Because so much has been made of Trump and Tweets, we risk missing the big story. The big story is about big data elections. 

Not so long ago, election tactics were about opinion polls, focus groups and mass media. Find out what voters think, test some new messages and then push them out the news and advertising cycle. No more. Cloud computing and analytics tell a new story. People tell you what they think without asking. The Internet is one large focus group and channel of influence.

With big data, election tactics get very personal. And at a massive level. Probably there is a line between persuasion and manipulation. I’m not sure where that line is drawn anymore.

Consider Obama in 2012. His campaign wanted to re-build the coalition that put him in office four years earlier. How to do this? Well, you start with a database of nearly 70 million Americans who voted for him in 2008. Read that again. Political campaign management and data collection have become so good that a candidate actually knows – with a high degree of confidence – who voted for him. Consequently, campaign workers are big data collectors. Because voters are individuals that can be micro-managed.

Political campaigns buy consumer data and merge databases together. Web browsing cookies are collected and tracked. You are a Facebook profile. Therefore, good data about your habits and beliefs is available. Social media is the key battleground. A candidate identifies supporters and matches them other voters who have similar affinities. Affinity matching drives e-mails, Facebook friends and even personal contacts. In short, voters are micro-targeted. Citizens are consumers who can be sold.

Big Data Elections Demonstrate Analytics Work – Sometimes

A debate is raging between big data and traditional science. Science seeks to understand causality. Big data finds correlations instead. Since causality is painfully hard to understand in the social sciences, correlations might be good enough. With masses of data and correlations, it is fairly easy to conduct thousands of experiments in persuasion and find out what works. Modern campaigns run thousands of experiments daily to fine tune their messages and manipulations.

But data science is not enough to compensate for uncertainty in social systems. Causality does not go away. Hillary just found this out the hard way.