Chris Wilson of WPA Intelligence explains why the value and importance of data varies in political campaigns.
CNET's Dan Patterson interviewed Chris Wilson, CEO of WPA Intelligence, about the importance of past voting history in political campaigns. The following is an edited transcript of the interview.
Campaign 2018: Election Hacking is a weekly series from TechRepublic sibling sites, CBS News & CNET, about the cyber-threats and vulnerabilities of the 2018 midterm election.
Dan Patterson: When you're building these communities, how do you initially scale those platforms, and how do you find the evangelists who really will shout to the rooftop, and help your communities expand and scale in a way that Facebook or Twitter or social media might have in a previous age?
Chris Wilson: You know, we're still figuring it out. There's a lot of experiment going on, and I don't know that in 2018 that I can answer that question about what the best way to do it is. I know in 2016 on the presidential campaign, we had 92,000 people download and utilize Cruz Crew app, and we did that just through the spread of word of mouth, and individual evangelists.
We tried taking those universes and those people and we also build in models to try and identify what type of person they were, and score someone as being likely to be, an app evangelist. That has had reasonable amounts of success, but nowhere near the kind of success we had in 2016. Dan to be honest with you, I don't know that that nut has been cracked yet in 2018. It may take for another presidential campaign, and the kind of enthusiasm that existed behind a candidate like Ted Cruz, or I think the UC developmental left really with Bernie Sanders. The communities were built behind these candidates to become kind of wave style candidates. It may take that type of experience again for us to understand how to do that. Our success has been reasonable, but it hasn't been anywhere like what we had in 2016.
Dan Patterson: If the social media spigot is turned off, what data sources are valuable to you now?
Chris Wilson: Well, the most valuable source is always going to be just past voting history. People are most likely to do what they've done in the past. Kind of like buying a stock. You buy a stock, you buy CVS stock because it's performed well in the past, it's most likely to keep performing well in the future. That's what it gets down to, whether or not people voted in the past. But on top of that, and you could have in some cases where it keeps being written about, for instance in Texas, that no democrat has won statewide since 1994. If I was doing a Democratic campaign there, I'd want to go find data that goes back to 1994 and figure out if there was something about that election year, or previous election years that I might be able to tap into, to help win an election in 2018.
SEE: Network security policy template (Tech Pro Research)
But outside of that, there are just the kind of things that you would expect, that people give to you every day. You fill out a warranty card, when you subscribe to a magazine, and home values. All that kind of factors in, and at this point there's really three types of data as you know. There's first-party data which is data that we build on our own. There's second party data, which is an action someone's taken, and third party data which we go purchase from another source.
Third party data, which is what I think people get most concerned about is actually the least valuable from a standpoint of what we build, and what goes into our models. What I believe is the most valuable is the first party data which is what we are continually refreshing and building ongoing basis. And that's going to be things like likelihood to turn out and vote. In 2018 that's really that important to us. Who do we talk to? Do we go to a persuadable voter, which there just aren't a lot of those left in today's polarized society, or do we go to potentially a voter who cast a vote in 2016 in the presidential campaign, but maybe isn't likely to turn out in 2018.
That's voter turnout. The second one is who they're most likely to support. If you think about it on an analytical scale, zero would be the way that we would do it on your X axis, would be from the far left side would be the Democrats, would be like a .001, all the way over to .9999. Anybody .499 and to the left would be someone who's voting for the Democrat. .5 and to the right somebody who's voting for the Republican. We kind of pull out that .4 to .6 and consider those to be swing voters, that's different and going to be by race and by the scoring. But those are going to be the types of the universes we look at. Likely to support a candidate, likelihood to turn out and vote. But we also build in models directly on issue bases too. So, whenever we utilize those for the purpose of communication in the campaign, what feeds into it is going to be different based on the issue itself. The value of the data and the importance of the data is going to vary by what sort of model we're building.
- Midterm elections 2018: How 7 states are fighting cybersecurity threats from Russia and other attackers (free PDF) (TechRepublic)
- How Florida is bolstering election security after being targeted by Russian hackers (TechRepublic)
- State of Washington has new laws and the Air National Guard to help secure 2018 midterm election (TechRepublic)
- West Virginia moves forward with first mobile voting app, despite fears from security experts (TechRepublic)
- Ohio taps college cybersecurity experts to audit election systems before 2018 midterms (TechRepublic)
- How Colorado voting became a cybersecurity leader long before Russians tried to hack it (TechRepublic)