worldwide userbase to participate, global sentiment could be factored in.
Until we know more about what data Facebook is providing to CNN, however, Election Insights cannot fulfill its potential. As with any poll, greater transparency equals a corresponding increase in credibility.
If Facebook or CNN does provide this transparency, Election Insights could become a useful tool. But without it, this might be the polling equivalent of the ”hologram” CNN used during the 2008 election: something to provoke conversation among viewers, but adding little value to the company’s core mission – delivering relevant news.