BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Solon Barocas (Microsoft Research / Cornell University) DTSTART:20200429T213000Z DTEND:20200429T223000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/1 DESCRIPTION:Title: What Is a Proxy and Why Is it a Problem?\nby Solon Barocas (Microsoft Research / Cornell University) as part of NYU CDS Math and Dem ocracy Seminar\n\n\nAbstract\nWarnings about so-called ‘proxy variables ’ have become ubiquitous in recent policy debates about machine learning ’s potential to discriminate illegally. Yet it is far from clear what ma kes something a proxy and why it poses a problem. In most cases\, commenta tors seem to worry that even when a legally proscribed feature such as rac e is not provided directly as an input into a machine learning model\, dis crimination on that basis may persist because non-proscribed features are correlated with — that is\, serve as a proxy for — the proscribed feat ure. Analogizing to redlining\, commentators point out that zip codes can easily serve as a stand in for race. Yet\, unlike lenders\, a machine lear ning model will not seize on zip codes because the model intends to discri minate on race\; it will only do so because zip codes also happen to be pr edictive of the outcome of interest. So how are we to decide whether a var iable is serving as a proxy for race or as a legitimate predictor that jus t happens to be correlated with race? This question cuts to the core of di scrimination law\, posing both practical and conceptual challenges for res olving whether any observed disparate impact is justified when a decision relies on variables that exhibit any correlation with class membership. Th is paper attempts to develop a more principled definition of proxy variabl es\, aiming to bring improved clarity to statistical\, legal\, and normati ve reasoning on the issue. It describes the various conditions that might create a proxy problem and explores a wide range of possible responses. In so doing\, it reveals that any rigorous discussion of proxy variables req uires excavating the causal relationship that different commentators assum e to exist between non-proscribed features\, proscribed features\, and the outcome of interest. Joint with Margarita Boyarskaya and Hanna Wallach.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/1/ END:VEVENT BEGIN:VEVENT SUMMARY:Momin Malik (Berkman Klein Center for Internet and Society at Harv ard University) DTSTART:20201005T213000Z DTEND:20201005T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/2 DESCRIPTION:Title: A hierarchy of limitations in machine learning\nby Momin M alik (Berkman Klein Center for Internet and Society at Harvard University) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nIn the immor tal words of George E. P. Box (1979)\, “All models are wrong\, but some are useful.” This is an important lesson to recall amidst hopes and clai ms that the impressive successes of machine learning will extend to wider branches of inquiry\, and that its high-dimension and low-assumption model s can overcome what previously seemed to be insurmountable barriers. In th is talk\, I review the fundamental limitations with which all quantitative research into the social world must grapple\, and discuss how these limit ations manifest today.\n \nI cover sociological and philosophical aspects of the process of quantification and modeling\, as well as technical aspec ts around implications of the bias-variance tradeoff and the effect of dep endencies on cross-validation assessments of model performance. I metaphor ically structure the set of limitations as a tree\, where the root node is the choice to undertake systematic inquiry\, the leaf nodes are specific methodological approaches\, and each branch (qualitative/quantitative\, ex planatory/predictive\, etc.) represents tradeoffs whose limitations percol ate downwards.\n \nThis talk will serve as a useful overview about modelin g limitations and critiques\, as well as possible fixes\, for researchers in and practitioners of data science\, statistics\, and machine learning. It will also be useful as a primer for qualitative and theoretical social scientists on what are solid grounds on which to accept or reject applicat ions of techniques from these areas\, as well as where there are promising areas for developing new mixed methods approaches.\n\nThe talk format is a Zoom webinar.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/2/ END:VEVENT BEGIN:VEVENT SUMMARY:Audrey Malagon (Verified Voting / Virginia Wesleyan University) DTSTART:20201026T213000Z DTEND:20201026T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/3 DESCRIPTION:Title: Votes of Confidence: Leveraging Mathematics to Ensure Election Integrity\nby Audrey Malagon (Verified Voting / Virginia Wesleyan Uni versity) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nOur democracy relies on fair elections in which every vote counts and ensures a peaceful transition of power after the election. Concerns about foreign interference in our elections\, unreliable voting technology\, disinformat ion\, and last minute changes during the COVID-19 pandemic make this elect ion the most challenging in recent history.  In this talk\, we’ll discu ss how statistical post-election audits play a vital role in ensuring a fa ir and trustworthy process and other ways that mathematics can help ensure the integrity of our elections.