- / -
Collect information from the field about the field; structure the campaign accordingly.
Seek the help of initial volunteers to learn about the neighborhoods. LOs and FOs work together to divide precinct assignments in logistically-sensible groupings. Acknowledging that time constraints and technical problems make 25 precincts per organizer unfeasible, select 15 out of a possible 25 (taking into consideration demographics, geographics, quality of calling lists). Create satellite offices if necessary, using volunteer homes if possible.

Match the skills to the tasks.
Among the organizer staff, identify a few effective public speakers and experienced trainers; assign them to regularly run recruitment meetings. Identify a few organizers with the best computer skills to grapple with the chronic technical difficulties, and to facilitate the implementation throughout the office.

Likewise: as volunteer teams grow, diversify volunteer functions. Many volunteers have no computer skills whatsoever, others are computer-savvy and much more willing to do hours of data entry than to knock on a single door. Many people can't or won't canvass (be they elderly, unhealthy, or just plain shy) -- however, they could be trained to make recruitment calls, which would relieve the pressure on the organizers to spend a third of their day on the phones. With that time freed up, work closer with teams: identify and specifically train canvassers, canvass trainers, computer literates, computer savvies, map-makers, data-enterers, etc. Distribution of work demands is even more essential in the face of a malfunctioning computer system.

Adapt the model so that the work is smarter, not harder.
The model schedules, scripts, and monitors the results of every single interaction on a campaign -- you can anticipate that x hours of recruitment calling with a properly-delivered rap will yield y meeting attendees, which will then yield z volunteers. But the numbers in the model are vulnerable to uncontrollable factors: phone lists might be bad and yield fewer contacts than expected; precincts might be too far away to yield meeting attendees; gated communities or apartment are often not conducive to canvassing. Reassess the model's parameters accordingly; resize goals and redeploy resources if necessary.

Account for regional specificities: in Boca Raton, don't schedule a community meeting on Shabbat; in Las Vegas, try cold-calling at different times during the day, to see if there's a better period in which to reach the service industry-based populace at home. Ensure that time is being used most effectively: if a part of the model is not working, determine why and adjust to correct.

Remember that the quality of volunteer experience matters as much or more than the quantity of volunteers.
A single volunteer who is trained, focused, and confident is more effective than ten volunteers who signed up one night, never to be further engaged. Facilitate team bonding. Spend extra time with volunteers who need it; it will pay off (up to a point, of course).  Ultimately, a smaller number of well-trained teams would be able to expand into the rest of the original 25 precincts, and approach the original goal of voters. But almost all volunteer leaders will only lead others if they themselves are being well led.

Network the group together, and help it solve its own problems.
Provide an online forum for volunteers (there were, even then in the morning of Web 2.0, plenty of free and easy options!), complete with canvass scheduling, FAQs, social networking, and more. Use this forum to facilitate the information flow as pertaining to all of the above objectives.

    - / -

Now. On any campaign, or any intensive business operation, one might expect to hear gripes from lower-level staffers like 'we need more autonomy,' or 'there is not enough time to meet the goals.' In a system with so many 'moving parts' and so little time in which to move them, it is inevitable that some resources will mismanaged and time is going to be wasted, because decisions are being made that can't possibly take into account all of the available information. There is even a specific economic term for this: 'hierarchy cost,' which is the productivity loss associated with information that is 'unavailable' to the decision-makers in the higher levels of a firm. That information is most available to the people on the ground who are engaged with the production -- accordingly, the scenario outlined above is that of a campaign engaging itself in an ongoing process of reducing hierarchy costs and adapting the campaign model to best fit the reality.

But in Leave No Voter Behind, every one one of the dynamic processes listed above was deprioritized, discouraged, or often explicitly disallowed by GCI management. All functions of staff and volunteers were determined by rank alone, with no variation therein. Assignments and implementation was dictated from above, and the goals and processes thereof never changed, except for a steady increase of work hours. There were no viable channels for feedback from organizer staff. Without any functioning online networking, volunteers remained largely disconnected from each other. The ability of managers to closely monitor their organizers was more important than the ability of the organizers to closely work with their volunteers. Training was almost entirely sacrificed for the sake of sheer recruitment.

This last failure --the near-total neglect of volunteer training-- was egregious and irresponsible on the part of senior GCI management. Everything above it, however, was "on point" -- by the very terms of the PIRG/Fund model itself.

