Democracy and Big Data

utah data center

NSA Data Center — Bluffdale, Utah

In a recent post to this blog, I outlined how the debate regarding the National Security Administration’s data gathering activities pitted privacy against national security and sought to “balance” the two competing values.  I suggested that framing the debate in these terms misses the more important concern that the NSA’s data gathering activities are a significant threat to democracy.   In what follows, I will explain my concerns.

Although most reporters suggested that Edward Snowden was primarily concerned about the invasion of privacy when he revealed the NSA’s data gathering activities, Snowden himself made it clear that his primary concern was for democracy itself.  In an interview about the reasons for his actions, Snowden worried that through his work for the NSA, he was “extend[ing] the capabilities of … [an] architecture of oppression” and that the government unilaterally was “grant[ing] itself powers to create greater control over American society and global society.”   Snowden was calling on us to see the dangers of the NSA’s surveillance programs more broadly.  These programs do not simply pose harms to individuals, they have the potential to transform the character of all political life in the country.

But what is this “architecture of oppression” that Snowden mentions, and how will it “create greater control over American society?”  The answers lie in understanding the significance of collecting and accessing Big Data which is really the core of the NSA surveillance activities.

Far from merely poking into the privacy of individuals, Big Data potentially provides its owners with the ability to modify the behavior of individuals and entire demographic groups.  The most obvious example of this is the data collected by internet companies like facebook and Google.  By collecting information about a person via their voluntarily constructed on line profile or through recording their search behavior, facebook, Google, and other such companies are able to craft advertising messages that are increasingly able to direct our behavior on and off line.   To be sure, the algorithms used to customize advertising and search results are not perfect, but one need not succeed in every instance to increase the odds that members of a market segment will be persuaded to make a purchase or view a website.  Such is Big Data’s role in commerce – a role that is not especially worrying.

We should be concerned more, however, with the political use of Big Data.  In the past, political strategists employed data collected by Boards of Elections.  One’s voter registration record usually contained one’s name, address, date of birth, political party affiliation (if any), and the elections in which one voted.  From this, campaigns tried to identify likely and unlikely voters as well as sympathetic and unsympathetic voters.  Door-to-door campaigns could then be run more effectively.  Furthermore, the campaign message could be tailored to specific groups to maximize voter turnout in favor of the candidate and suppress turnout for the opposing candidates.

Now, with the availability of Big Data, a campaign can understand the voting population much better. This data often is available freely on government websites, e.g., the US Census Bureau and the Federal Election Commission.  These sites can inform a campaign about the socio-economic status of a precinct, the breakdown of renters versus home owners, an individual voter’s history of campaign contributions, and much more.  Conceivably, other Big Data repositories could be made available from the private sector.  Knowing which voters purchased SUVs, have health insurance, shop at discount stores, take advantage of “back to school” sales, subscribe to specific magazines, purchased home security systems, or visit certain websites can help identify individuals with specific interests that then could be exploited by the campaign.  The candidate who has the most extensive access to these data sources and can hire the data analysts capable of mining the data will have an enormous advantage over candidates who do not.

This style of campaigning is not merely a prospect for the future.  During the 2012 presidential campaign, the Obama reelection committee employed Big Data (or at very least a lot of data and very sophisticated data analysis) to contact voters with messages that brought them to the polls in numbers far greater than anyone expected.  According to Jonathan Alter the analyses were sophisticated enough to tell the campaign “why placing ads on reruns of The Andy Griffith Show made sense on some cable systems but not on others.”  Furthermore, data was collected to test campaign messages and to measure the persuasiveness of particular door-to-door volunteers.  The data analysis used by the Obama campaign, however, mostly focused on creating a nation-wide database that linked likely voters, volunteers, and donors in order to make donors of volunteers and volunteers of donors.  So in this sense, it was not as sophisticated as it could have been.  Still, it was really only the first concerted attempt to run a Big Data campaign.  It often has been credited for winning the election, and so it likely will become the model for future political campaigns that will make greater and greater use of data analysis.  (For an illuminating account of the Obama campaign’s use of data in 2012, see Jonathan Alter’s recent book The Center Holds.)

But where is the danger to democracy in this?  After all, it is still the voters who are deciding the outcomes.   Well, the danger arises long before the voters have anything to say about the election.

As campaign data analysis becomes more sophisticated, voters will only be presented with candidates who have access to the largest data sets about the voting population and who have the resources to analyze these sets.  All others will be screened out of the electoral process long before any serious campaigning begins.  For a campaign to be successful, it will need to have supporters who own important data sets and can provide the technical expertise to exploit them.  Such friends cannot come from the working or underprivileged classes.  Obama’s digital campaign had a budget of over $25 million dollars and costs for future campaigns surely will be higher.  Consequently, the only entities capable of amassing the financial and digital resources will be extremely rich individuals, major corporations, internet companies, and broad industry groups.  The ability to affect an election will not be based on the democratic principle of one person – one vote.  It will be proportional to the donor’s wealth.  Even more so than today, these groups will have effective veto power over who will be a “viable” candidate for state and federal office.  If the Supreme Court’s decision in Citizens United advanced the cause of plutocracy, then the private ownership of Big Data and its use in elections will ensure that plutocrats will be unchallenged in perpetuity.

George Orwell’s 1984 warned that video surveillance might ensure that a political party would one day establish unassailable control over a society.  He wrote, “If you want a picture of the future, imagine a boot stamping on a human face – forever.  There is no way the Party can be overthrown.  The rule of the Party is forever.  Make that the starting point of your thoughts.”  Today’s surveillance technology is not just Orwell’s simple video cameras.  It is also the ubiquitous data and metadata harvesting by public and private entities.  The NSA is merely one institution that is amassing this data, though it is doing so on an unimaginable scale and with an enormous budget.  It currently is constructing a data center in Bluffdale, Utah, containing four 25,000-square-foot halls, filled with servers that will be able to handle yottabytes of information.  (A yottabyte is equally to approximately 500 quintillion or 500 x 1018 pages of text.)  Meanwhile, the NSA has only the slightest democratic oversight and ominously, it is working in support of a bloated National Security State that defends a plutocratic government.  One might be tempted to call it an “architecture of oppression.”

I suspect (hope) that Orwell’s image of the future as a boot stamping on a human face is too extreme, at least for US domestic politics.  More likely, if you want a picture of the future, it will not be much different from the present, but it will be less corrigible.  We will see a wide disparity of wealth with a large, struggling underclass that is alienated from the benefits of economic progress.  These condidtions will be guaranteed by governments that first of all serve the owners and managers of society.  The pretense of democracy will survive only in the carefully manipulated elections contested by competing elements within the ruling class, and one of their most important tools for social control will be Big Data.