Sunday Reflection on Big Data

Living in a digital web....

"…faceless Big Data's predictions are hard to opt out of.
"One Minnesota man learned that a few years ago. He stormed into a suburban Minneapolis Target department store, demanding to speak to the manager. 'My daughter got this in the mail,' the man explained. The manager took a look at what the customer had brought. It was a Target mailer, like millions sent out every year, addressed to the man's daughter. This one looked harmless enough, with pictures of adorable infants, baby furniture, and maternity wear.
" 'Are you trying to encourage her to get pregnant?' the man sputtered. His high-school age daughter was unmarried.
"The manager apologized and said he'd look into it. When he did, he learned that Target was using predictive analytics. It aggregated the information it had on its customers -- from visits to the website, purchases in bricks-and-mortar stores, calls to customer support, and the use of coupons or rebates. Software parsed this haystack of data in order to make specific, actionable forecasts of individual customers' future behavior.
"One secret initiative was to predict which customers were pregnant. Expectant mothers have to buy a lot of products that they may never have purchased before. With novelty comes indecision. This makes mothers-to-be receptive to advertising discounts, and anything else that might nudge them in the direction of buying at Target. A customer who comes to depend on Target while expecting a baby may decide to do her grocery shopping there -- for decades to come.
"Target's pregnancy predictions were much more accurate than random guessing, though of course not 100 percent certain. A few wrong guesses were acceptable. The awkward exception is when a customer gets really, really upset by a wrong prediction.
"A few days later, the manager called back the irate customer to apologize a second time.
" 'I had a talk with my daughter,' the customer said. 'It turns out there's been some activities in my house I haven't been completely aware of. She's due in August. I owe you an apology.'"
"Witness the new human condition. A department store's software can guess that a woman is having a child; her own father can't. Should we marvel at how clever our algorithms are, or at how bad we are at listening to and understanding our fellow beings?"

-- from "Rock Breaks Scissors" by William Poundstone

...and I now have an interview up with Mr. Poundstone over at MathTango

[…If you have a favorite math-related passage that might make a nice Sunday morning reflection here let me know (SheckyR@gmail.com). If I use one submitted by a reader, I'll cite the contributor.]