Rumors of a Facebook “sentiment engine” analyzing aggregate user data, or a new form of the company’s Engagement Ads that offer rapid polling to advertisers, have been flying around the web since the start of this week’s World Economic Forum in Switzerland. The reality behind these rumors seems to be much less exciting (or creepy, depending on your perspective) than many people claimed – but we’d like to entertain some thoughts on just what Facebook could do with such a system.
Remember when Google published the most popular searches being performed during the Presidential debates? That represented a sea change in real-time awareness of what people care about. A Facebook “sentiment engine” has that same kind of potential, and we’d love to see some of this data be put to use in service of such innovation.
Best Case: A Living History
Facebook is a place where 150 million people are conversing in real time, with people that they know, about the world around them. Those conversations are going on primarily in text. While most of the discussions around a Facebook sentiment engine have referenced the collection of data through active engagement in polls, there’s also a whole lot of passive discourse that could be mined in interesting ways.
This author would gladly opt-in, especially through the kind of granular privacy controls that Facebook is so good about offering, to allow the company to add what I say on Facebook to a big collection of data for analysis. Our individual identities need not be tied to our words, but the demographic data associated with our confirmed identities is invaluable.
Think of the non-commercial, public interest kind of data that could be acquired. When the economic stimulus plan of 2009 was first announced on national television – what was the reaction of people in their mid twenties who lived in the Mid West of the US? Was that collective reaction substantially different from the reaction of self-identified queer people of color living in the North East US? How did the public reaction to the proposed plan change one hour, one day or one week after the announcement? This is all very interesting and potentially valuable data that could be, for the first time in history, available in near real time. Just by listening to what people are talking about in status updates and comments.
Privacy concerns are very important, but with enough opt-in required, there could be a system set up that could funnel data straight from Facebook into the US Library of Congress or historical archives.
Facebook is one of very few companies we can think of with the access to public sentiment to make something like this happen – plus the imagination and willingness to survive backlash for a good idea.
More Likely: Polls and Product Feedback
The above scenario is a pretty far-out one, unfortunately. Far more likely is that Facebook will perform political polls and sell access to product feedback survey systems to advertisers. Those paths could be interesting as well, though they fail to solve the problem of public discussion data running unused down the drain.
We’d love to see even these approaches used seriously. Polls could be performed quickly, even as part of corporate crisis management processes, if companies have the guts and Facebook can handle the sales requirements. The age of quieting down a controversy are over – why not buy some polls on Facebook to see how real people are actually responding to bad news about your company?
Likewise, both active and passive consumer sentiment about products could lead to some very cool features. RWW reader Scott Aikin says he thinks a sentiment engine would work well in powering a consumer goods search engine. I need a pair of headphones, which ones are most liked by my friends, people in my own demographic group or people who are older and wiser? I’ll do the search on Facebook and they can have affiliate revenue when I make a purchase.
That sounds like a great idea. There are a number of great ideas that could be actualized by a Facebook “sentiment engine.” Most promising, though least likely, is a system for monitoring public reaction to historic events. We have huge concerns about Facebook’s being a proprietary set of technologies, but we’re fascinated by the company and its potential at the same time. A Facebook sentiment engine would be like a real-time, hyper detailed census – and as such Facebook should make freely available as much of the data it gathers as it can.
Will the company do anything like this? Only time will tell, but if 3 years ago we had told you that the Facebook Newsfeed would become the way that 150 million people found out when a friend’s relationship status changed, when a photo of a friend was posted online or when two people you knew were having a conversation – would you have believed it?