Inside Baseball: What the 2019 Convention Vote Data Tells Us About the DSA Membership

Two months after the 2019 National Convention, the election numbers for the National Political Committee (NPC) have been released to the membership. Though we already knew the results of the NPC election, the ballots are the first public data we have on the membership since the 2017 Convention. The Democratic Socialists of America (DSA) reports its total membership figures as it has grown, but we have relatively little information to help us objectively identify member and chapter trends.

Per the DSA Constitution, Article V Section 6, “No voting at National Conventions shall be secret.” The votes of convention delegates are recorded with the delegates’ name and chapter, and this information has since been compiled into a spreadsheet that was emailed to members on October 11th, 2019. The spreadsheet shows how each of the 1026 ballots were cast, specifically the way that each delegate ranked candidates on their ballots. We can pair the data with the information we have about each of the candidates, their affiliations or leanings, and from there trends and patterns start to surface.

In what follows, I’ll go through the voter data to share my observations and impressions and approach some of the questions about DSA’s political dynamics. Fair warning to the reader, this will include a good amount of organizational “inside baseball”. Luckily, I’ve noticed that DSA members, like baseball fans, often like the inside baseball. To keep this semi-readable, I’m going to assume the reader has some knowledge of the national caucuses and basically knows what happened at the 2019 Convention[i]. For the purposes of disclosure, I’m not a member of any caucus (you can look at my votes along with everyone else’s).

Caucuses, Revisited

In the articles I’ve written on DSA, I put forward a framework to give a general understanding of the internal politics of the organization. In sum, DSA tends to have agreement on political positions but disagreement on organization. While there are certainly political differences, DSA is animated by two poles of attraction: towards greater coordination or towards decentralization¹. Bread and Roses (B&R), Socialist Majority (SMC) and Collective Power Network (CPN) all weigh in on the coordination side, while Build and Libertarian Socialist Caucus (LSC) tend towards decentralization. Without this framework, it’s difficult to understand the underlying issues behind convention motions, maneuvers and arguments.

That said, it matters that these are separate formations. B&R has the most articulated theoretical perspectives on socialist strategy, a deep commitment to the Sanders campaign and what they call “class struggle elections,” and appear to be internally well-organized. CPN has specific perspectives on how to structure DSA with more regional organization, a focus on labor but they are also hostile towards the “Rank and File Strategy”, advanced by Bread & Roses. Socialist Majority formed largely as a reaction to B&R, an amalgam of tendencies whose main coherence is in wanting a stronger organization and not to be in B&R. SMC’s base includes former staff and organizational leaders with administrative objectives, electoral and labor activists who prefer DSA’s traditional approaches (allowing more support for Democrats and labor officials), and those who for whatever reason just don’t want to associate with B&R (even if they have similar positions).

LSC houses the organization’s anarchists, Wobblies and council communists, with the usual array of ultra-left politics and deep suspicion of leadership structures. Build is not as ideologically motivated as LSC but is committed to localist politics. In form, the two have a large cross section and so they act as de facto organizational partners. These distinctions are useful for understanding times when there are exceptions to the framework: for example, the vote on the “rank and file strategy” motion at convention crossed organizational lines and saw Libertarian Socialist Caucus voting with Bread and Roses on one side, and Collective Power and Build on the other with Socialist Majority splitting their votes.

These national formations are the most visible, but there are several smaller local formations that elected delegates to the convention. Most notably, the San Francisco Slate ran members of the Red Star caucus[ii], a revolutionary Marxist grouping, for NPC and won a seat for Jennifer Bolen. Before the vote data, the weight of these formations was pure speculation. Below we’ll look at what the voter data tells us about the local formations.

The NPC Vote and STV

While in the past, DSA Conventions have elected the leadership through a modified Borda Count, a motion made from the floor[iii] altered the election format to follow Scottish-style Single Transferrable Vote (STV). STV has become popular in DSA chapters, especially through the website Opavote, which can digitally distribute secure ballots and can calculate the results that would otherwise be rather burdensome to do manually. The move from Borda Count (a points-based system) to STV was significant in that it was arguably more proportional and it changed the tactical calculations delegates (and more specifically caucuses) would have to make to elect as many of their desired candidates as possible. To understand voting patterns, you have to understand how the voting system works.

