Big Data: Knows You Better Than You Do!
Before the boom of online engagement and Web 2.0, the natural method of data collection was through human input into records or computers. These days, any individual that participates online essentially creates their own data imprint; automatically recorded whenever they click on an ad, ‘like’ a Facebook page or enter a website. The ever-present role of the Internet in our everyday lives means data now grows by 2.5 quintillion bytes per day (Siegel, 2013, 2). The sum of this huge, growing resource can only properly be measured by what Siegal (2013, 3) describes as “machine learning”: the use of computers to quantify and process this data to develop new knowledge and capabilities, such as predictions. "Predictive Analytics" describes the method of using data mining to determine likely outcomes, which organisations use determine the likely outcome of their products and services. Examples include the prediction of mouse clicks for websites with pay per click advertisements (Siegal, 3), dating predictions for online sites such as Match.com; and the ability of companies such as Hewlett Packard (HP) to predict the chance of an employee’s 'flight risk’ (whether they will quit their job).
Our willingness to contribute data to companies like Facebook allow them to monitor personal usage and on-sell information to interested parties such as advertising agencies. Algorithms developed by Cambridge University and Microsoft allow Facebook to make predictions based on characteristics such as demoraphics, page likes, Facebook usage, location and language and purchasing activity (Cohen, 2014). This following video provides a really good simple explanation of the type of data Facebook mines and how.
Facebook advertisements and increasingly their ‘page suggestions’ (subtly weaved into News Feeds), are based on individual user characteristics. For example, both my housemates are females in their late twenties – their Facebook pages constantly include ads for weight loss products, bridal advertisements and dating websites with ‘single, faithful policemen’ in their local area: all products for which they would be classified as the target audience. I recently went kayaking and posted a status about it along with photos and ever since, I have received ads for kite surfing on the side of my home page.
Similarly, I recently changed by mobile data plan and ever since, there has been an advertisement for Vodaphone’s new data plan on the side of my page. It seems impossible Facebook could know the details of my phone contract, but perhaps the mere fact that I am on Facebook regularly means I am a good target audience for affordable data products.
Below is a snapshot of Facebook’s Data Use Policy. Whilst it is extremely simple and easy to read, it raises some questions like:
What does it actually mean that we own our data?
What legal avenues are available to enforce these rights?
Is there a way we can extract our data and move or save it elsewhere?
Policies such as this are helpful and important but often vague, and fail to give meaningful examples to users of what the actual consequences of their participation could be. Facebook has certainly faced criticised for this before.
For those that are concerned about the invasive accuracy of data mining and predictive analytics, one has to wonder whether it is that different from any sort of audience research. Television ratings and research have been used to sell audiences to advertisers for years. Classic methods of measurement include written viewing diaries and the people-meter (introduced in 1987), to track television consumption in the home (Sullivan, 2013). Both these methods rely on volunteer provision of information by participants. In many ways, data mining is not that different, except that it has extremely widespread participation and as a result, is conducted on a much grander scale. This can result in targeted, efficient advertising and information about products that consumers might actually enjoy or value.
Whilst data mining is considered by many as a form of surveillance, it is important that participants educate themselves about how their data could be used, however companies such as Facebook must also ensure honest and transparent policy disclosure.
References:
Dara Kerr. 2013. "Facebook faces criticism over its privacy policy." CNET. September 5. Accessed May 12, 2014. http://www.cnet.com/au/news/facebook-faces-criticism-over-its-privacy-policy/
David Cohen. 2014. "Facebook Audience Insights gives marketers an overview of their potential audiences." Inside Facebook. May 8. Accessed May 12, 2014. http://www.insidefacebook.com/2014/05/08/facebook-audience-insights-gives-marketers-a-better-overview-of-their-potential-audiences/
Siegel, Eric. 2013. “Introduction – The Prediction Effect.” In Predictive Analytics, 1-16. Hoboken, NJ: John Wiley and Sons Inc.
Sullivan, John L. (2013), “Chapter 4: Media ratings and target marketing”, Media Audiences: Effects, Users, Institutions and Power, Los Angeles: Sage
Woodford, Darryl, April 24, 2014. “Week 9 – Big Data and Telemetrics.” Accessed May 13, 2014. http://blackboard.qut.edu.au/
DNews. 2013. “Your Facebook Likes Reveal Everything.” YouTube video, posted March 12. Accessed May 11, 2014. https://www.youtube.com/watch?v=3SNeQXWn3xk