Our comparative analysis unfolded through an iterative process of analytic memoing and bi-weekly discussions over several months. This sustained engagement enabled us to trace recurring patterns across sites while remaining attentive to local specificities. The case studies—airline surge pricing during Hurricane Helene, the restricted deployment of Starlink in disaster zones, and the collapse of Texas’s power grid during Winter Storm Uri—were selected not only for their illustrative value but also for their nearness to our lives. Together, these cases illustrate three algorithmic modes of disaster capitalism: computational price exploitation, infrastructural fragility from deregulated predictive systems, and privatized control over essential connectivity. They also show how practices of mutual aid and collective refusal emerge in response to these computational shortcomings. Critically, while this paper focuses on U.S.-based cases shaped by our own lived experiences and research contexts, we recognize that disaster capitalism and mutual aid take radically different forms globally. We do not claim universality, but rather offer these cases as situated provocations toward broader inquiries into the intersection of algorithmic harm and grassroots resilience.

Our methodological approach—combining autobiographical proximity with discourse analysis of news coverage, government reports, and organizational communications—necessarily privileges breadth over depth. We trace how algorithmic systems and mutual aid responses were publicly framed and narrated, rather than offering the thick description that ethnographic immersion would provide. This scope means we do not capture the full complexity of on-the-ground organizing: the interpersonal negotiations, the contested decisions, the moments of friction and failure that characterize any collective endeavor. We view our contribution as complementary to future ethnographic work that would attend more closely to the situated practices and emergent governance structures of mutual aid networks in crisis contexts.

…

These responses illuminate a fundamental contrast with the logics of algorithmic extraction. Where surge-pricing algorithms treated evacuation as a marketplace—translating desperation into demand signals and vulnerability into profit—mutual aid networks operated under an entirely different calculus. Transportation was allocated based on need rather than ability to pay; resources flowed toward those most at risk rather than those with the greatest purchasing power. The crowdfunding efforts that covered bus tickets for strangers enacted a form of collective risk-pooling that inverted the individualized burden imposed by dynamic pricing. Rather than optimizing for revenue extraction during peak demand, these networks optimized for collective survival, demonstrating that the “efficiency” claimed by algorithmic systems is always efficiency for particular ends—and that communities can organize around radically different ends when given the chance. Moreover, these efforts were not simply reactive. Still, they drew on pre-existing relationships and infrastructures of care, underscoring that mutual aid is not ad hoc but relies on the slow, ongoing work of building trust and solidarity before disaster strikes.

The persistent issue of surge pricing during mandatory evacuations raises critical ethical questions about fairness, accessibility, and corporate responsibility. However, the response to Hurricanes Helene and Milton also underscored the power of mutual aid in filling the gaps left by disaster capitalism. As climate change intensifies the frequency and severity of natural disasters, such community-led initiatives will remain essential in ensuring that disaster relief prioritizes human lives over profit (particularly when regulation fails to rein in the practice). By supporting and strengthening mutual aid networks, communities can build resilient systems of care that counteract the exploitative practices of algorithm-driven markets.

…

Firstly, algorithmic forecasting systems failed to anticipate the scale of the demand surge during the storm. ERCOT’s reliance on historical data for predictive modeling proved inadequate for an event outside typical parameters, a phenomenon often described as the “black swan” problem in computational analytics [23, 130]. This failure underscores the risks of over-reliance on machine learning models without accounting for rare but high-impact scenarios [75, 134].

Secondly, digital control systems—which manage the distribution of electricity—struggled to respond to the cascading failures within the grid. Automated mechanisms designed to balance supply and demand instead triggered widespread rolling blackouts, as they were unable to distinguish between recoverable outages and critical failures. While efficient under normal circumstances, this automation amplified the crisis by prioritizing grid stability over localized human needs [23, 59].

…

Scholars like Pasquale and O’Neil [96] argue that the secretive nature of algorithms often undermines public trust and exacerbates feelings of helplessness during crises, as affected populations struggle to understand or challenge the decisions made by computational systems. The opacity of algorithmic decision-making often leaves affected populations with little recourse to address grievances or to understand why specific neighborhoods experienced longer outages than others. Scholars argue that this “algorithmic opacity” can undermine trust in public institutions and exacerbate feelings of helplessness during crises [96, 134].

…

Perhaps most significantly, these responses revealed what algorithmic opacity had obscured. While affected residents struggled to understand why their neighborhoods faced extended outages, mutual aid networks operated through radical transparency: needs were articulated openly, resources were tracked collectively, and decision-making was visible to participants. This transparency was not incidental but constitutive—mutual aid depends on shared knowledge and trust in ways that proprietary algorithmic systems, by design, do not. By leveraging existing social networks and local knowledge, Austin’s residents demonstrated that resilience emerges not from computational optimization but from the slow accumulation of relational infrastructure—the very forms of care work that market-oriented systems consistently devalue.

…

We contribute to CSCW by advancing what we call infrastructures of mutuality, a way of linking critique of algorithmic exploitation with design orientations inspired by mutual aid. Across our cases, we show that the systemic failures of surge pricing, privatized satellite connectivity, and deregulated energy markets are not aberrations but constitutive features of algorithmic systems optimized for profit. These systems translate vulnerability into market opportunity, thereby reproducing disaster capitalism. In contrast, mutual aid emerges not as a stopgap or ad hoc tactic, but as a generative counter-logic. It demonstrates how infrastructures can be oriented toward relationality, transparency, and reciprocity, directly challenging the opacity and exclusion embedded in algorithmic systems. Extending critical and reflective traditions in CSCW, we describe infrastructures of mutuality as the design and maintenance of systems that both meet immediate needs and enact alternative social relations. Drawing from practices of care and commoning, this orientation highlights how computing might support solidarity, redundancy, and collective governance, rather than transactional optimization. In doing so, we seek to move beyond the descriptive reframing of market-driven crisis informatics and toward an approach that researchers and designers can actively adopt in theorizing, evaluating, and creating technologies for resilience.

I think this paper is really interesting because it places algorithms designed by corporations in contrast with mutual aid in a simple but generative way that illustrates well how suffering and confusion are often a product of the architecture of systems (designed this way intentionally or not).