What is field?
First, a quick primer on the nitty-gritty details of field, as I understand it.  I've worked as a volunteer in perhaps 5 or so local races in the last few years, in a variety of different capacities (phonebanking, door-to-door canvassing, etc.).  These races have ranged from city council up to state senate.  And I did a fair amount of volunteering for the Dean campaign back in the day.  I've always been a volunteer, never a campaign worker or strategist.

Our field method, at least here in eastern MA, is as follows.  The months before the election are consumed mostly with canvassing and phone banking; from the state, we gather the list of people registered to vote in the district (i.e. the "voterfile"), and make a local copy within campaign HQ.  If we're working on a primary, we filter the list by party.  Then we contact as many voters as possible.  Face to face contact (i.e., door-to-door canvassing) is preferable, but phone contact works as well.

The purpose of all this activity is to determine which voters are supporters of the candidate, undecided, or opponents (usually coded as "1"s, "3"s, and "5"s, respectively, with 2 and 4 reserved for people on the fence in either direction.)  A second purpose is to "clean" the list of registered voters - voterfiles are notoriously cluttered with people who have moved and/or no longer eligible to vote.  It's very important to know who those people are and to clear them out of our private copy of the voterfile.  As we call and canvass voters, we annotate our private voterfile with these supporter codes, and cross out voters who have died or moved.

The week before the election, the campaign switches into get out the vote ("GOTV") mode.  GOTV mode is preoccupied with getting the "1"s to vote, not finding new voters.  Each supporter gets bombarded with phone calls, door-to-door visits, and day-before door-hangers specifying polling place location and other details to remind the supporter to vote.

On the day of the election, we create a new version of the private voterfile, which I call the supportive voter list.  This version contains only the "1"s, or possibly the "1"s and "2"s, and their address and phone number.  We are primarily concerned with making sure that everyone on the supportive voter list votes.  As we find out that people have voted, we cross them off the list; hopefully, by the end of the day, there's no one left on the list.

How do we find out who voted?  There are several ways.  One way is to canvass and phonebank supportive voters - repeatedly, again and again throughout the day - until they say they've voted.  Another way (and this is legal by MA election law, I'm not sure about other states) is to place a volunteer in the polling place with a copy of the voterfile; this volunteer is called the pollchecker.  The pollchecker's job is to listen in as voters declare their name and address before getting a ballot.  If a voter's name is on the pollchecker's supportive voter list, the name is crossed off the list.  Other volunteers (precinct captains) visit the poll-checker throughout the day to "pull" the poll-checker's list; the list will be synchronized with HQ's copy later on.  Meanwhile, the poll-checker gets the latest copy from HQ, reflecting results from earlier pulls, canvassing, phonebanking, etc.  The lists for canvassers and phonebankers are similarly updated periodically throughout the day.

What is wrong with this method?

If you've been carefully following this discussion, or you have some experience with field/GOTV, you'll note a few problems with this method.  Please be sure that I am not trying to downplay the work that campaign workers and volunteers do.  In many cases, campaigns choose this method because it's the cheapest one available.  The workers, strategists, and volunteers who implement this method are hard-working and well-meaning people, by and large.  In many ways, it's an extremely grassroots, anti-big-media approach to politics, and it is commendable for that reason.  But it's far from perfect, and it suffers from a lot of flaws:

• It is conservative in terms of who the voter universe is; it assumes only currently registered voters will vote, and doesn't encompass much new voter registration.  In a sense, this is a very undemocratic assumption.

• It is very labor-intensive and requires a lot of boring, repetitive, hard work.  It's very difficult to turn volunteers out for this kind of work; what's worse is that grumpy (or just untrained) volunteers can lead to bad data (e.g. an undecided labeled a "1").  From the point of view of getting your supporters to the polls on election day, this is disastrous.

• It relies on stone-age technology.  If each poll-checker had a Blackberry which was automatically sync'd up with a server at campaign headquarters, and if that Blackberry were sync'd up with reports from canvassers and phonebankers who are hearing first-hand accounts of voters who claim to have voted, then a lot of precious time can be saved on Election Day.  Think this is bad?  It gets worse.  Many voters vote between 6 and 8 pm.  Guess when the last refresh of supportive voter lists (for the pollcheckers, canvassers, and phonebankers) is?  5:30 pm, usually.  That means that in the crucial 6 - 8 pm time frame, some of the people who our canvassers and phonebankers are working hard to contact, are people who have already voted.

