Social Media Data Mining and Targeted Advertising

"People aren't the customers on Facebook, they're the product being sold."  

-Jennifer Golbeck

Just now while studying for an exam, I just came across an assignment that I completed last year and realized that people might be interested in it.  I think it's interesting anyway, and alarming at the same time.  The assignment was to watch a Ted Talks about  technology and summarize it.  I chose this one:  Ted Talks - Jennifer Golbeck - The Curly Fry Conundrum and you could easily just click on that link and watch it, or read my summary of it here.  Or both.  Or neither.  That is entirely up to you.   

Summarizing a Ted Talk About Technology

Zachary Wakefield

Ted Talks speaker Jennifer Golbeck gave a very interesting and thought provoking presentation titled "The curly fry conundrum:  Why social media "likes" say more than you think".  It was filmed in 2013 at the TedxMidAtlantic conference center in Washington, District of Columbia.  Golbeck is a computer scientist, and is the director of the Human-Computer Interaction Lab at the University of Maryland.  She has been able to extrapolate enormous amounts of demographics data, behavioral patters, political tendencies, preferences etc. from social media sites.

Golbeck points out that there are 1.2 billion Facebook users per month, or about half of the entire internet community.  People post huge amounts of personal data online, and if filtered correctly the information is easily compiled into targeted demographics data that is very specific.  She makes a very interesting example of Facebook 'likes' on a page titled 'curly fries'.  As it turns out, people who 'like' the curly fries page are likely to be smart.  The reason is thanks to a few facts completely unrelated to curly fries (2013). 

First, she points out that people tend to be friends with similar people, i.e. smart people have smart friends, young people have young friends, etc.  As it happened, a smart person ('smart' as determined from a host of other criteria, of course) was one of the first people to like the curly fries page, so the news of that 'like' was broadcast to that person's friends, a few of whom clicked 'like' as well, starting a propagation through people with other similarities other than their mutual appreciation for curly fries.  Liking curly fries has nothing to do with being smart, it's just that people not in those circles of friends never saw the curly fries page pop up in their news feeds (2013).

Another example that Golbeck illustrated was of a teenage girl who started receiving specialized advertisements in the mail from Target for baby clothes, strollers, bottles, etc. well before she had prepared herself to tell her parents that she was pregnant.  Her demographics data had been compiled and algorithm kicked out her 'pregnancy score', which had determined that she was likely to be pregnant based on changes in her purchase habits.  The pregnancy score computes not just whether or not she's pregnant, but what her due date is, based on her purchase history.  Then the targeted advertising starts to swoop in.  The internet knew before her own parents, and she hadn't told anybody.  That's the power of the data extrapolation algorithms (2013). 

Golbeck raised concerns about the ease of which personal data is extrapolated, and suggested the possibility of alerting users of the type of data that could be gleaned from their 'likes' or posts before making the information public.  She admits that it will be difficult to wrangle control of personal data which is used by social media essentially as currency back into the hands of the users.  She goes on to mention the saying "…people aren't the customers on Facebook, they're the product being sold", which seems very true (2013).

While I had a small amount of understanding about what was going on, this talk was still eye opening.  I've seen it in action on my own Facebook feed; minutes after I did an Amazon.com search for geodes, Facebook tried to sell me geodes right in my feed.  While I appreciate a solid capitalist marketing scheme, it still frightens me a little knowing that people are so vulnerable to this method of target advertising, as if we don’t have money flying out of our pockets already.  

References

Golbeck, J. (2013). The curly fry conundrum: Why social media "likes" say more than you think.   Ted Talks. Retrieved from http://www.ted.com/talks/jennifer_golbeck_the_curly_fry_conundrum_why_social_media_likes_say_more_than_you_might_think#t-150455