Navigating the Data Swamp: A Political Insider Explains Voter Privacy and Campaign Vulnerabilities

Allen Pace: Picture two people that live on the same street, looking at the same election, holding the same ballot. One of them is being shown and fed messages engineered to press one set of buttons, when the other one is being shown and fed messages engineered to press the opposite set of buttons. Neither of them knows that the other version exists, and neither knows the message was built off of a record that started the day that they registered to vote. In last episode, I followed the record from the county clerk to the data broker, and I showed you how a public registration becomes a scored and resold profile of your life. the 30-second version if you missed it. You registered in nearly every state that creates a public record that anyone can buy. An industry buys all of it, stamps you with a permanent ID, bolts on your money. your purchases and your predicted beliefs then sells the result to campaigns, parties, nonprofits, and most recently the federal government that buy their way around the rules through a data broker. You only agreed to vote, and the rest you were never even told that it was happening with your data behind the scenes. That was part one. And it's the foundation for today's episode. Today we go inside the campaign with somebody who sat in those rooms. How the file actually gets built, how it gets aimed at you, and how it gets leaked. One ground rule, and it counts double in the conversation like this. This is not a story of red versus blue or right versus left. Every side runs this machine. And the quickest way to get played is to convince yourself that the other team is the only one that does it. We are here to talk about your data, not your team, and not your politics. Today we're talking about what happens to your voter record once a campaign gets it in its hands and what it does with it. I am Alan Pace. This is the Opset Podcast. My guest today is Matthew Durden. He has spent over 20 years working elections at the state and national level as a public policy executive consulting with governments and NGOs. He has spent his career inside the political world that feeds this data swamp, which means he can answer the parts I can only gesture at from the outside. Matthew, welcome to the Hopside Podcast.

Matthew Durden: Alan, thank you very much. Really appreciate the opportunity to to be part of this.

Allen Pace: Yeah man, great to have you on. So let's start with how the brokers enrich these data packages and let's go ahead and start with the state file, which in our previous episode we called the spine. When a broker gets it, what is the first thing that gets added?

Matthew Durden: Like you described, that state file is the spine. It is the original data record. This is all the information that electors have provided to the state about their personal information. It's their home address, full name, their license number, oftentimes their phone number, email address. That file gets transferred to a broker. And The match begins. the broker starts finding the information about what phone numbers, especially cell phone numbers, and email addresses for each of those consumers, and they add it to the file. Then comes demographics. So age bands, household composition or modeled ethnicity gets added to the file as well. So all of these presupposed determinants are the demographics that are being added to the file. And then on top of that becomes the consumer data. So this is all the credit card transactions, online transactions that take place that helps enrich that data and creates the voter into a profile or a consumer.

Allen Pace: so how much of that first layer is bought and how much of it is scraped or inferred from those consumer based sources?

Matthew Durden: Most of this information, this is the most expensive transaction. Once data brokers get a hold of that data file and they start to add scores, ethnicity, and this assumed demographic data, that's the most expensive part. And so campaigns or interested parties are going to pay the premium for this enriched file.

Allen Pace: So the first file, like we covered in the previous episode, as you're aware of, is the government layer file, which is essentially the voter records unenriched, correct?

Matthew Durden: That's correct.

Allen Pace: How does a data broker know that, for example, Mr. Matt Smith in the voter file, is the same Matt Smith in a credit bureau file? what are the matching keys there?

Matthew Durden: At this point it becomes a probability question. So if the Matt Smith has registered to vote at a particular address at a location, there is longitudinal data that puts Matt Smith at that address. And so any kind of purchases or credit card purchases, online purchases that can match back to that address, it's just probabilistic and it it is assumed that it's the same Matt Smith. That's not a hundred percent guarantee that the data is correlated to that consumer, but it is a high probability. So much of this information is based on probability scores. So even ethnicity or religion, when those are tied to the voter file, it's not an exact match. It's based off of kind of assumed principles within the survey or polling world where we're making assumptions based on where the individual lives. what their voting patterns look like, what their purchasing looks like, and so that becomes this new model for how the file is enriched.

