Friday, October 18, 2019

Two Sources of Privacy Attacks: Satellites and Interconnections

I had just finished looking over a piece from Slate, September 19th’s “You Don’t Want Facebook Involved With Your Health Care” by Kirsten Ostherr, when I discovered a related one.  That was “Are We Ready for Satellites That See Our Every Move?”, by Sarah Parcak in the October 15th New York Times.

The effect of their material has much in common.  Both are progressing rapidly, both are behind the scenes as most Americans perceive them, and both can be used for much more than these articles suggested.  As well, both have real potential to get out of control.

The first showed that seemingly unrelated websites could combine their personal data troves to provide further information on individual health care decisions.  Its first sentence, “could your Netflix viewing habits predict that you will develop inflammatory bowel disease?”, may seem silly, but there are correlations everywhere.  Some of these statistical relationships are meaningful, as, for example, the purchase of Tums and Mylanta going along with stomach discomfort.  Some describe situations which are not as they seem, such as the high rates of respiratory problems in southwestern states not meaning problems with Arizona or New Mexico but the opposite, as many people with such issues choose to move there.  Many more, probably most, correlations come from “hidden variables” which affect both factors in the same way, such as, as I have written about, the lower likelihood of female high school athletes using illegal drugs, both actually in large part from higher social class.  Some are splendidly meaningless, such as the long-lasting extremely close relationship a sociologist documented between population in the Indian state of Hyderabad and membership in the International Machinists Union.   These bogus correlations fool a lot of smart people, and an almost infinite number can be found when comparing the contents of large databases.  Accordingly, even if there is no sensible real-life connection between choice of movies watched and diseases, it may look as if there is.  That is scary.

The scope of such data excursions goes well beyond recommending unjustified medical treatments.  The article mentions the possibility, which I have long since noted, that purchases of some products could trigger verdicts of poor health practices, leading to higher insurance premiums.  Some are easy to see, such as those buying motorcycle equipment tending to have higher fatal accident rates.  Some may or may not have merit, such as people acquiring more than a certain amount of bacon or butter having an increased chance of heart problems.  Some will depend on current views about food safety, which, like the 1990s oat bran craze, may only later be shown to be erroneous.  And most if not all are susceptible to failing to identify purchases for others; a father buying his son racing car parts may come through as the one with the risky lifestyle. 

While health care and health insurance costs are real issues, there are plenty more where such data could be generated and then used both fairly and unfairly.  A short brainstorming session got me auto insurance (higher rates for people buying a lot of beer), renting decisions (that and other things associated with rowdy lifestyles), credit-related decisions (what might be deemed excessive spending on non-necessities), membership organization acceptance (opinions they don’t like), employment decisions (already being made, based on anything that could interfere too much with work), or anything else where they know even only your name and address (this time, you say what would stop them).

Parcak’s work described how satellites, which 15 years ago could clearly “see things the size of 40-inch TVs” and can now handle “those the size of smart tablets.”  A big difference as, for one reason, the new capability includes viewing license plates.  Anything not indoors, including you and I much of the time, can be observed and followed, leading to combination with other information to predict future activity, including that none of the authorities’ business.

The real long-term problem with both forms of tracking is not a lack of privacy.  It’s enforcement of conformity.  As we are seeing now with Chinese efforts giving people points for displaying what their government considers socially positive behaviors, we can be motivated with money to do the same.  From there it’s just one step to the political party in power incorporating their ideology as well.  And what will stop it?  The best outcome might be what I think now exists on a smaller scale, where police departments use information, either legally or illegally obtained, to catch criminals, but little else is actually done with it.  If that continues it would hardly be harmless – police do make mistakes – but would be the most positive we can hope for.  From there, we will as always need to do the best we can. 

No comments:

Post a Comment