2025-2026 CISAC HONORS PROGRAM PRESENTATIONS
A Game-Theoretic Analysis of the Relationship between Domestic Volatility and Alliance Dynamics
Ellie Greyson | Watch presentation
Abstract: Existing formal-theoretic scholarship studying alliance dynamics tends to assume that states are unitary actors with fixed long-term interests. However, particularly in the United States, different administrations may have different interests and approaches toward their allies. This is especially relevant because alliance decisions are often made with the expectation of relatively secure, long-term payoffs. How does volatility in the political preferences of a patron state affect the dynamics of its alliance with client states?
In this thesis, I answer this question by modeling the strategic interaction between an allied client state and a series of U.S. patron administrations that may or may not be “transactional,” demanding payment in exchange for the alliance. I find that client states are more willing to provide payment when they are more vulnerable to outside threats, or less able to independently militarize outside of the alliance. While these results are in line with our existing intuitions, I also find that transactional administrations are able to extract less payment when future administrations are more likely to be transactional.
In other words, a transactional strategy is self-undermining—it only works when very few administrations are likely to do it. I validate this insight with three formal-theoretic extensions and two process-tracing case studies. Each adds nuance and complexity, but all support the central finding that the transactional strategy is self-undermining. This finding may caution future U.S. administrations against an overly transactional strategy—such strategies may have worked in the past only because of their rarity.
Aftershock: Modelling Resilience in U.S. Critical Mineral Supply Chains
Sidharth Gopisetty | Watch presentation
Abstract: The United States faces strategic vulnerability in its critical mineral supply chains, and existing policy frameworks rely on aggregate-concentration metrics that cannot be evaluated counterfactually or calibrated against the historical disruption record. Policymakers asked to size a strategic reserve, fund a domestic-capacity programme, or evaluate an export-control announcement currently lack a model that forecasts how the affected market will respond, how sharply the price will move after the shock, and how many years recovery to baseline will take. No publicly available engine produces these forecasts using both causal and structural modelling. This thesis develops a causal-inference framework that produces all three for any geopolitical shock to a critical mineral supply chain, applied to a panel of four minerals across twelve documented disruption episodes from 2006 to 2024 with graphite as the focal case.
The framework couples a country-resolved knowledge graph with a stock-flow ordinary differential equation on supply, capacity, inventories, and price. It accepts both public-sector shocks (export licences, tariffs, strategic-reserve injections) and private-sector shocks (cathode-chemistry substitution, anode-material substitution) as inputs, and is calibrated by differential evolution against bilateral price-and-quantity series from CEPII BACI, the Energy Information Administration, and the World Bank.
The framework is validated against the twelve historical episodes through three convergent diagnostics (in-sample fit, cross-episode out-of-sample transfer, and joint per-commodity calibration with shared demand-elasticity) yielding directional accuracy of 88% to 99% depending on the protocol. The framework's parameters are commodity-level structural properties rather than episode-specific fits.
This thesis demonstrates that critical mineral price direction, magnitude, and recovery duration can be forecast credibly across regime changes. The model supplies decision-makers with a quantitative basis for stress-testing planned interventions before they are announced, and supports informed planning rather than retrospective response to shocks. The framework is publicly auditable, episode-validated, and extensible to additional minerals and shock types as the policy environment evolves.
Who Can See the Earth? Earth Observation's Defense-Market Dependence and Civilian Access to Commercial Satellite Imagery
Lara Rudar | Watch presentation
Abstract: Within hours of Russia’s full-scale invasion of Ukraine in February 2022, leading commercial satellite firms released imagery of struck Ukrainian airbases, a testament to the transparency promised by the Earth observation (EO) sector. Four years later, the same providers imposed sweeping restrictions across much of the Middle East, limiting imagery access for journalists, humanitarian organizations, and open-source intelligence practitioners at the moments their work mattered most. The contrast is less a reversal than a revelation: commercial transparency has always been contingent on the alignment between operator interests and government preferences.
Existing scholarship explains militarization primarily through state procurement, implicitly assuming a healthy commercial market. I argue that this "Healthy Market Fallacy" misreads the causal direction. Capital-market pressures push firms toward defense independent of, and often prior to, direct government demand. Drawing on financial analysis of public filings and earnings calls, archival coding of company websites, and semi-structured interviews with executives, investors, attorneys, researchers, and OSINT practitioners, I trace how reliance on defense revenue intensifies precisely when projected commercial demand fails to materialize.