Allen Pace: So what I'm hearing is that the enriched file, as it's being enriched, the voter record includes your name and your home address, it can easily infer that a purchasing record or financial records that match that same thing obviously make it a pretty easy assumption that it's the same person. on the data side, what happens when that match is incorrect? And does anybody downstream in this chain ever find out or care or does it matter?

Matthew Durden: There's no real way for a, the consumer, to actually see what their voter file looks like or what their profile looks like. there's no recourse that you would be able to address, you know, your credit score or your FICO score. If you find an error or something, there's an opportunity for you to address the inconsistencies or the incorrectness and challenge those. When it comes to the voter file, there is no avenue for that. So those assumptions will stick with you. the only way that you would be able to challenge something is perhaps during a survey or poll where you're talking to a campaign worker and maybe they let you know a piece of the information that's in your file and you have an opportunity to correct it. But beyond that, those scores, the files, the demographics that have been assigned to you are now kind of a living piece or a living part of that data profile that sticks with you As long as you continue to register, update your information, your license, et cetera.

Allen Pace: one question that just popped into my head as a follow-up. are voters or citizens allowed to demand that their data be removed when somebody knocks on their door?

Matthew Durden: No, I don't think that there's any state or federal right that would guarantee somebody a voter that if somebody were at their door and some information was shared that you'd even be allowed to even look at it, let alone challenge some of the information that's in it.

Allen Pace: Yeah, that's interesting because my mind goes to obviously the credit bureaus and the data harvesting services where you can request data removal, but this seems to not even be allowed in this system. it's pretty mind-boggling that there's not even a ability to view your own data. That's pretty shocking. So speaking of that data. We threw around up to 1800 fields in the last episode that are data points on each individual user or citizen. can you make five of them real for the listeners here and what actually is on somebody's record?

Matthew Durden: Sure. one such we'll call scores and typically a score is a number between zero and a hundred that is assumed. And so if you have a partisanship score that's super high, that would mean that you are likely to participate in in electoral activities. You're likely to take action. There's a high probability that you're going to participate in a primary election and a general election, whether it be through an absentee ballot or voting in person. And that can be determined by just looking at your vote history. And so if you do participate in primary elections and general elections, you're seen as like a habitual voter. You show up to the ballot whenever there's something on the ballot. That Gives you a high score when it comes to your participation in politics. A campaign would then use that as seeing you as a high probability target, as a persuadable universe for their electoral profile. Another example would be your religiosity or what religion you belong to. this is also just assumed data. Often in the United States you'll have just a general Christian moniker for most folks. But as the consumer data goes into the file and starts to enrich that data, it gets much more granular and much more specific a likelihood of which church you attend based on your distance traveled to and from church. because that kind of data that exists both on phones or on email servers, text messages, et cetera, that can be gathered and added to your campaign data. and it's easy then to determine perhaps what church you even attend.

Allen Pace: I see. as I know that you've been a long time listener to this show and as everybody else knows, I harp heavily on information related to money and that being one of the first things that you should secure. we talked about that data file being enriched with household income data and even more troubling, home value, which to me is one of the most invasive things that a random stranger can have about you, I don't understand how that is justified to be even in the file. can you expand on that a little bit and tell me are those two fields the ones, well at least the household income is that the one that is most accurate most of the time or least accurate most of the time? And how is it used? And what's the justification of it being there?

Matthew Durden: The most accurate information when it comes to income or property ownership is going to be whether or not you own or rent your property. And that's because inmost counties, all of the property information is public. You can go online and you can look parcel by parcel to determine what the price is, who the owner is. So that scraped data can go into the file and you have automatically An understanding of if that person's a renter or owner and what their home value's worth. the information about your income is going to be a little less accurate. But of course, every time you apply for a credit card, you have to tell them what your income is your banking information has all of your income. all of that data then goes into the data file as well. And while it's assumed, you, as a consumer, have shared that information often enough to make the household income as accurate as you're willing to make it by providing that data out there. Another such example is also whether or not you have children in the home. And while you know we don't register kids until they're 18. Parents who are of a particular age in a particular school district who answer questions online or shop for particular products online, we can assume that then there's children in the home based on some of their commercial activity as well.

