-/-

At the national training for Leave No Voter Behind, there was a considerable amount of buzz about a recently-published book called Get Out the Vote!. For this landmark text on turnout methodology, Donald Green and Alan Gerber used scientifically-rigorous statistical analysis to sort through the effects of various GOTV strategies. The book's most prominent conclusion was that the surest way to increase the voting odds of an "unlikely voter" -- that is, a registered citizen who has failed to vote in a recent election -- is through face-to-face contact with another person who lives in the neighborhood. This conclusion surprised no one, but the margin by which the canvass method towered over other strategies was striking: "one additional vote is produced for every fourteen people who are successfully contacted by canvassers." Gerber and Green further note that, despite this potency, massive electoral canvass campaigns are not common -- the canvass is far and away the most resource-intensive of all campaigns, requiring a daunting amount of resources and presence within neighborhoods.

Hence the buzz in our campaign: MoveOn's one great strength is its members, at that point more than two million strong -- informed, active, tech savvy (well, tech capable), geographically diverse, and loyal to MoveOn for its demonstrated eagerness to put such resources to creative and highly visible work. In the fall of 2004, MoveOn could run the kind of campaign that Gerber and Green (whose book is really a practical guide for your friendly neighborhood campaign manager) would hardly even allow themselves to dream about.

Leave No Voter Behind's strategy would channel this strength solely into canvassing, in a very thorough plan consisting of three distinct phases. The first phase was to identify John Kerry supporters out of a list of the 'unlikely' Democratic voters in a precinct. In 'Phase II,' during the 72 hours before election day, these identified Kerry supporters were to be recontacted and asked when they planned to vote. Finally, on election day, each precinct team would be able to track the attendance of its 'unlikely voters' one by one. By doing so, we'd be able to hunt down the inevitable group of stragglers and "stand on their front porch until they come out to vote." Rather than just canvassing to increase the chance (by one in fourteen) that each canvassed voter would show at the polls, Leave No Voter Behind intended to grab the unlikely voters in each precinct and eliminate chance almost altogether. Get Out the Vote! doesn't even consider the potential effects of this kind of triple-tiered voter engagement.

But the truly unprecedented element of Leave No Voter Behind was its size. There were 500 organizers stationed throughout the swing states (around a third in Florida and Ohio), each to be assigned 25 precincts. In each precinct we would recruit a team of 5-10 volunteers. Those precinct teams would each target a list of 150 voters, and turn out 65 of them on Election Day. And it was all going to happen in six weeks.

A seasoned campaign manager would probably expect this kind of operation to be on the ground three months in advance. But Leave No Voter Behind had three things, supposedly, that most campaign managers don't: the urgency of the most important election in recent history, the organizing skills of a largely PIRG-trained staff corps, and a cutting-edge internet-based "Web Action Center," which would streamline and automate the whole process. (Hint: that last one was the important one.)

Which brings us back GCI's post-election claim, as repeated by Eli Pariser and MoveOn: that Leave No Voter Behind actually saw almost half a million voters check in with our volunteers at the polls on Election Day. You see, reaching this goal would entail that not a day was wasted and the system worked up and down. That, or, volunteer angels flying out of our butts.

Now, far be it from me to begrudge a couple of enterprising progressive organizations for buffing their numbers column up a little bit. But while a few offices did manage to pull through and meet their goals (like Philadelphia, West Palm Beach and a handful of others), they were the best-case exceptions. A full week after deployment, many offices were still waiting to move past the Start line; two weeks in, a wave of attrition wiped out a quarter or more of the organizers throughout the country; from that point on, attrition stayed at a high rate, burning through hundreds by the end. Many offices came close to breakdown, and at least one was fired summarily. By the Friday before Election Day, Phase II (recontacting voters to establish their voting plans for Election Day) was effectively jettisoned. As for Election Day itself, it is hard to find any organizer (even from those best-case offices) who claims that their precinct-monitoring system actually worked well enough to get significant numbers of voters to check-in at the polls.

