FIELD WORK:
An Experiment on Striking STEM Research
at Dartmouth College
David Freeman, Jake Willard, Logan Mann, and Rendi Rogers
23 November 2025
The Dartmouth graduate worker strike, launched on May 1st, 2024, was a direct outgrowth of the years of struggle and militant organizing by the workers who now comprise the long-haul tendency. Our strategy and entire framework was structured by lessons learned from those who came before us, and our eagerness to learn and apply these lessons was perhaps our greatest strength. With this in mind, we aim here to analyze how our implementation of these ideas unfolded in the context of Dartmouth’s particular social and organizational structure.
Background
Following the 2016 Columbia decision permitting private-sector unionization for grad workers, Dartmouth College was the only Ivy Plus campus on which grad workers were not organizing. This changed in 2020, as the entrenched local landlord class (including the college itself) capitalized on the flight of wealthy urban dwellers from New York and Boston to the idyllic northern New England vacationland during the pandemic. In fall 2021, as rents surged and the Dartmouth workforce faced new and intensified forms of immiseration, graduate workers began organizing a union campaign.
Despite its (mostly false) reputation as a small liberal arts college, Dartmouth’s graduate student population mostly consists of STEM researchers, with a different orientation and relationship to academic work than the cadres of humanities and social science grads that largely spearheaded organizing drives and strikes in the post-2016 era. Those previous efforts typically relied on the leverage of teaching assistants and course instructors to extract gains from the universities by disrupting the production of grades and degrees and the smooth transition from one term to the next. As we began developing our campaign, more experienced organizers warned us that grad workers in the sciences possessed politics and personalities that made them difficult to organize. But in our experience, the biggest organizing challenges came from the uniquely exploitative working conditions of the hard sciences and engineering fields.
Researchers typically work under a single principal investigator (PI) for the duration of their program. Most of the PI’s time goes into writing grants to acquire funding for their labs, mostly through federal agencies like the National Science Foundation (NSF) and National Institutes of Health (NIH). The PI then acts as a manager of their small fiefdom. In the absence of a union (and sometimes despite the presence of one), the PI holds unlimited power to determine hours, working conditions, and the production of research; and this apprenticeship model makes the advisor-advisee relationship near totalizing. Challenges to that relationship can end a prospective scientist’s career. Letters of recommendation and academic networking are extremely important for career progression. PIs have sole authority to decide whether the quality, content, and volume of a researcher’s work justifies progress towards their degree, and can fire grad workers arbitrarily over petty disputes, by issuing failing grades for the research course, framing issues of worker discipline as matters of academic freedom.
Further, the idea that research is a self-interested intellectual pursuit rather than labor performed for an employer is deeply ingrained in academic culture, despite the eye-watering returns that research grants generate. Universities often use this narrative to undermine attempts to organize researchers, as evidenced by the 2023 NLRB ruling that MIT graduate research fellows are not considered employees under the NLRA. Dartmouth cited this decision during our own recognition election when they (unsuccessfully) attempted to challenge the “worker” status of over 400 Dartmouth grads one week before the vote.
These conditions of work present new opportunities for strike leverage. Our teaching load was uneven across departments, but relatively light overall. We felt that even a very strong teaching strike alone would be insufficient to win us many of our demands at Dartmouth, since it accounted for such a small fraction of the total labor of our unit, as well as a relatively small amount of the total teaching labor at the College. By contrast, graduate workers supply the vast majority of the labor input into the research enterprise. In labs and departments where the research strike was effective, withdrawing our labor could halt production, and the damage would increase over time as grant timelines drew out and funding expired without results. So figuring out how to do an effective research strike became integral to our strategy. Our approach to power analysis naturally arose from these conditions.
The Sources of Our Leverage
Despite the relatively low teaching load, the impact of a given teaching job being struck was easier to quantify than that of a research job. The impact of the teaching strike was immediate and noticeable. We could clearly see class sections being forced to combine, faculty being made to take on additional teaching duties, and tutoring sessions ceasing to run for many classes.
