Glad you found your way here. WELCOME!!
I’m Ann Shi, a bad programmer, a travel lover and a real foodie who now lives in New York City. Avocado is my love. Making two avocado toasts is always the last step of my morning routing, then it’s time to go and start off the day!
This blog is all about digital marketing. If you are interested in digital marketing or interested in me, do subscribe my newsletter. Email me, I’ll reply if you also love avocado 🙂
There is a classic example of how powerful big data is, and you may have seen it over the years online. Target kept sending pregnant magazines to an unmarried high school girl based on her purchasing patterns like unscented lotion, mineral supplements, and cotton balls. Her father was angry and sued Target for defamation, but later he was told by his daughter that she was indeed pregnant.
As crazy as it may seem, the case is absolutely true. Using big data and predictive analytics can tell a frightening story.
When we shop online, we are sharing our highly detailed data through our user IDs, which tied to our credit cards, emails, and address. When we are on social media, we are sharing our locations, networks and personal interests.
Big data is everywhere, and it seems that big data has no weakness. But is big data truly almighty? Does it have limitations?
In my opinion, yes, it does! The first limitation of big data is the absolute accuracy. It is impossible to drive insights that are nearly 100% accurate.
Take my experience as an example. Smart thermostat uses data to predict your schedule and control the heat of your house accordingly. But there were several times that my house was colder than the outside when I came home because my thermostat did a poor job of predicting when I’d return.
Even though this kind of mistake happens only a few times a month, it makes me feel pretty bad. People online also say it tries to be smart but uses dumb data.
The problems become even more serious when they extend to something like driverless cars, which rely on big data. We’ve already seen some accidents of Tesla car mangled in fatal crash was on autopilot.
No one wants our cars or our daily applicants go wrong because the data analytics were only 90 or 95 or 99 percent accurate. And they may result in serious accidents.
The second limitation is the difficulty of big data answering “why”.
Years ago, researchers in google developed the GFT system (Google Flu Trends) to “nowcast” the flu based on people’s related searches online. It could produce estimates two weeks earlier than the CDC’s data, which may save more lives!
But it failed in missing at the peak of the 2013 flu season by 140 percent.
The main reason of the failure is that Google researchers didn’t know the exact correlation between searching keywords and the flu trends. They didn’t even want to figure out why.
In March 2014, researches from Harvard University published a paper, The Parable of Google Flu: Traps in Big Data Analysis, in Nature journals. They said the Google flu shows a wrong way of thinking, which is, significant correlations of big data can be the replacement of all the cause and effect in our world.
There is also an example of myself. Big data will show what I bought on internet or at a supermarket and link them with my demographic profile. But it doesn’t tell whether I would have chosen another product if it were available. Nor does it tell why I chose that particular product.
Big data provides information on our behavior but it does not explain. This is why big data is not enough. Big data is easy, but it won’t replace the brainwork behind careful in-depth statistical analysis.
Big data also faces limitations because of privacy concerns. Public privacy are the bottom line of big data. Companies that collect data are tasked with the big responsibility of protecting that data.
Last month, The Facebook’s data breach enraged millions of people. No matter whether it blamed on Facebook or not, the crisis brought very bad effect to the company and caused panic among the public.
Except for Facebook, Uber’s 2016 data breach effected 5.7 millions users with their user name, mobile phone number and their email address. In 2014, ebay reported a cyber attack and said exposed names, addresses, dates of birth and encrypted passwords of all of its 145 million users!
I made a infographic about the biggest data breaches of the 21st century. (Click the infographic)
Let’s turn back to the example of the Target’s pregnant girl. From the point of view of users, obviously Target was not thinking of the privacy of that girl.
Data is unlimited, but there are limitations of using big data, such as inaccuracy, misinterpretation, and privacy. Big data is a great thing, but it’s not a panacea. It’s better to use big data cautiously, intelligently and legally, and avoid bringing results that cause serious problems for users.
Thank you for reading and have a good day!
Ann Shi
Ann Shi 