Single Transferable Vote creates a threshold that must be met to “elect” a candidate, based off the number of offices and voters: (votes/number of seats + 1) + 1. Voters rank their ballots from first preference to last. STV then goes through rounds, redistributing (“transferring”) votes that either surpass the election threshold, or eliminating the lowest scoring candidate. Transferred votes are redistributed to next candidates in order of voters’ preference. Votes can be “exhausted” if the votes to be transferred don’t have a next ranked candidate to go to — if you only vote your first preference with no second preference and your candidate is eliminated, your vote that could have been transferred is now exhausted and goes to no other candidate. The first round is arguably the most important because it determines the candidates’ standing for the successive rounds, where the candidate with the fewest first-ranked votes is at risk of being knocked out of the election.

For the 2019 Convention, we have 1026 votes and sixteen (16) seats, so the threshold formula is (1026/16+1) +1 = 61.35, so 61 votes needed to be elected. Thirty-two (32) candidates ran for office, meaning you had a 50–50 chance of being seated. By comparison, the 2017 Convention had forty-one (41) candidates. The DSA Constitution requires a quota for the NPC according to gender and race[iv], but it did not come into play in this election as the candidates fielded met all the requirements upon election.

The National Office also provided the graphs from Opavote to show how STV transferred votes until all sixteen NPC seats had a candidate who met the 62-vote threshold, which took 32 rounds to satisfy. The progression is interesting, though there’s not all that much that I can say that isn’t evident by just looking at the Opavote Rounds.

Austin Gonzalez, favored by Build and LSC[v], was the front runner with 87 votes, well past the threshold in the first round. Blanca Estevez (CPN), Natalie Midiri (B&R) and Hannah Allison (SMC) were all also elected in the first round (in that order). Each caucus had a leading candidate, and as a rule transferred votes (from surplus votes or eliminated candidates) went to candidates’ other caucus members. LSC did not “elect” a candidate until Round 12 when Michelle Bruder was eliminated and her votes put Sauce over the line.

On the flip side, each caucus had a sacrificial lamb. With STV if a tendency has many candidates, first-rank votes can get spread too thin and penalize them, winning fewer seats relative to the overall support a tendency has. Each of the national caucuses deprioritized a candidate[vi] for the first round (intentional or not): B&R — Rachel Zibrat; SMC — Tim Zhu; LSC — Austin Smith; Build — Zac Echola and Ravi Ahmad. None of these candidates were elected, showing the value of a strong first round showing. It is a little bizarre that for all the stink about process, the Top 14 first place vote winners were elected; only Erika Paschold and Maikiko James were boosted through the transferred votes.

Tendencies were elected with the following number of seats:

Framing this in terms of the policy for the organization, you get Coordination (11)[vii] to Decentralization (5) with Red Star (1) not falling clearly on these lines.

Membership Data

This brings us to the more interesting part of the vote data, which is what the ballots tell us about the DSA membership. Though we’ll use the delegates (the only and best data we have) as a sample of the DSA membership at large, it is certainly not a perfect representation.

Each chapter had their own methods for selecting delegates, so a chapter’s delegation may be over-represented in a tendency depending on the politics of that chapter’s elections. Further, it is unwise to interpret the data when it comes to smaller chapters that have fewer than three (3) delegates to try and extrapolate meaning about that chapter, since the delegation as a sample is too small to reasonably infer that one or two people will represent the political positions of the 60–120 members in their chapter. At best, we can merge the small chapter delegations to see if that tells us anything about smaller chapters based off their size as a category, but we cannot reliably target any individual chapter to say that the entire chapter likely swings one way or the other. Lastly, At-Large delegates were elected by the entire pool of At-Large members and Organizing Committees, rather than locally, and might represent at-large members as a category but no geographic inferences can be made.


