Tuesday, June 18, 2013

No Place to Hide

Throughout history farmers and scientists have crossbred crops to improve them—to make them more durable, healthier, more abundant. Few people would think this is a bad thing.

When scientists began doing the same thing by manipulating the genetics of plants at the level of DNA, however, a lot of people did think it was a bad thing. Europe in particular has been very aggressive in regulating or, in some cases, outright banning genetically modified (GM) foods. That’s just one of the topics involving trade being argued among G10 leaders, as I type this, a mere two and a half hours up the road from me.

The controversy over GM foods is derived from a question: at what point do we become so efficient at doing something natural that it becomes unnatural—and in fact deleterious?

While I think it’s always good to be cautious with new technology, at the same time I don’t think we should react like superstitious Luddites to new technology either.

I’ve been thinking about this quite a bit lately, but not because of GM foods. I’ve been thinking of data mining. And, no, not for the reason you might expect.

Data mining is the process of gathering large amounts of information on computers and then analyzing it to find patterns and connections that otherwise are not immediately obvious. It is a term that is getting bandied about these days because of all the recent news stories about the National Security Agency’s ongoing collection of telephone and internet data inside and outside the United States.

The fact that the government is collecting all this data scares a lot of people. Personally, it doesn’t worry me a whole lot. As my father used to say, if you’re not doing anything wrong, then you don’t need to worry about what people might find out about you.

But there are more compelling reasons that the NSA’s intelligence collection doesn’t worry me too much. One is that I tend not to worry about things I cannot do anything about. Another is that, mathematically, my data is figuratively a needle in a haystack and the government seems determined to build the biggest haystack possible. As long as I don’t plan to start a Tea Party group, I don’t have much confidence the government can even find me. (In fact, one wonders whether, if the government had not been so determined to build such a big haystack, it might have had better luck picking up on the Boston bombers.)

No, what got me to fretting about data mining is an article in the January/February issue of the MIT Technology Review Magazine.

For years political consultants have used marketing tools to win political campaigns through market research, advertising and targeted campaigns. But what the Technology Review article describes is as far removed from traditional marketing as GM food is from winter wheat.

As described in the article, President Obama’s masterful and hugely successful reelection campaign took micro-targeting of voters to a whole new level. If you’re worried about the NSA knowing too much about you, it’s nothing next to what the Obama campaign knows about you. The campaign had such good and extensive data that it was nearly able to individualize campaign volunteers’ talking points for each phone call they made, each email they sent and each house they visited. They managed to get out a lot of people who otherwise would not have voted and to get people to vote differently than they otherwise would have. And where this really paid off was with women.

Think about it. In an election where everyone agreed that the overwhelmingly most important issue was the economy and in which neither political party was proposing any major change to abortion laws—and certainly nothing that would affect access to birth control—the Obama campaign managed to convince large numbers of women who would have normally voted Republican not to vote or to vote for Obama. The only concrete “women’s” issue raised during the campaign was whether church-run businesses had to pay for birth control under Obamacare, and that was settled well in advance of the election (by a fudge in which insurance companies, and not the employers, “paid” for birth control). Yet multitudes of women were somehow convinced that their rights were at stake in the election, and at the hands of a former governor of Massachusetts.

Looking at it objectively, one could fear that any candidate with a good enough data mining consortium (Obama used the Washington-based Analyst Institute) could not be stopped from manipulating his way to election. But this distortion of the electoral process will be self-correcting. In 2012 Mitt Romney’s data science unit was one-tenth the size of Obama’s. You can be sure that will not be true of 2016’s Republican nominee.

Unfortunately, while these new techniques have now been shown to work with incredible efficacy, there is no sign at all that they contribute in any way to making the successful candidate, once in office, a better leader or president.

No comments:

Post a Comment