Allen Pace: Well, the next question I had was about scores, but you kind of already answered that. So let's go ahead and move on to how the campaigns enrich this data. I'm an outsider who has very little knowledge of any type of politics or campaign structure and behaviors other than what I see on the news, which is Beyond the scope of this conversation. When a campaign touches a voter file, what gets written back to the file? Specifically like support ratings, what's the deal with these issue IDs, that kind of stuff, refused information, or is it like how they behave when you knock on their door? What what kind of information gets added to these things?

Matthew Durden: A campaign is number one is going to be looking for voters who support their candidates. So they're going to with this voter file, they'll have a phone number, email, likely a cell phone number, and they're going to reach out to voters and they're going to ask them a series of questions. hi, as Matt Smith available. with that, they have Matt's age, and much of his demographics, Based on his score system, somebody has determined that he's a high probability to be a persuadable individual for this campaign. Once you start interviewing the individual about some of the particular concerns or interests that they have in campaigns, you can add that to the voter file. Then if they show interest in the candidate that you're working with, you apply a score and then that voter is either considered a likely voter in favor of your candidate, opposed to your candidate, or you know, neutral, somebody who could be persuaded. and each of those data points means that that voter is gonna get pushed into a new stream of communication, be it we need to continue with the persuasive campaign work. So that individual will now get targeted for social media or actual mail or additional phone calls. Where we're gonna make the persuasive call and an argument to that voter. and so every time we communicate, we're able to validate: is the phone number, is that information correct, name correct, are all of the demographics that are on our list correct once we reach that individual? So every time we're canvassing the voters. we're affirming or disproving the data in that file and so we're able to refine it with every touch.

Allen Pace: So you mentioned canvassing. Tell me about free text notes and what does a canvasser actually type about somebody after they leave and the door closes. you know, is it their behavior, maybe more information about their demographic, elderly lives alone, aggressive dog? person is not approachable because they're angry somebody knocked on their door in the first place. What what kind of stuff goes in there and what impact does that type of information have?

Matthew Durden: If I was canvassing, I would have included or outlined or collected all the data that you outlined. I would confirm, first of all, all the assumed data. So is the individual actually the age, name, does their partner or spouse still live there? Are there any other voters in the house? If that individual has, you know, since deceased or moved, those are options where you can continue to track data once I've recorded that back into either my phone, tablet, whatever electronic device I'm using to canvas and survey voters, that will now take that either deceased or moved voter and place them into a different category for that data to get scrubbed and find Where have they moved to so that we could re add them to our universe?

Allen Pace: Can a person request to be added to the do not knock list just because they want to maintain privacy?

Matthew Durden: They can, but I guess it would then be up to the discretion of the canvaser to add them or mark the do not knock out of courtesy.

Allen Pace: does any of that information stay with the campaign or does it travel to the permanent file and get added to other campaigns in the future when they collect these files?

Matthew Durden: Yeah, all of the above. So that information would stay with the campaign for the duration of that campaign because it would be valuable. But part of the agreement that the campaign has with that data broker is that the data that comes out and the data that goes in also belongs to the parent company. So when you're enriching that data from a campaign side, you are enriching it for future campaigns and for that broker. and however they're going to use that file in the future as well.

Allen Pace: Okay, so that answers my follow-on question, which is who inherits the data? So it's the data broker that whatever campaign is working with,

Matthew Durden: Yeah, absolutely. their parent file, their megaphile basically is the primary file that's just continually being enriched over and over by campaigns, by data brokers, and by consumers themselves.

Allen Pace: I assume there's no retention policy on any of this information that's added in the field. has anybody ever worked to get some of this stuff deleted or get a retention policy added to this type of data, or is that just not ever even been talked about? Because as far as I know, the general public does not know that this is even happening.