For the purpose of this analysis, we’re going to assume that candidates function as political representatives of their tendency first and as individual actors second. As discussed above, we already knew the outcome of the NPC election. Seats were filled by tendencies:

The problem is that looking at the data like this collapses a lot of the relationships into monolithic representations of caucus power. So how did the delegates actually vote? Were there simply large blocs of caucus voters mobilized for their candidates?

There were six slates, who ran an average of five (5) candidates each[viii]. If we look at the delegates’ top five ranks, did they fill their top choices with a majority of any single slate’s candidates? If a delegate ranked three[ix] or more candidates in their top five, I take this as evidence that they have a significant preference for that slate. When we look at that data, we find the following[x]:

Most voters preferred a slate: 82% ranked one faction’s slate as the majority of their top five rankings. Of the remaining 18%, 9% had mixed rankings for the “coordination” slates — a blend of B&R, SMC and CPN where no one tendency had a plurality of the ranks; 4% likewise had “decentralization” ranks that didn’t show preference for either Build or LSC. Only 2% didn’t conform to the paradigm; 1% (9 voters) placed all independents as their top 5, where “independent” isn’t a meaningful standalone category, and 1% (7 voters) had a mix of every tendency.

How does this hold up against the election results? Pretty well, actually. If we take these figures for preference and test against the first-round results, they’re within range. “Independent” doesn’t work as a category in itself, but their seat count starts to make sense when you move the uncommitted “coordination” votes to Sean Estelle and Dave Pinkham as the Independents, along with a few B&R and CPN votes.

More on the Candidates

At this point I think it’s worth revisiting the candidates-as-representatives framework. Looking at candidates as an extension of the tendency is useful at first but can’t account for everything. Its useful to go through some of the pertinent candidate information. Every candidate is more than just their tendency, but for the sake of understanding voter behavior we have to look at it in terms of what was broadly known. I won’t go through every candidate, which is not intended to dismiss the multitude of identities and dynamics, but for the sake of analysis I am highlighting information to help explain outcomes.

· Independents[xi]

o Dave Pinkham: Labor activist from Austin

o Sean Estelle: Ecosocialist from Chicago

o José Pérez: Immigration activist from Atlanta; anti-Bread and Roses[xii]

o Theresa Alt: Incumbent, long-time DSA member from Ithaca

· Build

o Zac Echola & Ravi Ahmad Haque: Incumbents, leaders of Build. Just before convention, other NPC members wrote public articles accusing Zac and Ravi of problematic conduct on the national leadership, especially surrounding RL Stephens amidst rape allegations

o Tawny Tidwell: Moved from Texas to New York, also a member of Emerge in New York.

o Austin Gonzalez: Anti-fascist organizer from Richmond, Virginia.

· Socialist Majority

o Hannah Allison: Former DSA staff organizer based in Kansas.

o Russell Weiss-Irwin: Labor activist from Boston, former NPC member. Just before convention, a Boston member released a public statement asking delegates not to vote for Russell because of his response to a sexual harassment case.

Going through these profiles helps to explain some of the data that can’t be explained by caucus alone. As noted above, all the national slates deprioritized a candidate, and understanding the candidates may help explain who that was and why. It seems clear that Build moved their votes away from Zac and Ravi, and such a hard pull to me suggests that the allegations were an important part of deciding to favor their other candidates first. There are a large section of votes (23) from New York City that break the usual bloc pattern and favor both Tawny Tidwell and Red Star, which would likely be the Emerge group — Tawny is a member of Emerge, and Red Star is the only revolutionary Marxist grouping that presented candidates for the NPC, likely winning Emerge’s votes as another revolutionary current.

I suspect B&R expected to rank ballots according to the Borda Count, and when it switched to STV they decided to prioritize Marianela, Natalie and Megan to ensure B&R representation on the NPC. Why they picked those candidates and not Rachel or Marsha, isn’t clear — all five B&R candidates were consistently ranked Top 5 among B&R voters, so it seems likely this is more of a tactical approach to the vote. One criticism coming out of convention was that Natalie Midiri unfairly benefited in the election, as she was one of the convention chairs. Of the 70 first choice votes, 65 of those (93%) came from Bread and Roses voters; the remaining 5 votes were SMC or No Slate-Coordination voters; this suggests to me that her role as chair played no significant role in convincing voters to support her.