As firms restructure around defense customers through board appointments, sovereign satellite financing, and contractual lock-in, data access migrates from statutory licensing reviewable through public rulemaking to operator-managed discretion shaped by anticipated government preferences. The result is a tiered architecture in which governments receive priority tasking and dedicated capacity, while non-government users face challenges such as retroactive withholds, archive enclosure, and licensing friction. Displaced demand flows to European and Chinese providers, weakening the United States’ position as the default source of commercial imagery without meaningfully denying adversaries access. The thesis closes with policy recommendations for preserving independent verification through firm-level commitments, community-level alternatives, and government-level institutional design.
Preventing the Proliferation of Bioweapons: Artificial Intelligence for Biodefense
Taeil Matthew Kim | Watch presentation
Abstract: The convergence of artificial intelligence (AI) and biotechnology has begun to transform biological weapons from a state-monopoly threat constrained by tacit expertise into a capability accessible to non-state actors. Existing biodefense doctrine emphasizes upstream prevention, but such measures address only whether an attack can be initiated, not what occurs once initial defenses fail. Of the three strategic postures available against bioweapons—deterrence by punishment, dissuasion, and denial—only deterrence by denial remains viable in the AI era.
Deterrence by punishment is undermined by structural attribution failures, and deterrence by dissuasion is eroded as AI lowers the capability barrier. The same technological convergence that has widened the offensive frontier also enables a robust defensive response that deterrence by denial requires. AI can be deployed as a biodefense capability, operating at two distinct levels: patient-level AI interventions operate through individual agent decisions and probabilistic care-seeking behaviors, and system-level AI interventions modify transmission parameters and detection capabilities at the population level. This thesis investigates the following research question: how effective are patient-level AI interventions compared to system-level AI interventions in responding to disease outbreaks?
To address this question, I develop a hybrid simulation framework comparing four intervention regimes (no AI, patient-level AI, system-level AI, and integrated deployment) across five pathogens spanning the relevant epidemiological parameter space. Three findings emerge. First, both AI interventions reduce attack rates and severity outcomes substantially relative to baseline with integrated deployment dominating either tier in isolation. Second, the two interventions exhibit complementary specialization with patient-level AI acting most effectively on transmission and system-level AI acting most effectively on severity. Third, against fast-incubation pathogens, the threat profile most relevant to engineered bioweapons, only integrated deployment produces a meaningful effect with the synergy between the two interventions far exceeding the sum of their individual contributions. These findings indicate that deterrence by denial in the AI era requires a biodefense architecture that integrates both layers. Neither alone is sufficient against the most strategically concerning threat profiles.
Why Successful Revolutionary Regimes Support Co-revolutionaries Across the Border
Amina Wase | Watch presentation
Abstract: Since World War II, few revolutionary groups have succeeded in capturing the state. Yet when they do, some go on to support co-revolutionary movements abroad. This support comes with devastating consequence. This thesis asks: under what conditions do successful revolutionary groups, once in power, support co-revolutionary groups across borders? I argue that two conditions primarily drive support: an unresolved territory dispute between the revolutionary state and the neighboring target state, and shared ideology between the successful revolutionary regime and the co-revolutionary movement. Shared social ties facilitate support by making cooperation easier to organize and justify, while domestic political structure shapes the level and consistency of support over time.
Using a small-N comparative design, this thesis examines three cases: Taliban support and tolerance of the TTP after the 2021 takeover of Afghanistan; Gaddafi’s Libya and its support for FROLINAT and GUNT in Chad; and the collapse of EPLF–TPLF relationship into the Eritrean-Ethiopian conflict. The cases show that revolutionary support is driven by territory disputes and shared ideology, and support is most likely when the post-revolutionary regime has the domestic capacity to pursue external support consistently. These findings offer a framework for assessing the foreign policy behavior of newly victorious revolutionary regimes, including contemporary cases such as HTS in Syria, and for anticipating when revolutionary victory may generate broader regional instability.