Matthew Durden: I don't know of any campaign or campaign staff or anybody who would be interested in trying to limit their access to this data because it's such valuable information

Allen Pace: Right. let's go ahead and move on to what gets sold and who decides. the universes and you touched on that briefly a few minutes ago. Can you define in plain language for the listeners, for somebody who's never heard of this product, like myself, And tell me who actually decides what goes into it.

Matthew Durden: For sure. So a universe is essentially just a list of people that a campaign has decided are worth spending time and money on. Everybody else is invisible so the campaign has made a series of phone calls, kind of like mining, where you're just taking, depth gauges and ratings. You've called a group of individuals, and the data has yielded that. People with a partisanship score of X over X have a higher tendency to vote in favor of your candidate. So now you're just going to take the entire population that has that swath that meets the model that that the campaign has determined suits their demographic need for winning. You've isolated that group. Now you're only going to concentrate on moving those individuals out to vote. because you have determined through the scoring system that those are the likely voters who are gonna turn out in favor of your candidate. Everybody else you don't communicate with because you're not turning out folks who are gonna vote against your campaign. And there's no need to spend time on folks who are not motivated by your issue. So there's no incentive for me to try to persuade a universe.

Allen Pace: You mentioned scores during that explanation. scores get licensed on their own, separate from any list. Who's buying the score itself?

Matthew Durden: it could be any vendor. digital ad platforms, mailhouses, other issue groups that are looking to make a persuadable environment. Literally any business can, ask the Secretary of State for a list as well.

Allen Pace: Let's talk about panels and match services. can you explain the survey panels and how they get drawn off the file What does all that mean?

Matthew Durden: So pollsters will draw a sample straight from the voter file, meaning that they are going to isolate a universe and they're going to pull a sample and then they're going to run basically A B testing on messaging with that group. And through a scientific methodology, they're determining what information is going to make the persuadable argument and what doesn't land. And once they find the most revealing information through those probes, those become the most popular talking points for the campaign at that point.

Allen Pace: Got it. so which of these products would surprise the ordinary voter the most? And which one of them would trouble somebody like me who is in the world of privacy and security and protecting my own personal data?

Matthew Durden: Yeah, I think the most surprising, most troublesome is just about how much personal information is is tied back to your voter data file and how rich that data is and how accurate it is, and how available it is. I think maybe the most valuable is going to be the universe itself because that's honestly where the money gets spent where folks are investing considerable hundreds and millions of dollars, right? Tens of millions of dollars.

Allen Pace: Well, at this point in this interview, I'm sure some of my listeners are curious of why we're doing such a deep dive on voter data files and whatnot, and why the OPSEC podcast would care about this. obviously. Since I harp so heavily on data collection and data brokers, this is deeply in that world. And since it's such a shadow economy that most people don't know about, that's why I wanted to cover it. But the next series of questions really gets into the meat potatoes of why this stuff matters which is data leaks. campaign staff. just download this mass quantity of data about all these different users, as you've explained. who gets access to that? And are there permission levels and who takes it away at the end of the campaign and what happens to it?

Matthew Durden: In a campaign, most of the work that's done is done by volunteers. there's just a handful of paid staff members and campaigns depend heavily on campaign volunteers to do the work. so whenever there's a data project, most often you have volunteers who are doing that. So I'll use the example of canvassing, which is the door to door or phone calls where we're Calling voters, knocking on their door, and asking them specific questions. Those accounts for those lists are given out just kind of willy-nilly. If you have 50 people who are doing canvas that morning, every one of those has the voter file list downloaded to their phone device, an iPad, a computer, or they're given a physical copy of lists that has all of this information. There's a tremendous amount of urgency and lists are just given out. There's no such thing as like permission levels, and the permissions are essentially just set by the campaign, whoever has time to tune it, fit it together, and get the data out the door. When it comes to offboarding, it doesn't belong to anybody. At the end of the day, there's no check to see did anybody take the app that has the voter file on their phone off. So as long as the app still exists on the phone and the campaign is live, an individual can walk around with a remarkable amount of stored data on their phone for entire precincts or the state altogether. So access availability is there's no no such thing as like two-factor authentication. It is wild, wild west, and everybody has access to the voter file at that point. point. It's it's loaded on cell phones, tablets, and and lists are are generated just willy-nilly.