Despite controversy around Russell, Socialist Majority voters stuck with him. Of the 205 SMC voters, 25 ranked him as their #1 preference, so 86% of Russell’s first place votes were from within the SMC bloc, including those from Boston.

For the two independent candidates elected, were they simply caucus-adjacent? Admittedly, I first considered Sean and Dave as unofficial B&R candidates since they were endorsed by that caucus. The data suggests otherwise. Of Sean’s 48 first round votes, only 16 came from B&R-aligned voters; 5 came from SMC voters; close to half of Sean’s votes came from voters who did not support any one tendency primarily but supported the Coordination candidates.

Dave’s votes are similar. Of the 57 first round votes, 34 were No Slate-Coordination voters, with only 8 B&R voters and 6 SMC voters. The 34 votes are interesting, because they nearly all come from Phoenix, Philadelphia and New York City, and all of them had essentially identical preferences for candidates down ballot. These voters are likely the group that was purged from the Spring caucus at the beginning of 2019 before Spring rebranded as Bread and Roses. That group of voters all heavily favor any labor candidates (Pinkham, Niemeijer, Weiss-Irwin), splitting support among the coordination-aligned slates but not siding with any one over the other (and potentially punishing B&R for expelling them.)

I would therefore argue that voters who favored Sean and Dave were those who generally wanted DSA to move in the direction of a stronger organization, but did not have preference for one faction over another — whether these votes are from voters critical of caucusing or are themselves a caucus-in-waiting, I couldn’t say. Together, these account for the majority (82) of the 86 No Slate-Coordination votes.

Voting Blocs and What They Mean

While there are undoubtedly solid blocs of caucus voters, there are a significant number of voters who, even if they have a preference, are not committed to a political tendency as such. If we look at delegate votes as an expression of their political leanings rather than as fidelity to a caucus and all its contours, how could we define political leanings in DSA? These are my interpretations.

First, there is a definite geography to DSA’s internal politics. This makes sense given how large the United States is, so politics develop in pockets and for a tendency to develop it needs to be transmitted through agents who will build and recruit to it. By tendency:

· Collective Power Network: Los Angeles, Denver, Ohio (Cincinnati, Dayton, Columbus), Washington DC, Miami, New Orleans, Sacramento.

· Socialist Majority: Boston, Denver, New York, Kansas City, Los Angeles, St. Louis, Minneapolis-St. Paul, Tacoma

· Bread and Roses: New York, Chicago-area (including northern Indiana), East Bay, Detroit, Baltimore, Los Angeles, Washington State (Seattle, Olympia, Everett), Austin

· Build: Cleveland, Pittsburgh, Richmond, Philadelphia, Michigan (Lansing, suburban Detroit, Boston, Silicon Valley

· Libertarian Socialist Caucus: Boulder, Eugene, DC, Orange County, Portland, Silicon Valley, Buffalo

Build and LSC have overlap, but LSC is located more on the West Coast (Oregon, California and Colorado), where Build is more in the Rust Belt. Bread and Roses are often in big industrial areas, but not exclusively. Collective Power finds its base along the southeast (DC, Miami, New Orleans), in LA and Ohio. Socialist Majority looks to have built its support more heavily around their candidates than other tendencies: New York (Abdullah), LA (Maikiko), Kansas and the lower Midwest (Hannah), Boston (Russell). Texas split between B&R (Austin), SMC (North Texas) and Dave Pinkham/labor (Austin). Of course, nearly every candidate took their home areas, but some tendencies were able to reach farther away than others. Chapters on the Coordination pole are either B&R or SMC chapters; except for very big chapters like New York and Los Angeles, if one caucus is present it generally dominates and there are few members of the other one.

Looking farther down ballot, we can see the caucus relationships to one another. Bread and Roses typically ranked their slate in the Top Five, followed by Sean Estelle, then Dave Pinkham, and either SMC or CPN (usually SMC first, but not always). No voters who voted all B&R also included all the LSC or Build candidates.

