The Big Problem with Big Data

• Trends
• October 17, 2014
• by Sally Ann Rivera

If you want to know which state listens to the most bluegrass, there’s an infographic for that. (Spoiler:  it’s Oregon). If you want to know which American cities most closely align with your dialect, there’s a quiz for that. It seems that our computing capabilities have surpassed our needs.

Big data has become the Internet’s favorite mind-numbing category. Big data, of course, has many practical uses, including powering “almost every successful artificial intelligence computer program in the last 20 years.” 

How do you decipher what’s useful and what’s not? Big data can too easily detect misleading correlations:

                -Use of Internet Explorer declined at the same pace as U.S. murder rates

                -Autism diagnosis and the sales of organic food increased almost identically

Big data, like any other statistical analysis, can be framed to support the researcher’s interests.

One of the scariest concerns is “the echo-chamber effect.”  This is where big data feeds into itself. For example, Google Translate works by using pairs of parallel texts, which could include Wikipedia articles. Then, Wikipedia articles in foreign languages can be written using Google Translate. See the problem here?

Has your company found a big data cure-all? Does your sales team want to present a snazzy infographic on the best markets for your product? A word of caution:  big data, like any numbers, need to be presented in context.

What’s the number mean to your audience? Tell the audience exactly what the number means. Giving your audiences something they can grasp and understand often means drawing comparisons and turning numbers into stories.

The new president at the University of North Texas announced at Dallas Regional Chamber’s 2014 Higher Education Luncheon, “Parents, send your kids to us,” Smatresk joked, “and at the end of four years, buy a Maserati.” Now that’s a number I can understand.