Allen Pace: So let me make sure I understand the summary of your answer is everybody gets all of the data and there is no checks and balances at all, including when those people leave. The data could just travel with them on their personal device and nobody would ever know or care.

Matthew Durden: That has been my relative experience, yes.

Allen Pace: That's great. all right. Well, If I audited as a cybersecurity professional a campaign's account in October, what would I find?

Matthew Durden: Perhaps one login shared across fifty different users. definitely multi-factor authentication. Interns, volunteers, they all have the same access as paid staff. orpaid staff may have shared their access with volunteers and interns. at least one vendor account that nobody can name the the owner of. Like there's no connectivity for this back to an actual staff member, meaning nobody knows who the administrator for that account might look like. and you know, campaigns think in terms of how many days to the election, but there's absolutely no interest in like security posture because that just slows down the program of the campaign.

Allen Pace: So in complete hypothetical world, a bad actor who could be a disgruntled college kid or anybody who has an agenda or wants to become an activist or make a name for themselves could volunteer for any campaign and then leave after that volunteership with the full data set and do what they want with it. They could use it for spamming or hacking or malware ransomware, whatever, they could publish it to the dark web and sell it.

Matthew Durden: Yeah, absolutely.

Allen Pace: Wow. This just keeps getting lighter and lighter, man. has anybody in your twenty-something years in in this field, has anybody ever even suggested that you guys need to tighten up protections for the data for the actual citizens?

Matthew Durden: I don't think that that's a significant concern for campaigns is privacy for success. No.

Allen Pace: So it's never even that talked about. Awesome. I got some more questions for you. So what happens the moment a list leaves a platform and becomes a CSV?

Matthew Durden: At that point, it's just now a document with all of the data in it. So unencrypted can be transferred easily with up to eighteen hundred data fields on it. I think that that becomes economically a piece of priceless gold in terms of campaign space.

Allen Pace: for those who don't know, a CSV file is essentially like a an Excel spreadsheet So you mentioned the canvassing apps before that run on volunteers' personal devices. what's actually on the device when a volunteer loads the canvassing app? Is it literally everything in the data file or is it universe only because they're working in that one universe? What does that look like?

Matthew Durden: Most often it's going to be just their one universe. It'll be the turf that is cut for their area. so it and it could be cached offline because oftentimes they're going to be working without a signal. So these be cached offline information. They use their own phones and so whatever security they have on their own phone, which is often none, and then we often have like the the lost phone issue where somebody has lost their device. and so they have access to name, address, age, party, what registration looks like, what states, their turnout history, and then often the scores that we've talked about. and they are ready to talk about predictive beliefs, and canvassers are usually they have a script in their hands that has been designed by a campaign that is supposed to be somehow provocative or persuasive so that the opening salvo is somewhat inviting for a conversation.

Allen Pace: So,these volunteers, do they receive any training at all related to NDA of personal information, security training or protecting it? any type of NDA for like Personal privacy, sensitive data, financial information. I mean, we have laws in place for HIPAA data for health-related stuff. I find it hard to believe that this data file isn't on that same level of sensitivity or higher.

Matthew Durden: gonna be up to the organization or the campaign to make a determination of what that exposure is going to look like. I've worked for organizations that have really strict policies about data sharing and lists, and I've worked with other organizations that have zero policy. So it depends on what kind of time crunch you're under and what kind of legal exposure your organization's willing to to take

Allen Pace: All right. this is an important subject related to people's data and manipulation that I have strong feelings about. So somebody listening to this is already in these files and cannot opt out. What do you tell them that they can actually do? Like what meaningful steps can they take to protect themselves?