• It is extremely annoying to voters.  It's not atypical for some voters - and these are the campaign's best supporters, mind you - to get ten or more contacts in a single day.  Generally the animosity created this way is surprisingly minimal, but it's certainly not pleasant for the voter or the campaign volunteer to deal with.

• It plans inefficiency.  There are some common voting patterns which this method doesn't handle very well.  Primarily: many voters tend to vote at night, after they return from work; and many couples vote together, i.e. if one partner is home, he or she will wait for the other to return before going out to vote.  There are similar patterns for elderly parents who live with their middle-aged children.  With this method, these people get the most contacts of all, since they typically vote last.  Unfortunately, these people account for about 40 - 60% of the Democratic electorate, by my rough estimates.

• It targets exactly the wrong people.  Guess which kinds of people are registered and tend to be home when canvassers come around, and phonebankers call?  Folks who are middle-aged and elderly, which is to say, the least Democratic generations of voters (they voted about 5 - 10% less Democratic than young voters in both of the last two federal elections.)  This method is horrible at registering, contacting, and turning out young voters.

• It does not take advantage of relationships.  One of the nice things about pre-GOTV canvassing is that it's a good way to establish relationships with voters in your neighborhood.  In a similar vein, the campaign sometimes has supporters who have a wealth of social connections within the district, especially if the supporter has a civic leadership role (as coach of the school softball team, for example.)  This method does not take advantage of either of these kinds of relationships.  During GOTV, a canvasser or phonebanker is very likely to be talking to a stranger, and a supporter is very unlikely to be contacted by his or her friend or previous canvassing contact.  Of course, a contact from a friend is much more persuasive then a contact from a stranger; one reminder from a friend is likely to be much more effective then ten reminders from strangers.

How Facebook apps can help

There must be a better way to do field, and I believe that a high-quality Facebook application can be part of the solution.  What would such an application look like?

The target audience would be volunteers who are active supporters of a campaign.  To kick off the usage of this application, the campaign would periodically ask people on its email list to add the "Field/GOTV" application to their profile, and subsequently sign up for the candidate's group within that application.

The goal of the volunteer would be to bring as many "1"s into the campaign as possible; to make sure that all of those "1"s are registered on time; and to be ensure that those "1"s vote.  The supporter would accomplish these goals using a variety of tools provided by the application:

• Finding friends and second-order friends in the district - As a first step, the volunteer would look for friends who live in the district, and friends of friends who live in the district.  The application would generate this list automatically for the user, and would provide the user with optional email and/or RSS updates when new friends appear.

• Determining the registration status of friends in the district - The next step would be to align friends and second-order friends with the state voterfile.  Non-registered friends would be reminded by the user to vote, and each registration could count towards a point system (if we want to get hokey.)  A similar system could handle absentee ballots.

• Recruiting "1"s - At this point, in my opinion, the interaction has to go offline.  Recruiting "1"s is a very difficult process, even among friends, and it can't be done by email or IM.  On the other hand, it might be done by house party or some other offline event, which could be facilitated by Facebook tools.  Or it could be done using old-fashioned one-on-one conversation.  Regardless, this part of the application would allow the user to mark each friend's level of support.  Needless to say, levels of support should be synchronized with the campaign's internal voterfile, so that the campaign doesn't need to worry about phonebanking/canvassing the user's friend (which would be both inefficient and annoying.)  On the flip side, if the volunteer registers a friend and turns that friend into a "1", that friend should be added to the campaign's voterfile but not contacted by the campaign.

   

  
   
  • More thoughts on recruiting "1"s.  One of the dirty secrets of a campaign, as I understand it, is asking "1"s to volunteer for the campaign, or to put up a yard sign.  This isn't because of the need for volunteers or additional signage per se, but rather because the process of volunteering, and the process of displaying a yard sign, makes the voter a "hard 1", i.e., a person who is extremely unlikely to change his or her mind.  This is a way to counteract the problem of mislabeled "1"s described above.  I think there might be some way to automate the volunteering process online, and thereby to auomate the process of recruiting a "1".  For example, perhaps anyone who organizes a successful house party or a fundraising drive can be considered a "hard 1" automatically, without any offline interaction.
     