Despite the significant impact of the teaching strike, the administration of most departments seemed able to adapt to the strike conditions and continue to deliver undergraduate courses, albeit in a haphazard and chaotic way. In the mathematics department, for example, we saw that the vast majority of our struck teaching labor was scabbed by faculty. While this pressure on the faculty contributed to our strike power by forcing them to spend time doing work that wasn’t initially theirs, faculty scabbing meant that the strike fell short of interfering with a vital product of the college, the production of degrees.
The research strike proved more difficult to organize and more difficult to analyze (both during and after the strike), but also seemed to cut deeper where it was working well. Research tasks were much harder to scab than teaching tasks. Random faculty could be tasked with teaching an undergraduate course for a few months, but could not replace the role of graduate students in the lab. Graduate workers often had very specific knowledge of how a given experiment was set up and how to continue the research process. Oftentimes even the advisors overseeing that experiment were not directly involved with the tasks assigned to their graduate worker apprentices, and could not simply pick up their work without a large overhead of time and effort.
It is difficult to get a concrete read on the impact of individual research strikers. Much of the finances of private universities are shrouded in secrecy. Apart from trying to use contextual information, such as grant deadlines, to guess the effect a given lab’s striking was having on their immediate management, we had to use information like the reaction of individual faculty advisors to gauge our impact. The research strike appeared to press a deeper nerve than the teaching strike. While the faculty seemed annoyed at the teaching strike, the cessation of research labor angered them. Despite the threat of a strike being public knowledge for weeks prior to its launch, many faculty advisors were particularly upset and even aggressive upon finding out that their students were planning to strike their research. Whereas a month or two of research stoppage could be worked around in the long term, the prospect of an indefinite continuation of the strike caused some faculty to begin openly fearing about meeting grant deadlines. While faculty were often tight-lipped about their concerns during the strike, the handful of instances we found of faculty despairing over their grants gave us good reason to believe our strike was working as intended.
This anger manifested in a schoolwide propaganda effort led by Dartmouth administration to loudly proclaim their absolute right to punish research strikers in any way they see fit, declaring it an “academic” matter rather than an “employment” one and thus outside the scope of the NLRA. Numerous faculty and admin stated that their “right” to have total decisionmaking power over “academic” progress was, for them, the most important issue at hand.
Ultimately, we believe the two-pronged nature of our strike was necessary for its success. Without the teaching strike, it would have been difficult to maintain momentum and cause sufficient immediate disruption to maintain the strike over an extended time period. It also possibly produced concessions in the short term that helped build our momentum. Without the research strike, however, the departmental administrations could have effectively scabbed enough of our labor to make our strike insufficiently disruptive going into the longer term struggle.
Research Strike Planning as Direct Worker Inquiry
Researchers, like all workers, can build power by organizing themselves into a strike-capable collective body. This is what we took to be the premise of our union at every stage of its development. The militant organizer core of our union was never shy about the eventuality of a strike in conversations with coworkers from the very beginning. However, it was not until shortly prior to the strike itself that workers themselves determined if or how a research strike could be executed. Instead, much of our efforts focused on building relationships and developing organizers in as many labs and departments as possible, such that most workers could be reached by an organizer that they know or with whom they share a workplace. Organizers learned to view the research strike as an unsolved problem that needed to be solved by us rather than a ready-made play-book to sell our members on. In this way, the strike strategy came together relatively quickly and at the 11th hour, much to the dismay of union staff and officials in the UE international.
Rather than presenting a ready-made, objective theory of power to workers, organizers operated under two basic premises. First, universities are very resilient to short strikes. If we want to threaten our employers with real pain, the strike we prepare for needs to be indefinite in nature and most likely, quite long. This means that workers must assume a posture where it is actually possible for them to strike for the long haul, without incurring undue damage or setbacks to their own research. Second, the economics of academia are such that leverage over the university is not uniformly distributed across campus. While some researchers could potentially cause the university to lose millions of dollars if their lab were to shut down for a prolonged period, other researchers’ refusal to work may have little to no financial impact.
