Why the Distributed Organizing Debate Won't Settle

Part of The Build. Start with the guide for the full map.

Distributed organizing—equipping volunteers to run their own local organizing efforts with minimal direct campaign staffing rather than centralizing every field decision through paid organizers—has one of the most devoted followings of any idea in campaign strategy circles. I’ve described it in this series as one of five claims that don’t hold up to the same scrutiny as their reputations. This is the full evidentiary case behind that claim including why the debate around it is structurally hard to settle with the evidence anyone currently has and what a real test would need to look like.

Does building a dedicated, resourced distributed-organizing department produce a measurable field-strategy advantage? And if the data doesn’t show one, is that because the tactic doesn’t work or because something else about the campaigns that adopt it explains the outcome instead?

Three 2020 Democratic campaigns built dedicated, resourced distributed-organizing departments, real budget lines, headcount, and a seat at the strategic table rather than a volunteer-coordination function bolted onto the existing field program. Sanders’s national organizing director came in with a specific relational and distributed-organizing pedigree; Warren’s campaign built a widely documented “organize everywhere” model with its own dedicated staff; Yang’s campaign went furthest by building self-organizing local “Yang Gang” chapters that ran their own canvassing and events with loose central coordination rather than a paid organizing corps directing volunteers. This was a structural departure from every other campaign in the field. None of the three won the nomination. Biden, who did, barely built a distributed-organizing function at all, relying instead on a more conventional paid-organizer field model.

That’s the entire empirical record this debate currently runs on. Three positive cases (campaigns that invested) with a uniform negative outcome and one negative case (a campaign that didn’t invest) with a positive outcome. I want to be precise about what that pattern is and isn’t. It’s not a controlled comparison, the three investing campaigns and the one non-investing campaign differed from each other in dozens of ways that have nothing to do with field strategy. Resource tier, candidate profile, coalition composition, campaign launch timing, etc.. It’s a real, if small, natural pattern to take seriously rather than dismissing as noise and it’s also nowhere close to enough to settle the question either way.

The strongest of the three explanations below is the selection effect. If the campaigns that invest heavily in distributed organizing are disproportionately the same campaigns structurally unlikely to win for entirely separate reasons, the null result on distributed organizing is really a null result on being an insurgent campaign, with the field strategy riding along as a marker of campaign type rather than a cause of the outcome. Here’s the full case for that, alongside the two alternatives it has to be weighed against.

The tactic doesn’t work as advertised. The straightforward reading is that distributed organizing sounds appealing in theory but doesn’t convert into the coalition-building and turnout advantages its advocates claim, at least not at the scale needed to justify the department-level investment these three campaigns made.

Selection effect. The campaigns most likely to invest heavily in distributed organizing are, for reasons that have nothing to do with field strategy, disproportionately movement or insurgent campaigns already betting on grassroots enthusiasm as a broader theory of victory. That’s a type of campaign that is—for entirely separate structural reasons documented elsewhere in this research such as resource constraints, coalition ceiling, or a higher bar insurgent campaigns face for consolidating institutional support—statistically less likely to win a modern presidential primary regardless of its field program.

Implementation quality, not the concept itself. A third possibility this dataset can’t separate from the other two is that maybe distributed organizing works, but only when implemented with measurement discipline and integration into the broader campaign strategy. None of the three campaigns that tried it did so with that level of rigor. A poorly measured, loosely integrated distributed-organizing department would produce the same null result as an ineffective tactic and this dataset doesn’t have the granularity to tell those two apart.

The three variables—campaign type (insurgent vs. institutional), distributed-organizing investment, and nomination outcome—are confounded in exactly the way that makes causal inference nearly impossible with only three positive cases. To isolate the tactic’s actual effect, you’d need either an institutional, well-resourced campaign that tried distributed organizing seriously (a case this field didn’t produce) or a controlled sub-experiment within a single campaign. Neither exists in the current record.

I looked for it rather than assume its absence. The strongest case for the tactic shows up in down-ballot and issue-advocacy organizing where distributed models have a longer track record and some real, if contested, evidence of cost-effective volunteer mobilization at scale. That’s a different context, though, since down-ballot and issue campaigns typically don’t face the same compressed, high-stakes calendar or the same need to rapidly consolidate a fractured coalition that a presidential primary does. Whether a tactic that shows promise in a slower-moving, lower-stakes environment transfers to a presidential primary’s compressed timeline is exactly the kind of transfer question this series tries to be honest about elsewhere. Here, I don’t think the transfer is established either way.

If the 2028 field produces a campaign willing to run this as an actual experiment rather than an article of faith, here’s what would move this debate—real measurement infrastructure comparing distributed-model regions against conventionally staffed regions on the same underlying metrics (volunteer retention, voter contact rate, verified turnout lift where measurable), built in before the investment is made rather than reconstructed after the fact. Absent that, another campaign simply trying it again and winning or losing doesn’t add much. It’s just a fourth confounded data point to a debate that already has three.

I mark this as speculative and deliberately leave it there rather than resolved in either direction. The claim that distributed organizing “doesn’t work” is not supported by this evidence. Three confounded cases can’t support a negative claim that strong. The claim that it’s “a proven force multiplier,” which is the version that circulates as conventional wisdom in organizing circles, is even less supported. It requires treating three uncontrolled cases as if they were a positive trial when they’re actually a uniform negative outcome for the specific implementation each of the three campaigns happened to run.

What the evidence actually supports, narrowly, is that the case for committing budget and headcount to a standalone distributed-organizing department, on the strength of the current record, isn’t there yet. That’s a specific, falsifiable, and more modest claim than either “it works” or “it doesn’t” and it’s the one I’m willing to publish under my own name. Any campaign considering the investment should treat it as a real, live hypothesis worth testing with measurement discipline.