In the comments, Peter from WI--who initially defended the PIRG/Fund model--seemed to acknowledge that the management ethos in LNVB "devolve[d] into irrational dogmatism and silliness.'  I wouldn't have chosen the word "devolve" -- what we saw in Leave No Voter Behind was a strict reinforcement of a core (though largely unacknowledged) principle of the PIRG/Fund model. I examined this principle in "Strip-Mining the Grassroots":

The canvassers and directors who actually do the work have no influence upon their work conditions [or] the distribution of the funds they raise...the 'raps' are created without input from those delivering them; all decisions are made without any accountability to those below.... [T]he model is, in a way, curiously apolitical. I don't mean that it's not 'liberal' -- but rather, that it does not engage in deliberative processes. ... Its participants (both the canvassers and those who are canvassed) do not participate in open-ended, discursive relationships with the organization or each other.

In this model, the function of "organizers" is only to execute the orders of those above them, motivate those below them, and relay bits of information in between. They do not assess that information, and they have no capacity to decide upon an action that will best achieve the goal. There is only work and harder work; there are only masters and slaves; the model is the master of them all.

Not to get too metaphysical here -- this point actually makes the discussion quite concrete: for all the bad decisions being made in LNVB, there were ultimately a handful of decision-makers. To find them, you can't look anywhere but up. This explains why, after the election, so many organizers were torn between two extreme opinions: either the people in charge were wholly incompetent or they were frauds.

Let's look at an unusually clear insight into the mindset of this "island of conscious power", which can be seen in a comment by Ricardo, an LO from one of the nation's best offices. Ricardo explains how the model was 'adapted' by senior GCI management:

A lot of the LOs would say [to their superiors] in a conference call: 'we're having trouble with the precinct leader retention rate, because we're on the phones for so long.' The response we got was, 'We have x number of hours that we have to be on the phone for the state. If your office feels that it's going to fall short of the amount you're supposed to fit in, call your AOD and tell her, and she will report that to the state director, and he'll divvy those hours up among the other offices.'

So management actively rejected information from below -- insisting that the feedback given by all individual units was irrelevant (or even wrong) in the face of the model itself. It would seem to suggest that senior GCI staff are frauds: that they had a contract with MoveOn to recruit z number of volunteers, so they sacrificed the rest of campaign to hit that number and 'win.'

I see two explanations for this behavior, and once again, I'll leave it to the reader to decide which one is more disturbing. One, they simply don't trust the intelligence of the people they're "leading," rather consider them mules who will work harder if kicked hard enough. For a while, that conclusion was satisfying enough for me. But reason Two was suggested by a number of managers who had seen several stages of the model's deployment (from the troubled 'alpha test' to the post-mortem debrief): GCI's senior management chose disallow adaptation in order to maintain their control. They understood that this would put unworkable constraints on some organizers, but they wanted the model to be implemented as rigidly as possible, in order to have more concrete results that could be analyzed and tweaked accordingly in future campaigns. As a result, the hierarchy costs were maximized, to the point where these costs were often greater to participants than the benefit of organization itself. This resulted in massive attrition of volunteers and organizers, who really are smarter than mules.  

Dogmatic? Yes. Silly? Oh, yes. But frauds? Well, not quite--arrogant control freaks, perhaps. But you couldn't even say incompetent, either -- since they got the numbers they wanted, and MoveOn rehired them. Of course, none of this information is reliable -- as Private Kicker noted, most of the numbers that GCI wanted so badly were simply falsified by organizers who had no other option.

  - / -

By now, MoveOn shouldn't have to be told that the model was broken during LNVB -- at this point, a year into its second try with GCI, it should know this perfectly well. In the last month, I have been contacted by a large number of ex-Operation Democracy field organizers, who confirm that the attrition rate for the second wave of MoveOn/GCI field organizers is at 80-90%! This clearly shows that the hierarchy costs are still overwhelmingly high; that the model is still broken.

My next series will get into specific about ways in which MoveOn might try to 'fix' the model in order to approach the kind of dynamic, effective campaign outlined above; I will also report on Operation Democracy itself. But the basic issue here is organizer agency -- the top-down hierarchy must become a network of principles and agents, rather than a chain of masters and slaves.

We're talking paradigm shift here, and it won't just come about by asking for it -- in such a closed system, there must be a formal imposition of viable channels for internal communication and deliberative dynamics. Legitimate worker representation would force management to operate by process, not by decree -- its result would not only be happier organizers, but better campaigns. There are a number of different possibilities that could begin this process, like bottom-level staff groups networked with volunteers, GCI management and MoveOn staff into cross-functional teams. It is the client's responsibility to ensure that its campaign is working properly and progressively -- and it the workers' right to try to influence it as such.

In my next post, I'll write more about the issue of hierarchy costs, and about one specific change that would significantly increase bottom-dynamics without fundamentally changing the model's top-down structure. Hint: it's quite literally right under your nose.