Matthew Durden: I think first you can actually reach out to the state and ask for your own voter file just so that you can have visibility on what is in the file. knowing what's in the file is going to be first and foremost the most important part because then you'll have visibility to what it is that you need to push back against.

Allen Pace: What about the thing I mentioned in the previous episode that's referred to as the address confidentiality program? Are you familiar with those?

Matthew Durden: I am, and I I think that, states really need to make the confidentiality program more available for more people. At this point, it is too difficult for an individual to prove to the authorities that that their information needs to be kept private that needs to be way more expansive and accessible for folks who are interested in keeping their information private. There's there's no reason that policymakers couldn't tighten that up.

Allen Pace: good deal. pretty much the only thing I gave as far as an actionable item in the previous episode was something that I touch on in many different ways and that is starving the commercial data sector. starving the data brokers of of information to add to it. So going after the enrichment layer in this subject with the data broker removal and all the commercial data being bolted on after the fact, how effective would you say that would be

Matthew Durden: I think if you can make yourself as a consumer extremely expensive to profile, that is the best way to to keep your data from reaching the file. the inability for data brokers to scrape data and build a file from you because you've taken proactive steps to protect yourself, your personal identity, and taking those measures makes it extremely expensive for data brokers to to find information about you. The more expensive it becomes, the less advantageous it becomes to the broker. And it becomes far more expensive for them to gather the data and build a profile on you than it is to actually have you as a consumer. So make their jobs harder make it more expensive for data brokers to harvest your information.

Allen Pace: So don't be the low hanging fruit, which is the thing I say all the time on this podcast. And also just assuming that the record is public and enriched and is being sold is a good baseline stance to just expect the worse and then work towards improving it from there,

Matthew Durden: Yeah, I think some of the best time that I spend is the twenty minutes listening to this podcast, finding out the new details of how my data is being distributed and shared. this is this information that you're providing, that we're providing to listeners right now is absolutely critical to understanding how we're being exploited.

Allen Pace: I love it. Well, thank you for that. the last thing I have is let's flip the script and let's say that you're advising a campaign Is there any way to convince a campaign that protecting voter data is a good idea and is valuable to them?

Matthew Durden: I think the best way to incentivize campaigns respecting private information is to make it really expensive for campaigns to to have that information and to protect that information. There there just needs to be carrot and stick here because across the board through this entire theme, the through line here is is that data has a remarkable amount of value and we have to make it more expensive to mistreat that data than it is to absorb that data.

Allen Pace: I don't know how that's accomplished, but I really like that perspective. what do you want voters to take away from all this?

Matthew Durden: You need to recognize that your data is doing a lot more work on behalf of the campaigns than you are. And you need to register, you need to vote, you keep doing it, but you really need to recognize how your data is being used and manipulated to persuade you and drive you towards particular political decisions.

Allen Pace: So understand how you're being manipulated. Even if it's a manipulation in a direction that might be in parallel to your beliefs, still understand how it's being used to guide you in that direction. Okay.

Matthew Durden: That's right. There is somebody there's somebody on your doorstep or there's a piece of mail in your mailbox for a reason and understanding why you've been pre selected, that's important.

Allen Pace: Well, I love that. I will let that be the final word because that is a really good piece of advice to understand. for anybody who hasn't heard episode 29, which posted two weeks ago, I recommend that you go listen to the full breakdown of the chain of events from data creation to data product for more information to potentially need to make sense of what we just spoke about today. Thank you, Matt, for being here and listening and answering my difficult questions. I appreciate it. all the resources for this episode are in the show notes. Thank you for listening to the OPSEC podcast. Don't forget to like and subscribe and follow us on socials. All of our links can be found on our website at OPSecpodcast.com. If this episode has earned its value, lock in the habit. Follow us and subscribe wherever you're listening so that you never miss a drop. And sign up for our newsletter on our website. Subscribing is one of the easiest ways to help us succeed and continue bringing you good content. I am Alan Pace and remember that your privacy and your security is your responsibility.