The strategic implications of these two insights flew directly in the face of conventional wisdom. Strikes are supposed to create immediate crises for the employer, and the potency of the strike threat is supposed to increase linearly with the number of members who pledge to participate in it. Seeing that neither of these pieces of wisdom held for us, our organizers learned to adapt on a department-by-department and lab-by-lab level. Thus, we developed an exercise in worker inquiry that we referred to as a “group strike plan.” Organizers were given explicit instructions to hold a “strike planning meeting” with a relatively small group of close coworkers who shared a lab, research group, or department. Our goal for these meetings was for workers to derive a theory of power for themselves and then decide collectively to act on that theory.
To begin developing a group strike plan, workers were asked what specific type of work they do: whether they are writing a paper, writing a grant proposal, doing experiments, preparing for conferences, etc. They were asked how their work is financially supported, and whether the work they do furthers scientific objectives of their PI beyond what is necessary for their own actual thesis. They were asked what kind of work they do outside of research itself, such as cleaning chores in the lab, or responsibilities to department events. With these questions, the mechanics of the complicated machine exploiting them started to become clearer, and a real conversation about throwing wrenches into it could be had.
Conversations about action inevitably provoked questions of risk. What are people potentially giving up by agreeing to any given form of collective action? Organizers in most contexts tend to rely on one-on-one conversations with workers aimed at connecting with their issues and letting them talk themselves into doing things. However, when workers express fears in this context, there is not much the organizer can do except assure them that “the union” will have their back, with little in the way of a concrete commitment to supporting the worker if something were to happen. Strike planning meetings were spaces where workers were highly prone to catastrophizing, but when they did so, they were catastrophizing to their immediate coworkers, the very people who are in a position to make concrete commitments of support. In the lead up to the strike, these department meetings replaced one-on-ones as our primary form of communication. They were the spaces where workers decided to take action together with the people they trusted most, and where they expressed anxieties to the people who could understand them the most. In retrospect, this single detail stands out as the most decisive feature of our practice, without which the strike likely would not have taken place at all, or had much endurance.
Organizers went into these meetings with an open mind regarding what specific actions it is reasonable to expect the workers to do. Having no “master theory” to fall back on, no way to “check their work,” we often had no choice but to take the workers’ word for it that their theories of power were correct. As a result, workers decided in a rather decentralized way where the “picket line” would be drawn – e.g, whether it is allowed for a mathematician to continue thinking about a math problem in their head, or if the experimentalist should throw away years of research by allowing their rats to die.
| Microbiology Lab – 6 grad workers | Math Department – 33 grad workers |
| Activities to stop during strike: – Lab chores – Bench work – Presence in lab – Presenting data to PI, collaborators, or department (exemption for all of the above when directly related to qualifying exam defense – applies to 1 grad worker, defense scheduled for 2nd week of strike) – Lab meeting attendance – Seminar attendance – Use maximum allowed journal club absences | Activities to stop during strike: – Presence in department building/grad offices (exemption for class attendance) – Teaching duties (includes teaching assistants, graders, tutors, and instructors of record) – Contact with faculty advisor – Submissions for publications, conferences, or grants – Attending lab/group meetings – Department chores (includes hosting socials and seminars) |
| Activities to continue during strike: – Literature review – Data analysis – Writing for qualifying exam or thesis (all of the above to be done remote ONLY) – Course attendance and coursework | Activities to continue during strike: – Course attendance and coursework – Qualifying exams – Dissertation defense – Progress towards dissertation cannot be tied to an external grant and cannot be communicated with faculty |
Throughout the strike, we held weekly general body meetings in which members would vote to either continue or stop the strike. At these meetings, organizers reported on strike participation in each department, and the bargaining committee outlined which contract demands we won or which outstanding demands members had decided to maintain or concede during the previous bargaining session. The discussions at general meetings and during bargaining caucus sessions were often highly contentious, emotional, and difficult to facilitate. Workers fervently argued to continue or end the strike, usually aligning with others in their department. Despite the uneasy nature of the discussions themselves, workers nearly reached consensus whenever the time came to make a concrete decision. Furthermore, those who vehemently argued for ending the strike during meetings largely abided by the group’s decision, refusing to cross the picket line whenever the general body voted to continue.
