It's hard to write about voter files without getting into the weeds a bit, but some context is helpful to understand the genesis of the debate about whether the DNC's voter database works.
Before the '04 cycle, the DNC asked a company (now called Plus Three) to help turn its old database, which literally resided on a dusty old computer, into a dynamic targeting powerhouse. Plus Three constructed a database called DataMart and merged with what Terry McAullife liked to call Demzilla -- a separate collection of DNC donor lists. Those donors -- the party's regular voters, certainly -- suddenly had full demographic and interest profiles.
The challenge: build on the technology and create a larger database of folks who weren't regular Dem voters and weren't regular Dem donors. The Dems appended as many as 150 data points to each person. Institutionally, state parties balked. The data was theirs, by tradition. They didn't want the DNC to have what was rightfully (wherein rightfully means by unquestioned tradition) theirs.
An agreement was cobbled together. The state parties would scrub their lists twice a year and give the data to the DNC, but in most cases, the DNC would not itself be able to do much more than give state party data manipulators -- many of them untrained and not too sophisticated about the scope and limitations of the system -- access to the coordinated data.
What the DNC had by the start of the '04 cycle was a computer system with a poor interface, jealous state parties, a mistrustful set of presidential campaigns, poorly trained managers and very few strategists who actually understood how targeting worked and how it could be helpful. They also had not actually put the data to good use -- or any use -- by time John Kerry won the nomination.
That's lightyears behind what the the GOP's Voter Vault project achieved by the start of the same cycle. The interface was intuitive. Depending upon the voter, the RNC could tell as few as 20 or as many as 60 different facts about them. The project was scientifically managed. Trials were repeated; assumptions were tested.
For Dems, Laura Quinn was brought on to manage the Datamart project. Data was acquired, though maybe not enough. The software was tweaked. There was little time for tests, though, and the wheels started to turn quite late -- in the summer of '04. As we've written, she and others had four months to do what the GOP did in two years. Technology does not follow the political calendar by itself, and for Dems in the '04 election, the Datamart project did not succeed.
The DNC asked Booz Allen Hamilton to figure out why Datamart failed. It blamed people and technology. The DNC hadn't spent the resources to train people who knew how to create and expectations for users of the system.
When Howard Dean was elected chair, he decided to spend more money to staff organizers in every state (requiring an enormous upfront disbursement) and to take the file project more slowly. The last thing the party needed, in Dean's thinking, was a second messed up system.
Today, under the direction of Ben Self, the DNC has produced a mighty fine online user interface. They're purchased some data sets and plan to spend (or in the Democrat's lexicon, "invest") in much more over the next several months. Right now, the database isn't terribly useful in terms of targeting because there are relatively few data points that come up when, say, the name "Karen Finney" is run through. But that is changing. Outsiders familiar with the DNC project worry that Dean has spent too little to train enough users for the system to be useful in 2006.
Who puts the "data" in databases? Well, the DNC can buy lists directly or it can purchase lists kept by other firms, like InfoUSA. Since there are a set number of "list firms" out there, both Data Warhouse and the DNC's effort will probably buy the same data from the same sources.
Who builds the "base" in database? The usual crowd, including Plus Three (which brags about its old DNC work here).
The good news for Dems is that the Ickes project intends to raise enough money to invest in both the purchase of lists and the tweaking of software necessary to catch up with the Republican Party. Data Warhouse has the luxury of testing its techniques, too.