Let me be clear here: this series is not trying to concern the reader with the 'minor' problems and sundry frustrations that will pop up in any campaign. I'm not crying "crisis of leadership!" because everything didn't proceed according to plan and we didn't make our turnout goals. Yet it is necessary to understand why we fell so short of those goals before this discussion can get into the real ugly stuff hinted at above. The explanation is simple, really -- two-fold and right up front.

One. We got out late in the game. This was actually an acknowledged problem from the beginning -- the entire campaign was conceived late (mid-summer), and the six-week plan was obviously going to be a cram. But then MoveOn and GCI didn't manage to finally sign their contract until September 1st, weeks after the organizers had actually been hired. (Apparently, they had a difficult time securing access to voter lists. I've heard a number of rumors about why, but I won't bring them up here -- yet some alluded to it in the comments, and I do think it's an important discussion to have -- please share if you know something.) The end result, whatever the reason, was that the dates for training and launch were pushed back weeks. In the meantime, infrastructure was not pre-arranged -- so that even once we got on the ground, many offices still did not have proper internet and phone access until October. Altogether, more than a third of the allotted time was lost up front -- but the Leave No Voter Behind goals were never adjusted to acknowledge this delay.

Still, there's a favorite catchphrase among the GCI staff: go big or go home. Considering the circumstances, even if you think this delay was irresponsible, you might still admire the chutzpah of the people who put the whole thing together.

But that brings us to reason number Two. Several people already nailed it in the comments: the 'cutting-edge' Web Action Center -- the internet-based spinal cord of our entire operation, through which volunteer recruitment, management, and voter data were all hardwired -- collapsed right out of the gate. This was totally immobilizing at its worst; when things got better, it merely became an enemy that we constantly had to fight against. I don't think it's quite right to blame the tech people themselves for this, since their system was not even put into beta before being dropped upon thousands of users. To make matters worse, GCI trained its organizers as if most of the actual work of organizing would be done for us by a computer -- in other words, it hardly trained us. What GCI got in return was a campaign that had to be rewired from scratch. But the Leave No Voter Behind goals were never adjusted to acknowledge this system failure.

This second point is where chutzpah curdled into hubris and the wings began to melt. Mind, both MoveOn and GCI share responsibility for these two up-front setbacks. But when I wrote about a "crisis of leadership," however, I was referring to what came next. As the sheer size of the operation began to buckle itself under, rather than cut losses and reconsolidate to run the best campaign possible under the given conditions, GCI shoved the burden of that responsibility right down to the staff on the ground. The consequences of this decision will be the subject of my next post.

Considering some of the comments we saw on my last post -- in which those who spoke of 'system failure' were derided by GCI management as not being "tough enough"-- it seems possible that MoveOn only really knew about these initial two setbacks. One would hope that -- since sentiments like this were not uncommon -- they'd have more interest in the quality of the campaign that was being run in their name. But the logic behind MoveOn's decision to re-hire GCI seems to be that 2004 was a test run--a hundred some-odd thousand Ohio votes and George W. Bush notwithstanding--and they'll be more prepared to run a better campaign going forward. With that in mind, I'd just like to make two final notes in regards to this second go-round.

MoveOn's GCI-run Operation Democracy launched well before the 2006 election -- so that's good. And their computer system is reportedly more stable now -- that's good too.

But the contrast between the two campaigns in terms of their goals is remarkable. Operation Democracy is every bit as modest as Leave No Voter Behind was ambitious. First of all, the field organizing staff for Operation Democracy is about five percent the size of Leave No Voter Behind. But the plan itself has been downgraded not only in scale, but in depth. Rather than building on-the-ground precinct teams throughout the country, Operation Democracy's electoral strategy is focused on a selected handful of races. And rather than canvassing, the GOTV strategy this time around is phone-banking, a method that Gerber and Green describe as a "hit-or-miss affair." A professionally-staffed and trained volunteer bank might turn out "roughly one voter for every thirty-five contacts" -- but MoveOn volunteers will oversee their own banks, and since they'll be calling into targeted precincts that are most likely not in their area code, that number can be expected to fall even further. Finally, Gerber and Green note that phone calls made before the final week of an election do not have any statistically significant influence upon turnout. Ironically, Operation Democracy is preparing to begin phone-banking as early as September.