What the Internet Is Doing with Your Data, Part II
As we mentioned last month. information about all of us is being collected and resold constantly by some of the largest corporations in the world. Our preferences, likes, and interests are being logged and examined for statistical correlations.
This isn’t entirely new. In the past, some patterns were predictable, too. If you lived in Chicago, you probably shopped mostly at stores in Chicago. If you bought hamburger meat, you might go to the ketchup aisle next. But now we have much more sophisticated tracking.
(By the way, not all internet companies gather and sell your data. Cruzio does not.)
New Tools are More Powerful Than Ever
Powerful data-crunching computers and artificial intelligence (AI) make it possible to store and analyze a lot more data and to make a lot more connections over a broad swath of the population. Social media and phone apps supply these powerhouses with the data they need. (What data do they collect? See last month’s newsletter.)
Market research companies use something called Predictive Analytics. One firm explains:
“Predictive analytics is the use of data, statistical algorithms and machine learning techniques to identify the likelihood of future outcomes based on historical data. The goal is to go beyond knowing what has happened to providing a best assessment of what will happen in the future.”
They go on:
“An analyst performs a regression analysis to spot strength of correlations between specific customer variables with the purchase of a particular product; they can then use the “regression coefficients” (i.e. the degree to which each variable affects the purchase behavior) and create a score for likelihood of future purchases.”
In other words, if we know what you did before, we can make a good guess at what you’ll do next.
Can Predictive Analysis be a Good Thing?
We’ve mentioned before that there’s a positive aspect to having your data analyzed and your behavior predicted. Companies can basically lay a smooth path before you, so that you are as comfortable in the vast online world as you are among friends.
But there are obvious drawbacks, too: giving people more and more of what they seem to like exaggerates the differences among us, putting us into silos of our own personalities. If you are elderly, you won’t see ads for baby food. Young people won’t get ads for arthritis. Politically, siloing has the effect of polarizing us as each person sees only what they already believe.
Some examples of predictive analytics are pretty impressive. For example, a Target customer was shown ads for products of interest to pregnant women before she knew she was pregnant. That’s amazing. The store was trying to be helpful, but that level of help might be unwanted, or even dangerous.
Some Funny Correlations
Other “insights” are bizarre. Scientific American captured a number of them. A few:
- Typing with proper capitalization indicates creditworthiness
- Women with large bra size spend more money
- Orange cars require less repair
- Vegetarians make their planes on time more often
- In general, 54 percent of people would rather have loud farts than stinky farts. But among those who are interested in calligraphy, 71 percent would rather have loud farts than stinky farts
Predictive analysis is often wrong. Even at Cruzio, we’ve experienced examples: One of our programmers has been mis-identified as a dentist and can’t escape the flood of ads for dental seminars. Another, who’s white, has been categorized as African-American. These make us chuckle; we seem to be hidden in plain sight, somehow winning the game.
But as data collection and analysis matures, it’s likely mistakes will diminish and more and more will be known about all of us. The small bargains that we accept every day — I’m alright letting my weather app know my location, I’m alright letting Facebook look at my email contact list — add up as data-mining companies purchase information from the vast number of apps, merge the data, package it and sell it. So our “identities” will follow us everywhere.