  • Turning "3"s into "1"s.  The campaign should notify the supporter of events where the candidate will be appearing and speaking, and encourage the supporter to bring his or her "3" friends to those events.  This often has the effect of making a "3" into a "2" or a "1".
     
  • Cold calls.  Facebook has some pretty good profile information, which might be useful in recruiting strangers to the campaign.  For example, the application could find all of the people marked "liberal" or "very liberal" in the district; or it could even use some secondary characteristics, like religion and hobbies, to guess the ideologies of potential supporters.  These lists could be provided to the campaign volunteer, who could email potential supporters individually, invite them to a house party, or do something similar to engage them in the campaign.
     

• Voting reminders - This is where things can get creative.  Facebook provides a number of ways the campaign can send voters (relatively) unintrusive voting reminders.  These include feed notifications, emails from the campaign, personal emails from the voter's friend, and SMS messages on the voter's cell phone.  I'm particularly excited about SMS messages because, delivered at the right time, they can be very effective in reminding people to vote; people who voted can also use SMS messages sent to the campaign to indicate "I've already voted".  Naturally, to the extent possible, the campaign should get permission to send these messages before election day, and of course, only send messages that are permitted.

• Voting checks - Where possible, the campaign supporter should be responsible for determining at what time during Election Day the recruited voter will vote, when the voter can't vote (i.e. is at work), and whether the voter needs an absentee ballot or ride to the polls.  The volunteer will annotate the voter's information in the voterfile with that information.  If this information is reliable - and it often will be, coming from friend-to-friend interaction - it can make Election Day a much more pleasant experience for everyone.  There's no reason to call someone during the day when he or she is at work.  There's no reason to bug someone in the morning if he or she will be voting at night.  And so on.  There are a variety of heuristics which can be applied using this kind of information.

• Election day reminders - If possible, the volunteer should be on hand to call voters who forgot to vote on Election Day, and remind them as the day goes on, and as appropriate according to the voter's schedule, to vote.  If it's not possible for the campaign supporter to do this herself, then she might be able to hand the task off to a friend who is also a supporter.  There will be far fewer reminders as a result, but they will be much more effective.

• Poll checking - If possible, the volunteer should train his or her friends to update their Facebook status as they vote, by joining a group set up for the day, i.e. "I voted for Jane Doe in the 2008 Capital City mayor's race".  This method, of course, has the benefit of informing the voter's friends that there's an election going on.  Another method, as I mentioned earlier, is to set up an SMS code where the volunteer's friends can text in their message to indicate that they have voted.  Naturally, these updates should be integrated into the campaign's voterfile to avoid redundant contacts on Election Day.
  

I should make one thing clear - no campaign should rely on such an application as its sole field/GOTV strategy, by any stretch of the imagination.  Facebook penetration is not deep enough to warrant that kind of decision.  On the other hand, if handled carefully, this kind of social networking strategy could be an excellent complement to the on-the-ground strategy, and could relieve the campaign staff of a lot of headaches and work.

This is a pretty tall order, and I'm well aware of that.  The cost of developing such an application are probably enormous, even if we ignore some of the more thorny problems (like integrating Facebook information with the campaign's internal voterfile).  And it's possible that even if we got started developing this application today, the application wouldn't be available until well into 2008.

On the other hand, this application is a perfect opportunity for Democrats to leap-frog Republican field efforts.  Facebook's audience is exactly the kind of audience Democrats largely ignore - young, mobile, technologically savvy, probably mostly progressive, and extremely creative.  We would do well to start recruiting this audience in the place where they hang out.  In doing so, we could win many more votes and open up a whole new avenue of pleasant volunteer experiences for savvy Facebook users.

I'd love to hear your thoughts about how this application could be made more robust and useful, for the campaign, the volunteer, and the volunteer's friends.  If you have thoughts about how to reduce the scope of the application but make it more feasible to implement in time, I'd love to hear that too.  And as always, if you want to jump on this idea, please do so!  I'd love to see someone turn this into a profitable idea.