For eight weeks, an overwhelming majority of workers chose to remain on strike. Strike participation held steady, and we gradually won contract demands. At the beginning of the ninth week, organizers began to report a decline in strike participation. A small contingent of workers argued to continue the strike in pursuit of back pay and other demands left unwon. However, with our strike fund running out and the summer term approaching, workers overwhelmingly decided to accept the offer on the table and go back to work. We soon ratified a new contract that included a 17.5% raise with annual cost of living adjustments to match or exceed inflation, free dental insurance, reimbursement funds for childcare, medical expenses, and visa renewal fees, and innovative just-cause protections for teaching assistants and researchers. After the dust settled, we found that when reflecting on the strike in groups, workers generally agreed that the strike was successful, it was worth it, and we would do it again.
At its best, the strike was a product of militant workers earnestly trying to hurt the university as much as they possibly could without insisting upon immaterial and purely symbolic commitments of self-sacrifice. At its worst, those clinging to the view that they hold no leverage over the university (and no responsibility to anyone else) found a path to legitimizing their position on the wrong side of the picket line. These paths diverged along departmental lines. Participation in our strike had one glaring predictive factor, and it was not attending a rally or signing a strike pledge: workers who participated earnestly in this process belonged to departments that had an established culture of practicing collective action on the shop floor, even at small scales.
Power Mapping and Departmental Strategy
Going into our strike, our strategy involved identifying labs of particular leverage and targeting them for strike participation. In practice, however, many of the labs we identified as particularly strategic were also the most difficult to engage in strike participation.
Many of the labs in large departments such as Engineering, Computer Science, and Cellular/Molecular Biology departments brought in the largest volume of grant revenue, due to their direct healthcare, military, or industrial applications, with some labs even producing profit-generating patents. Departments such as mathematics, Earth sciences, and physics tended to have smaller average grant revenues due to their tendency towards “fundamental” research as opposed to marketable applications.
This underlying material reality played a role in the departmental cultures with respect to striking research. Many of the graduate workers in the large, profit-generating, departments seemed to have the attitude that their future career prospects (including high-paying engineering jobs after graduating) were worth so much more than whatever raise we may win during a strike, or that upsetting their advisor or slowing their own research output was not worth it. Appeals to solidarity and communal support tended to have less influence there.
This made the targeting of “high-impact” labs a tougher proposition than we imagined. The departments which were more engaged in the project of the strike (such as mathematics and physics) tended to be smaller and more communal than departments like engineering, so the “targeting” strategy was unnecessary in those departments since we instead focused on engaging the entire department in the strike. In this way, targeting high impact labs fell to the wayside, in favor of driving entire departments to strike.
Where the strategy of understanding leverage did play an important role, however, was at the smaller, individual lab scale. In discussions with individual researchers or lab groups, we always put forward the question “what is your leverage over the college?” as the fundamental basis of forming a strike plan. Understanding strike plans in this way helped workers identify the forms of labor necessary for them to strike if they wanted to have a meaningful impact. This formulation, and our collective discussions about our leverage, played an important role in both driving workers to strike important labor, and in getting the general rank-and-file to internalize the acute pressure they were putting on the college even when obvious indications of the research strike impact were hard to find.
The Administration’s Strategy and Our Response
The Dartmouth administration’s strategy towards breaking our strike, apart from the standard strategy of cutting our pay, centered on heavily implied threats through open denial of our right to strike. While faculty were particularly upset at the research strike, Dartmouth administration extended its vociferous denial of our strike’s protected status to include the striking of teaching labor as well.
This denial took the convoluted form of Dartmouth constantly expressing that faculty have the absolute right to fail graduate workers and kick them out of their program if they are making insufficient “academic” progress (protection against such arbitrary dismissal was, in fact, an important demand in the strike itself). Both teaching and research positions are formally associated with a “grade,” which functions in practice simply as a ledger to indicate a given worker has completed their assigned job. Dartmouth leaned on this to argue that their right to assign grades is uninhibited by the NLRA, and that withholding teaching or research labor is grounds to fail graduate workers. Since failure leads to expulsion from the program, they defended their right to fire anyone they desire. These threats were repeated often in bargaining sessions, in Dartmouth’s mass communications, and by faculty loyal to the college.
