BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Did IBM Just Create The Killer App For The 2016 Presidential Election?

This article is more than 8 years old.

In 2012, a team of 100 staffers had been tracking millions of people since 2008 in an effort to influence their behavior on one specific day.  The team quickly realized that there were three levers to ultimately achieve victory:  the number of registrants, the number of people they had to persuade to act, and the turnout on that special day.  Yet one factor that wasn't available before was more important to the success of this campaign than all three levers combined:

The Obama campaign's focus on data.

When campaign staffers described their 2012 victory to me it was clear that data analytics was the primary reason that President Obama won the election. And, if you read “Inside the secret world of quants and data crunchers who helped Obama win” by Time’s Michael Scherer, you know that Obama campaign’s Chief Scientist led an analytics department five times as large as that of the 2008 campaign. That was a lot of rich analysis and consequently a significant amount of political firepower.

But that was two and a half years ago, so what does 2016 have in store for candidates? Well if IBM and its partners have anything to do with it, they’ll be creating 100 times the firepower and compared to 2012, doing so in a fraction of the time.

Here’s what they’re up to:

Why the IBM Tail Can Wag the Election Dog

According to a report released last month by Borrell Associates, “Total online political ad spending is projected to soar to almost a billion dollars in 2016. U.S. political races are expected to generate more than $12 billion for all contests — nearly $51 for every qualified voter.”

As you can imagine, social and other digital media promotions are expected to be a substantial and growing percentage of those numbers. Faced with this daunting situation, IBM has partnered with MutualMind to create a new approach known as ‘Adaptive Listening’ (AL) to help candidates automatically capture and act on election signals.

Inspired by the popular If-This-Then-That (IFTTT) service, this rule-based automation for social data analysis and action will help manage the mountains of social conversations being collected, parsed, analyzed and reported on. These conversations can be programmatically segmented into dozens of election issue categories across topics such as immigration reform, gay rights, environmental legislation, consumer protection, taxes, etc.

To understand how Adaptive Listening works, consider this example on Twitter:

Michael Smith is using social media to express his opinions about ‘immigration reform’. He’s concerned that taxpayer dollars will go into the housing and care for illegal immigrants pouring across the border between Texas and Mexico but not completely sure about where each of the announced Presidential candidates stands the issue. He is leaning toward Hillary Clinton, posting his thoughts and responding to replies via Twitter but he’s not opposed to a Republican candidate.  According to his Twitter profile, Michael lives in Texas and has over 100,000 followers.

Here’s his Tweet:

The AL enhanced Engagement Center (pictured below) powered by MutualMind and runs entirely on IBM Softlayer, will automatically pick up multiple types of matches here.  First, it picks up the keyword matches for #Hillary2016, Cruz and Rubio.  And from the post, its machine analyzed as negative (frustrated).  That’s not bad but it’s pretty much the table-stakes for most first-gen social monitoring tools.  Here’s how the Engagement Center goes further:

First, you can create an Adaptive Listening rule with multiple dimensions;

  • if keyword is #Hillary2016 AND content contains “Immigration” AND sentiment is “negative” THEN tag with “Immigration Reform” AND add to list “Immigration Reform”.

Since the rule includes reference to ‘Immigration’ this content will be automatically tagged and added to a list called ‘Immigration Reform’.

You can then take it further by creating another custom rule such as:

  • if keyword is #Hillary2016 AND content matches "Swing Voter" AND location in “Texas” THEN add to list "Swing Vote" AND tag with “Swing Voters” AND tag with “Texas Voters”.

Next, matching up a phrase contained in Michael’s Tweet such as-

  • ‘leaning’ in combination with the first keyword matches, adds this person to another list called ‘Swing Voters’!  

Finally, since Michael is on Twitter, we’re picking up his profile where he indicates ‘Texas’ as home. Now we are able to geotag this Swing Voter prospect for ‘Texas Voters’ and since his Twitter Profile also indicates he has 100,000 followers, he would certainly be qualified as a social media influencer.

Now imagine this process, at scale, without any human intervention – candidates would have a huge list, broken down by topic and sentiment. From here, campaign managers can easily target each of them with political advertising (like Twitter cards) to obtain their email address and to explain their immigration stance specifically for an influential Swing Voter in Texas.

In the last election, this process would have taken weeks with 25 staffers working full time to obtain the same information. And this is just one use case. Can you imagine all of the automated recipes that will be cooked up by politicians in the future?

But that’s only the beginning of what’s on the menu…


IBM’s Watson + Adaptive Listening = Killer Election App

MutualMind is also working with IBM’s Watson group to further enrich Adaptive Listening results for campaign analysts.  The company’s new Social Portraits feature dives deeper into voters’ personalities and behaviors in order to better understand their preferences.

In our last example, Watson will use its algorithms and artificial intelligence, to create a personality profile for Michael. Let’s for example, say his ‘openness to change’ score is 87 out of a possible 100.  If Ted Cruz’s campaign team were using the Watson enhanced AL solution, they could identify Michael as a high-value, influential swing voter in Texas and might be able to convert him from a Hillary Clinton supporter.

But the real power of Watson as I wrote in 2013, is using Watson as a scalable service - where they can observe large-scale shifts in perceptions and opinions as they are happening, and target or frame their message to speak to those populations.

For example, a political candidate or party could train Watson to understand its political base, then use predictive models to recognize undecided voters or potential voters that haven’t voted in previous elections.

But let’s get specific and apply the power of Watson to an election. Watson could proactively and intelligently test, measure and optimize digital content, ads, website pages even a series of messaging to a small group to efficiently maximize voter response. Once the most effective content was determined, Watson and the AL solution could be used to sway undecided voters and previous non-voters through an optimized content campaign.

In 2012, the Obama team took weeks to measure the impact of one message by testing up to 250 variations. Watson could do this in hours and with multiple messages. In other words, Watson the Political Analyst could do what Nate Silver does plus what 100 Obama re-election staffers did in 2012 – in a fraction of the time. But it does take months, possibly years to train Watson to be politically effective. Yet if a candidate or party started now, Watson combined with AL could be the deciding factor in 2016.


In Sum

When one considers how quickly people’s attitudes can change based on political activism on social media (e.g. marriage equality, confederate flags, welfare challenges, Obamacare, etc.), one must also consider that a system will be developed to make it more efficient. IBM and MutualMind have developed the beginning of that solution. And when Watson is taught politics, their solution can conceivably be used on auto-pilot and at scale to sway opinion during times of political uncertainty and during an election.

So the big question is, which political candidate(s) or party is going to take advantage of these next generation solutions?