In a news conference Friday, Bernie Sanders's presidential campaign manager, Jeff Weaver, angrily accused the Democratic National Committee of unfairly aiding the campaign of Hillary Clinton by shutting off Sanders's access to voter data.
It was the latest in a long line of fights between Sanders's team and the DNC -- a fight which includes skepticism over the closeness of DNC Chairwoman Debbie Wasserman-Schultz to Clinton and questions about the scheduling of debates. (The third Democratic debate will be Saturday, a day when most people aren't watching TV. It will be the second Saturday debate in a row.)
What's happening with the data, though, is much more complex than the fight over the debates. To understand it, you need to understand how and why political campaigns use data.
We've established before and will reiterate here that there is nothing more important to a campaign than knowing as many details about you, the voter, as it possibly can. In most places, obtaining a list of registered voters and their vote history is trivial. (For example, the state of Ohio's full voter file is online.) That data usually includes other important details: Your address, your gender, your political party, your age, etc. But in an era when political candidates have the ability and desire to target ads to 45-year-old car enthusiasts who are recently divorced and live in the suburbs of Cincinnati, the more data the better.
Since there's not much utility in, say, a senatorial campaign investing lots of money to build this thing out every six years, businesses stepped in to fill the void. There's a company called NGP VAN which runs a system that allows campaigns easy access to voter databases. Those databases are put together with the help of companies like TargetSmart, which helps the DNC take voter data from state parties (which get it from the state), cleans the data up, layers in consumer information and then slots it, in the case of the DNC, into the NGP VAN system.
You can think of it this way: Your phone has a "Contacts" app which displays all of your contacts. That's NGP VAN, a tool that lets you see and interact with your contacts. Underneath that, though, is a database system that you never see directly, that knows how to return "John Smith" when you search for "smi." That's the database itself.
What happened with Clinton and Sanders was apparently at the NGP VAN level. Both campaigns are using the same tool (NGP VAN) to access the same database from the DNC, but each campaign also has the ability to layer their own data on top of it. So if the basic record says "John Smith, Democrat, lives at 123 Fake Street," Clinton's campaign could add that he also made a donation to them in 2008, and Sanders's campaign could add that he volunteered at an event in Manchester.
The Clinton people aren't supposed to see the Sanders information, and vice-versa. But apparently a staffer or group of staffers for Sanders were using NGP VAN and noticed that some of Clinton's proprietary data was visible. The NGP VAN tool has one record for John Smith, and it's supposed to show the address to both Clinton and Sanders, the donation part to Clinton and the volunteering part to Sanders. But in this case, Sanders's team saw the Clinton information, too. (See update below.)
There are lots of ways that can happen. In a statement, Weaver of the Sanders campaign says that the NGP VAN "firewall" between campaigns was broken, which doesn't mean much in technical terms. NGP VAN is no doubt doing its best to figure out exactly where the bug was. But for our purposes, it doesn't really matter.
What happened next, as The Post reported Thursday night, was that the DNC shut off Sanders's campaign's access to the data.
— CNN Politics (@CNNPolitics) December 18, 2015
There are two ways to look at this. The DNC argues that, since Sanders's camp had access to Clinton's data — and exploited that to the extent that Sanders's campaign fired a key staffer — locking them out until the problem was fixed was essential. (It's not clear what the data was, but it could be significant. Voters that Clinton identified as supporters, for example.) Sanders's campaign, in the statement mentioned above, pointed out that NGP VAN had repeatedly been problematic, and that it was the software tool that was at fault, not them.
It's clear that this is a serious problem for the Sanders campaign. There is just more than a month until the Iowa caucuses, and the campaign needs to be knocking on doors and making phone calls. One of the things each campaign puts into its NGP VAN data is who they've talked to and what they've said. If Sanders can't make calls or doesn't know whether they've called someone before, that's a huge challenge — especially in Iowa, where such voter-level persuasion is important.
What's not clear is how quickly the issue will be resolved. Sanders's campaign has done a good job of looping this into their bigger fight with the DNC which, again, it thinks is overly sympathetic to Clinton's candidacy. But that victory is not worth being locked out of their data over the long term, so they're probably as eager to move past this as is the party.
The best days to contact voters are weekends, when most people are home. There are six more weekends (excluding the one after Christmas) between now and the Iowa caucuses. Sanders's campaign will apparently spend this weekend without access to data. Whose fault that is probably comes down to which Democrat you'd rather see win the nomination.
Update: Early Saturday morning, the DNC restored Sanders's access to the campaign data. The Clinton camp revealed the data that Sanders's staff had accessed, which appeared to include lists of people demonstrating particular levels of support for Clinton or likelihood to go out to vote in South Carolina primary.
This image from Clinton's public response shows the data that was accessed. Voters are often assigned scores from 0 to 100 on attributes.
If someone says, "I am voting for Hillary Clinton," they get a 100 or thereabouts. If they are demographically likely to support her, they get a high score, too. People who vote infrequently will get a low score in that regard.
The screenshot above appears to show Sanders staff trying to find people who weren't strong Clinton supporters ("HFA Support <30") -- perhaps to pitch another candidate to them.