Collective Power is consistent: they always vote CPN first, SMC second independents or Red Star and then B&R last.

Socialist Majority voters vary more depending which chapter they come from. Unlike B&R or CPN, SMC delegates vote their slate first but the order depends on their chapter — New York all ranked Abdullah first, LA: Maikiko, Boston: Russell, etc. Where most of the other caucuses are more ideological, this pattern makes SMC look much more like a convention vehicle that’s an alliance of chapter leaders. Very few SMC voters included all the B&R candidates down ballot; they often included some but interspersed with CPN and independents. Some SMC voters include no B&R votes at all.

This should give you an idea of how the caucuses on the Coordination spectrum relate to one another. CPN voters look to be reluctant allies of B&R, distancing themselves from the latter caucus as much as possible while guarding against Build and LSC. Bread and Roses is more open, but not much more. They are more willing to support CPN, but still prefer SMC. SMC as an electoral coalition doesn’t have a consistent preference as a caucus, but instead they decide by chapter who they support after their own slate.

One question that’s come up often is what do the “small chapters” think? It’s often been argued that the “small” chapters are more sympathetic to Build or LSC because of their localism. Of the 59 At-Large votes, 28 were “decentralization” votes and 26 were “coordination” votes. That’s a higher swing for decentralization (47%) than the organizational average (31%), but at best that means At-Large members and Organizing Committees are still split between the two poles; they don’t favor one more than the other. Of the 151 “Small chapter” delegates (those with three or fewer allotted delegates) the same is true: 46% decentralization to 49% coordination. This would suggest that chapter size in and of itself does not determine political tendency in DSA. (We can officially stop with the essentialist arguments about size and get to discussing politics.)

Bringing It Together: Political Currents in DSA

By now it should be clear that this convention cannot be viewed simply as a caucus slugfest. Beyond a slate’s immediate base, candidates drew support from uncommitted delegates because of who they were, what their work represented and where they were from. Only by understanding the combination of these factors are we able to make sense of why things happened as they did. The challenge is decoding what it is about each candidate and slate that made it appealing.

I’ve used the framework of “coordination” vs “decentralization” as central to much of my analysis. We can observe a pattern of preference for candidates who support greater coordination and a strong national organization (B&R, SMC, CPN) and likewise a pattern of those who support greater decentralization and localism (Build, LSC). Beyond that, why pick one slate over another? LSC and Build have enough political distinctions that its easy to parse out those who are for an organizational decentralization versus a political anarchism or ultraleftism.

For the coordination tendencies, there are nuanced differences and the reasons for affinity with one group over another is more difficult to figure out. At times, affinity is simply personal: you know the people, so you vote for them. There’s a geographic pattern, where candidates obviously draw support from their home chapters and those nearby as their networks can help establish the political relationships. Anecdotally, at convention I asked why someone had joined the caucus they were in, their response was, “My friend asked me to.” There can be a danger in overemphasizing the political differences, where we see these as fixed instead of the product of relationship-building.

Still, the politics matter. Labor is a core difference between Collective Power and Bread and Roses: their opposing views on the Rank and File Strategy draw a line in the sand between them (seemingly in the shape of SMC). History between groups has severed some relationships, where we can see the purged Spring members voting together and not in support of Bread and Roses. The politics of caucusing concerns enough voters that some appear to have deliberately split votes so as not to ally with one group or the other. The election of two independent candidates is testament to this. There are many “labor voters” who are not particularly interested in one group or another but voted primarily for candidates who were labor activists across the spectrum: Dave Pinkham, Russell Weiss-Irwin, Marsha Niemeijer were their clear firsts, despite coming from different currents.

An interesting phenomenon I hadn’t expected is the prevalence of the “far left” vote. The San Francisco slate didn’t make much sense to me at the beginning of convention, but looking at the data you can see that Red Star actually rallied a substantial number of delegates (10%) who are not themselves a national formation, but respond positively when a revolutionary current presents candidates for DSA’s leadership. Jenbo’s election was due in large part to local groups such as the Red Caucus in Portland-area, Emerge in NYC, Red Star in San Francisco, communists in Denver, Bellingham, Fresno, Santa Cruz and Tucson. In the past, this is the space that probably would have been occupied by the Refoundation Caucus, which collapsed in 2018.