The union’s response to these threats was narratively straightforward. We constantly emphasized to our members, in general body meetings (GBMs) and on the picket line, our inviolable right to strike our labor. This had been a major point of discussion in GBMs leading up to the strike as well, where we collectively discussed that perhaps the most important precondition for ending the strike – beyond any contractual win – was our refusal to allow Dartmouth to punish anyone for engaging in strike activity. The organizing we did around Dartmouth’s attempt to flagrantly violate one of our most important rights as workers brought intense momentum to our fight against retaliation, and fired up people who were even somewhat passive when it came to the contract fight itself.
To turn this strongly held collective belief into actual power we could wield against the College, however, required more effort. At the strike’s beginning, we had to organize workers to stand firm against the deluge of messages from their respective faculty advisors telling them that striking their research is liable to discipline. We believe that organizing members to collectively refuse faculty aggression pre-strike helped prevent individuals from being targeted for discipline on a large scale early on.
There were nonetheless some attempts by Dartmouth management to ramp up strike retaliation. Our union’s strategy was to respond swiftly and intensely to the first signs of such retaliation, in order to nip them in the bud and prevent strike discipline from becoming accepted. The most stark example of this occurred in the math department, when an international worker was threatened by a faculty member with effective termination were she to refuse to perform her teaching duties the first week of the strike. The math department workers responded by packing the department admin office with a majority of math workers that same day and demanding protections for us all from retaliatory discipline. Taking early, decisive action such as this wherever signs of retaliation cropped up appeared effective at sending a strong message back to the faculty and administration schoolwide. Despite Dartmouth’s initial heavy-handed declaration of their absolute right to discipline anyone, it was clear that their confidence was shaken as soon as workers began fighting back. After these early struggles, the administration continued with similar rhetoric. However, no attempts at serious discipline were made, and our union considered our battle over retaliation to have been won.
However, as the strike went on for the long haul, direct pressure from PIs decreased. Their anger turned into a more panicked concern as the strike continued into the summer term. It became clear they had little interest in dismissing high skilled workers, whom they had invested years of their time and funding into training. Their interest was primarily in a resolution to the strike, on terms either favorable or unfavorable to us. Worker-to-worker solidarity this was not. Rather, we found the faculty to be a disorganized middle strata, which would side against whoever they found to be the weaker party, exerting their influence either downward in an attempt to break the strike, or upward to management to press them towards an agreement. Demonstrating our power to stop production earned, in some cases, their begrudging respect, which endured in our working relationships beyond the strike.
Reflections on Effective Striking in Our Conditions
In retrospect, it is clear there are many granular strategic decisions that could have been made differently to improve our strength, both during and in the lead-up to the strike. We believe that the most fundamental point underlying both our successes and failures, however, relates to the degree to which we centered strike preparation in our organizing, and specifically when we correctly identified the correct scale (lab, departmental, schoolwide, etc.) at which to organize collective actions.
The series of recent strikes in higher ed continues to reveal that departmental organizing is a vitally important component of building a strike. Many grad organizers, ourselves included, have experienced a school-wide strike as being best understood as a sort of confederation of departmental strikes. This framework is helpful in part because it emphasizes building socially coherent and independently functional subgroups of the total unit which can then act all together, rather than haphazardly attempting to bring together disparate individuals.
This lens can be extended by adapting the scale of what we consider the basic components of a graduate worker strike to the particular conditions present on the ground. There are times when we think an overly dogmatic attempt to fit our organizing approach to a given scale slowed our progress. This took forms such as trying to force the formation of a departmental structure before sufficient groundwork had been laid (specifically in larger departments) or in trying to create campus-wide movements when smaller-scale, departmental organizing would have built more enduring structures. Better understanding the natural social contours of our unit, and when to focus on activating which groups, would have improved our strike preparation.
