Pulling Off Houdini's Trick

On election day, 2008, the Obama campaign experimented with a new system designed to fix one of the strategic problems that can plague election efforts: not knowing which of the potential voters field organizers work so hard to identify as supporters have actually made it to their polling place when it finally counts. Called "Project Houdini," the campaign planned to used technology to vanish voters from its highly-cultivated contact lists in real time. If it could pull of the trick, the campaign could gain a strategic advantage by not wasting critical resources (people, vans, etc.) to pulling in votes that had already been banked. Newsweek's Special Election Project described what Project Houdini aimed to do:

The Obama campaign's New Media experts created a computer program that would allow a "flusher"—the term for a volunteer who rounds up nonvoters on Election Day—to know exactly who had, and had not, voted in real time. They dubbed it Project Houdini, because of the way names disappear off the list instantly once people are identified as they wait in line at their local polling station.

But, while there's still digging to be done, it's clear that in some critical swing states Houdini didn't quite work as planned. More importantly, when it collapsed, it took down some critical reporting channels with it.

Let's back up. The way that Houdini was supposed to work was simple: trained "Houdini volunteers" patrolled polling places to keep tabs on which voters showed up. Each voter had been assigned a unique four-digit ID number, based on his or her precinct. When a critical mass of voters was reached, Obama volunteers were to call into the Houdini hotline, punch in the voters' codes, and "disappear" them from the rolls. According to campaign documents I've seen, though, Obama voter-file team told some state field offices as early as 10 a.m. to tell volunteers to abandon the phone-based Houdini. The problem: the IVR (Interactive Voice Response) lines were swamped. The campaign scrambled for fixes. If they could get to a computer or WAP phone, volunteers were directed to a web-based interface at watchforobama.com instead. And, if all else failed, they were told to resort to the age-old tactic of having staff and volunteers schlep paper poll-watching sheets back to campaign offices.

Of course, even the best-laid plans often go awry on election day. But where Project Houdini's problems were more serious than they would otherwise be is that the IVR phone lines rendered useless were the same ones used to call in the turnout numbers from each polling place -- hard data that campaigns use to model turnout predictions and craft high-level strategy in the waning hours of election day.

Project Houdini's ambitious vision sparked debate inside the Obama campaign. The new media team's plans raised hopes that Houdini would smooth some election day wrinkles. But some -- including organizers steeped in the old ways of doing field -- were concerned that Houdini wasn't the campaign's best use of time and resources. If it had worked without flaw, Project Houdini would have unequivocally joined the list of politech innovations the Obama campaign got very, very right. But, given the snafus, there are likely still hearts and minds of hardened field organizers still not entirely convinced about how far field technologies can go towards winning the election-day ground game.

UPDATE: The National Journal's David Herbert has some great reporting on how Houdini worked and didn't work on the 4th.