If it isn’t painfully obvious, the point I’ve been trying to make is that we should resist the temptation to look for easy answers. Caucuses alone, geography alone, personality alone cannot explain the dynamics of the organization and we should be prepared to look at how these elements interact. There are political questions at stake here: we have to be prepared to understand the experiences of our membership as a dynamic interaction with forces in motion. These questions aren’t static, and our politics shouldn’t be either.

Corrections (10/15/19): I’ve corrected two errors in this article. 1) Austin Gonzalez is not a member of LSC. 2) Bellingham DSA leans towards Red Star rather than Bread and Roses — I mixed up Whatcom and Snohomish counties in Washington.

Addition (10/19/19): I included remarks about Socialist Majority and the limitations of the coordination framework.

[i]I’ve written a number of articles about DSA and the Convention this year if you want to know more.
DSA’s Growing Pains on the state of DSA at the start of 2019:
DSA 2019 Convention Breakdown on the proposals submitted for convention and what they mean:
2019 Convention: NPC Candidates on who the candidates for National Political Committee are and who they are associated with:
If Sanders Should Lose — An article arguing for why DSA should not support other Democrats in the 2020 Presidential Elections:

Notes from the DSA 2019 Convention — rough notes on the proceedings of the convention from the floor:

DSA is Leading the Charge: On the 2019 Convention written for Jacobin about what happened at the convention:

Moving Targets: DSA’s 2019 Convention on the convention’s successes and failures, data from delegates before the convention and questions for the future:

[1] The use of words “coordination” and “decentralization” is limited in that those interested in decentralization see it as a different method of coordinating activity. The use of terms is imperfect, but was chosen to avoid known pejoratives on either side.

[ii] Jennifer Bolen and Jen Snyder are members of Red Star, though Darby Thomas is not.

[iii] This motion occurred while the convention rules were still being negotiated by delegates. Technically the convention did not begin until after the rules were adopted.

[iv] DSA Constitution, Article VIII Section 2:“Of the elected members, no more than eight shall be men and at least five shall be racial or national minority members of DSA”.

[v] Austin Gonzalez and Sauce are both dual members of Build and Libertarian Socialist Caucus (LSC). Looking at the voter preferences for those who listed Austin and Sauce first, Austin appears to be favored more by Build voters and Sauce by LSC. For that reason, I’ve decided to associate Austin with Build and Sauce with LSC, though there is overlap.

[vi] The exception here is CPN, who only officially fielded one candidate: Blanca Estevez.

[vii] I’ve been asked to include the YDSA vote, though they technically had their own convention. YDSA elected two co-chairs who share a single vote (though they can split it). Both are from the “Militancy” slate, which also favored coordination.

[viii] B&R ran 5, CPN 2, SMC 6, Build 5, LSC 6, with 4 independents. Mean of 4.57, median: 5.

[ix] For CPN, I lowered this number to 2, since they only have two candidates. Voting for both should be evidence that they are a CPN voter.

[x] Again, CPN complicates matters somewhat because they only have two candidates. 51 voters ranked both CPN candidates as well as three SMC candidates in their top 5. The data shows that they placed CPN above SMC, so I’ve chosen to count these as CPN voters rather than SMC.

[xi] I’ve chosen to list Daniel Merrill as CPN. Daniel ran as an independent but joined CPN immediately following convention. At first I thought that it would make more sense to keep her as independent, but based off of the voter behavior it seems clear that many voters identified her with CPN during the vote, so keeping her as independent would actually be more confusing.

[xii] Departing the narrative, Jose’s vote share almost all comes from his home city of Atlanta (10 of 13). Of his 13 votes, only one delegate was a B&R voter, which follows the notion that Jose ran largely as against B&R.

Socialist and labor activist in Wisconsin.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store