In the course of preparing for our strike, we think this played a role in the case of the large departments, namely the engineering and MCB departments. For months, we struggled to build coherent “departmental unions” within these departments, instead having a number of isolated individuals who participated in the campus-wide union. As the strike drew near, organizers in engineering found that certain groups of labs, who all happened to share a given sub-discipline of study, were much easier to cohere into an organized “unit” than was the whole department. Shifting towards lab focused organizing in MCB in the strike lead-up also led to success in bringing in more prepared strikers.
While shifting towards this approach where necessary helped us prepare some workers to strike in difficult departments, our approach to forming these smaller coherent groups occurred only in the last few months before the strike began. Even in smaller, stronger departments like Math and Physics, truly cohering the workers into a powerful and autonomous decisionmaking structure happened very late in the timeline of building the strike.
This lag arose from the other significant mistake we made in the strike preparation phase. During bargaining, we put a good deal of effort into organizing what we refer to now as “direct actions” – particularly pickets or protests targeting certain administrators over specific contract demands. This was in addition to departmental organizing, but nonetheless took a significant amount of our limited effort and time. We invested time in such actions essentially believing that we could help build participation in our union by engaging people in these direct actions, and eventually use the expanded structure we built to launch an even stronger strike.
We found that these direct actions did very little to actually expand our union’s base of participation. While they may have helped us gain some contract concessions, they were a tremendous expenditure of energy. People who joined these actions in general did not then join in as core union organizers, but instead seemed drained. The clearest example of this came when we organized a large group to infiltrate an alumni dinner event and chant at the College president over her unwillingness to give us meaningful sexual harassment/discrimination protections. While yelling at the president in front of Dartmouth alumni who were trying to eat dinner was fun (and funny), and may have even contributed to their concessions on those matters, those newer workers who came along did not seem invigorated, but rather tired out by participating in it, and doing it created no enduring social structures.
It wasn’t until four–five months before the strike that we began seriously organizing the infrastructure necessary to meaningfully strike at all. Throughout our entire union organizing effort, we would consistently speak to people about the idea of striking – why we would do it, how it would work, etc. We realized, however, that we simply did not have our departments well enough organized structurally to carry out something much bigger than a large rally. When we switched our energy towards the deliberate work of preparing individual lab groups and departments for actually striking, and focused our GBMs on preparing for the reality of a strike, our union’s strength grew rapidly. We also began to see more clearly what organizing methods worked well in what departmental conditions.
Conclusion
We can view the Dartmouth strike as a case study for understanding and improving the application of the research strike, and of organizing STEM workers more generally. The research and the teaching components of the strike had an impact on different time horizons, the synthesis of which increased the damage dealt to the college and the social momentum of the union. The faculty in particular seemed affected strongly by the research strike, and the college administration postured itself strongly against our right to strike research at all and indicated a willingness to punish those who engaged in it. It seemed that only quick and aggressive pressure from departmental organizing prevented them from carrying out serious retaliation.
Major questions still remain, however. The departments which tend to be most difficult to organize seem to be those whose research has more direct industrial value and whose graduate workers will tend more towards industry than academia. How to effectively organize such departments, like Engineering, is a common problem across graduate unions. Power mapping served well as a tool for strengthening lab-level strike plans, but is difficult to apply to larger-scale strike strategy without developing a strong organizing base, composed of smaller subunits, in these more difficult departments. Focusing on sub-departmental and lab organizing in such departments had some degree of effectiveness for us over organizing the whole department as such – but the development of a truly effective approach is still an open problem.
Quantifying the effects of the research strike is also a difficult problem, particularly in real time during a strike. Figuring out how to do this more effectively would allow striking unions to better measure their real leverage and make strategic decisions, and help motivate striking workers to continue wielding this leverage.
As research workers build strikes in the years to come, they will no doubt confront and solve these problems and continue developing on the strategies we used at Dartmouth. We look forward to seeing our movement develop creative new ways to bend our employers to the workers’ will.