How Generative AI Is Reshaping Non-State Malicious Actor Operations: An Empirical Case for Reorienting U.S. AI Governance
A New Trade Route or a Construction of Legitimacy? Thailand's Landbridge and The Politics of Persistence
Plengrhambhai (Pleng) Snidvongs Kruesopon | Watch presentation
Abstract: Why do large-scale infrastructure projects persist despite repeated evidence of economic and technical infeasibility? To answer this question, this thesis examines Thailand's proposed Land Bridge – a 90 km transport corridor linking the Andaman Sea and the Gulf of Thailand as a case of long-term political persistence with implications for regional trade and strategic connectivity.
A trade route originally conceived in the 17th century as a canal cutting across the Isthmus of Kra, the concept has since evolved into a proposed Land Bridge — estimated to cost over one trillion Thai baht. Despite decades of delays and repeated independent analyses questioning its economic viability, environmental impacts, and logistical feasibility, the project has endured as a political fixture. This thesis asks why. Drawing on twenty semi-structured interviews with policymakers, diplomats, academics, conservation leaders, and fishing communities in southern Thailand, alongside a quantitative media analysis of fifteen years of English-language press coverage of the Land Bridge, I argue that the project endures not because it is a sound infrastructure proposal but because it serves critical political functions within Thailand's hybrid regime: stabilizing elite coalitions, channeling infrastructure rents, centralizing authority, and projecting a narrative of national development that substitutes for democratic legitimacy. Yet the Land Bridge is more than a domestic political instrument. Positioned as a potential alternative to the Strait of Malacca — one of the world's most strategically sensitive maritime chokepoints — it situates Thailand within intensifying Indo-Pacific competition over supply chain resilience, energy security, and great-power infrastructure diplomacy. Decisions taken for largely domestic political reasons therefore carry significant consequences for regional trade routes and international security.
To explain how a project sustained by promise rather than performance can survive across Thailand's coups, elections, and civilian-military transitions, this thesis integrates three bodies of scholarship: megaproject management literature on strategic misrepresentation, anthropological work on infrastructure as promise and political performance, and comparative political economy scholarship on institutional lock-in and iron triangles. Together, these lenses reveal the Land Bridge as simultaneously a domestic instrument of regime maintenance and a geopolitical signaling device, two functions that reinforce one another.
Existing scholarship has largely debated whether the project should be built or whether it will be, but the political mechanisms behind four centuries of persistence remain underexplored. This thesis addresses that gap while centering the voices of communities whose livelihoods hang in the balance of a project designed to benefit those who propose it.
Title: In Pyongyang's Own Words: Audience Costs, Nuclear Narratives, and Summit Diplomacy in North Korean State Media
Abstract: Can a nuclear-armed state talk peace and keep the bomb at the same time? When states engage in international crises, a central problem in international relations is whether official rhetoric can be trusted as a signal of genuine intent. Fearon's domestic audience cost theory holds that the more democratic a state, and the more public opinion shapes political power, the more leaders risk domestic punishment for making public threats and backing down, a constraint that makes threats credible and disciplines official rhetoric. Weeks extends this to authoritarian regimes, arguing that among autocrats, personalist dictators face the lowest such costs of all due to minimal elite punishment capacity. Yet neither framework provides answers to what authoritarian leaders with minimal audience costs actually do with that freedom.
This thesis extends the audience cost framework into the underexplored terrain of summit diplomacy, using North Korean state media coverage from 2017 to 2022 to examine how a personalist regime exploits the near-absence of domestic accountability to sustain structurally contradictory rhetorical postures across diplomatic cycles, simultaneously signaling peace to an adversary, credibility to domestic public, and sovereignty to fellow authoritarian allies. Drawing on a bilingual close-reading of the Korean Central News Agency (KCNA), North Korea's state-owned news agency, across Korean- and English-language outputs and two comparative cases, the 2018 Singapore Summit and the 2019 Hanoi Summit, the thesis demonstrates that Pyongyang employed rhetorical sophistication rather than genuine nuclear moderation.
Rather than producing incoherence, the near-absence of domestic audience costs enabled a distinctive and stable rhetorical architecture, one that this thesis terms dual-track signaling, in which peace language and nuclear deterrence narratives are preserved simultaneously rather than traded off against each other. These findings complicate the existing literature by confirming that North Korea faces negligible domestic audience costs for rhetorical inconsistency, while revealing that the absence of such costs does not necessarily collapse the state’s incentive to manage credibility altogether.