Data mining often conjures unsettling thoughts — corporations tracking every click, search and purchase. In some cases, that concern is valid.
But data mining is not inherently sinister. It often drives scientific and social progress. Epidemiologists, for example, use vast sets of health data to predict upcoming flu strains — helping create vaccines that can save thousands of lives.
In the wrong hands, that same predictive power could deny health coverage, exclude students or unfairly target consumers.
A social media manager and author, Carol Stephen, views data mining as a decidedly mixed bag. So too do Larry Mount and Yvonne C. Conway-Williams.
Precious Info Falls Into Corporate Grasp
Together, they considered the prospects — and pitfalls — of data mining: Where to draw the line, how to ensure data is used ethically — whether in forecasting disease, supporting students or securing banks — and whether companies should be allowed to use personal data without asking.
On the bright side, scientists can use data mining to predict disease outbreaks. The caveat is that they stay within ethical guidelines.
“Connecting with support groups that assist with particular afflictions can create relationships where patients and families consent to share health data,” Mount said. He specializes in data privacy, security, information management and digital transformation.
Conway-Williams takes an even more discerning and cautious approach to the data mining challenge. A board member of Women Who WordPress and ParadoxPrime Information Assurance, she has addressed cybersecurity at WordCamp.
“Scientists can use data mining to analyze patterns in public health records, travel data, social media and search engine trends to detect early signs of disease outbreaks,” she said.
Scientific Restraint
From her research, Stephen took heart that for now, privacy generally holds in laboratories.
“Scientists anonymize data with masking, generalization and other methods to protect privacy,” she said.
In contrast, commercial platforms often seem eerily perceptive — or wildly off-base.
“I’m probably fairly predictable on Netflix,” Conway-Williams said. “Other websites and their recommendations — that’s a different story.
“Despite all the time I spend on YouTube, it still hasn’t figured me out,” she said. “This is where ADHD is a superpower — I confuse the heck out of their algorithm.”
Key Decisions Depend on Knowing Good Data
Those lacking such hidden strength notice the difference.
“Netflix’s recommendations got a lot more interesting after I started watching Korean dramas,” Stephen said. “YouTube gives me something totally different. What if they shared data? That’d be terrifying.”
More important to daily life, data could be used to predict behavioral challenges in students.
“Schools might analyze attendance, grades, social media — when allowed — and disciplinary records to identify students at risk of dropping out or experiencing emotional difficulties,” Conway-Williams said.
The goal, she stressed, should be support — not surveillance.
Many of these possibilities and practices are not new.
“Lower test scores and grades can be used to head off depression or changes in students’ home environments,” Stephen said. “Teachers have already been doing this for years. Is it ethical, though?”
In the financial arena, banks might use data to enhance their security.
“Banks could mine transaction data to detect fraud, such as unusual withdrawal patterns, login attempts from odd locations or purchases that deviate from a customer’s norm,” Conway-Williams said. “They should allow customers to challenge false positives and offer transparency into decisions.”
Employ Predictive Analysis
Mount envisions an array of predictive analysis for multiple scenarios, location, spending and the like.
Stephen fantasizes about exotic fraud detection, joking, “If your spending suddenly includes three Ferraris, a call from your bank might not be such a bad thing.”
When it comes to day-to-day meteorology, Conway-Williams does not see major ethical concerns as long as everyone plays by the same rules.
“Selling weather predictions to corporations before the public could raise serious concerns about fairness and public safety,” she said.
When companies use personal data without permission, that is a different story.
“It’s the ‘without permission’ part I find most offensive,” Conway-Williams said. “If it’s not that big a deal, why not be transparent about it? Why not allow everyone to opt in rather than opt out or bypass our option either way altogether?”
Digital Marketing Does Its Best With Truth in the Data
Concerns enter another realm when law-enforcement agencies use data mining to discriminate.
“Using historical crime data can reinforce racial or socioeconomic biases because policing practices are often biased,” Conway-Williams said. “Predictive policing tools might lead to over-policing certain neighborhoods based on biased data sets.”
Critics warn that predictive policing risks codifying existing bias into future enforcement decisions.
“Discrimination is broadly unacceptable, but there are arguments for positive discrimination to redress imbalances,” Mount said. “Governance on such issues is a further complexity. It takes a lot of tolerance, empathy and patience to nurture agreement.”
Nevertheless, prospects for law enforcement are unsettling.
“Predictive policing is a scary thing,” Stephen said. “There are already such deep-seated biases in policing.”
All-Seeing Eyes
More common for typical consumers are instances where marketers use data mining to exploit a person’s weaknesses.
“We see targeting of vulnerable individuals — people with gambling addictions — with tailored ads,” Conway-Williams said. “Emotional triggers found in personal data are used to pressure buying decisions.”
Data reveals a big picture of a person’s preferences, which marketers use with targeted ads. It is not a matter of “could” but “are.”
“We’re already getting targeted ads,” Stephen said. “Sometimes data miners know more about us than our families do. Insurance companies are already using it — and not always in our favor.
“We’ve already seen many instances of data mining to exploit people’s weaknesses,” she said. “Those trying to lose weight might get ads about unhealthy — but delicious — snacks, for instance.”
Privacy Risks Grow as Thieves Get Smarter
Inevitably, it seems, a government will overstep its boundaries when using data to predict future behavior.
“It could predict who will commit a crime — ‘pre-crime’ scenarios — without due process,” Conway-Williams said. “What about monitoring individuals without probable cause or oversight? Behavioral predictions could deny services, benefits or freedoms.”
For those truly wary of surveillance, she suggested a radical approach: Create new email accounts, use unrelated profiles, and carry a burner phone instead of a personal device.
For his part, Mount operates with well-founded paranoia.
“At the outset, before ethical considerations are presented, the government oversteps,” he said. “It’s around that time they bank hard cash selling out citizen data.”
Unfit Wearables
To get away from data mining, why not take a casual stroll with a wearable device? Then again, maybe not. Insurance companies are quite interested. This is one reason Mount eschews high-tech when not needed.
“I still wear a standard watch all day long,” he said. “There is a lot of activity among friends collecting certain watch brands.”
Good Cyber Habits Stop Hackers from Really Ruining Your Life
Some advocates push for legislation requiring transparency and consent in how wearables and other personal devices feed insurers.
“The way the U.S. deals with health insurance, in general, sucks,” Conway-Williams said. “Citizens shouldn’t worry that they will be denied insurance or deal with higher rates at all, especially because of a wearable device.”
In a world where data mining shapes medicine, marketing, education and even law enforcement, the question is not whether it happens — but how, and to what end.