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/3/ END:VEVENT BEGIN:VEVENT SUMMARY:Ariel Procaccia (Harvard University) DTSTART:20201207T223000Z DTEND:20201207T234500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/4 DESCRIPTION:Title: Democracy and the Pursuit of Randomness\nby Ariel Procacci a (Harvard University) as part of NYU CDS Math and Democracy Seminar\n\n\n Abstract\nSortition is a storied paradigm of democracy built on the idea o f choosing representatives through lotteries instead of elections. In rece nt years this idea has found renewed popularity in the form of citizens’ assemblies\, which bring together randomly selected people from all walks of life to discuss key questions and deliver policy recommendations. A pr incipled approach to sortition\, however\, must resolve the tension betwee n two competing requirements: that the demographic composition of citizens ’ assemblies reflect the general population and that every person be giv en a fair chance (literally) to participate. I will describe our work on d esigning\, analyzing and implementing randomized participant selection alg orithms that balance these two requirements. I will also discuss practical challenges in sortition based on experience with the adoption and deploym ent of our open-source system\, Panelot.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/4/ END:VEVENT BEGIN:VEVENT SUMMARY:Kate Starbird (University of Washington) DTSTART:20210308T223000Z DTEND:20210308T234500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/5 DESCRIPTION:Title: Online Misinformation during Crisis Events: The “Perfect Sto rm” of Covid19 and Election2020\nby Kate Starbird (University of Was hington) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nThe past year has been a difficult one. A pandemic has taken millions of lives and disrupted “normal” routines across the globe. In the United State s\, we have experienced an unprecedented political situation with a sittin g President refusing to concede after losing an election. Each of the even ts has been accompanied by uncertainty and anxiety\, as well as massive am ounts of false and misleading information. In this talk\, I will explore s ome of the mechanics of online misinformation\, explaining why we are part icularly vulnerable right now — due in part to the nature of these crise s\, and in part to the current structure of our information systems. Using examples from both Covid19 and Election2020\, I will explain how we are l iving through a “perfect storm” for both misinformation and disinforma tion. And I will describe how disinformation\, in particular\, can be an e xistential threat to democratic societies. After laying out the problems\, I aim to end on a more hopeful note\, with a call to action for researche rs and industry professionals to help “chip away” at this critical soc ietal issue.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/5/ END:VEVENT BEGIN:VEVENT SUMMARY:Mike Orrison (Harvey Mudd College) DTSTART:20210426T213000Z DTEND:20210426T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/6 DESCRIPTION:Title: Voting and Linear Algebra: Connections and Questions\nby Mike Orrison (Harvey Mudd College) as part of NYU CDS Math and Democracy S eminar\n\n\nAbstract\nVoting is something we do in a variety of settings a nd in a variety of ways\, but it can often be difficult to see nontrivial relationships between the different voting procedures we use. In this talk \, I will discuss how simple ideas from linear algebra and discrete mathem atics can sometimes be used to unify different voting procedures\, and how doing so leads to new insights and new questions in voting theory.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/6/ END:VEVENT BEGIN:VEVENT SUMMARY:Hakeem Angulu (Google / Metric Geometry and Gerrymandering Group) DTSTART:20210510T213000Z DTEND:20210510T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/7 DESCRIPTION:Title: The Voting Power Gap: Identifying Racial Gerrymandering with a Discrete Voter Model\nby Hakeem Angulu (Google / Metric Geometry and Gerrymandering Group) as part of NYU CDS Math and Democracy Seminar\n\n\nA bstract\nSection 2 of the Voting Rights Act of 1965 (VRA) prohibits voting practices or procedures that discriminate based on race\, color\, or memb ership in a language minority group\, and is often cited by plaintiffs see king to challenge racially-gerrymandered districts in court.\n\nIn 1986\, with Thornburg v. Gingles\, the Supreme Court held that in order for a pla intiff to prevail on a section 2 claim\, they must show that:\n\n1. the ra cial or language minority group is sufficiently numerous and compact to fo rm a majority in a single-member district\,\n2. that group is politically cohesive\,\n3. and the majority votes sufficiently as a bloc to enable it to defeat the minority’s preferred candidate.\n\nAll three conditions ar e notoriously hard to show\, given the lack of data on how people vote by race.\n\nIn the 1990s and early 2000s\, Professor Gary King’s ecological inference method tackled the second condition: racially polarized voting\ , or racial political cohesion. His technique became the standard techniqu e for analyzing racial polarization in elections by inferring individual b ehavior from group-level data. However\, for more than 2 racial groups or candidates\, that method hits computational bottlenecks.\n\nA new method o f solving the ecological inference problem\, using a mixture of contempora ry statistical computing techniques\, is demonstrated with this work. It i s called the Discrete Voter Model. It can be used for multiple racial grou ps and candidates\, and has been shown to work well on randomly-generated mock election data.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/7/ END:VEVENT BEGIN:VEVENT SUMMARY:Tina Eliassi-Rad (Northeastern University) DTSTART:20210329T213000Z DTEND:20210329T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/8 DESCRIPTION:Title: What can Complexity Science do for Democracy?\nby Tina Eli assi-Rad (Northeastern University) as part of NYU CDS Math and Democracy S eminar\n\n\nAbstract\nWe will discuss the following questions. What is dem ocratic backsliding? Is democratic backsliding an indicator of instability in the democratic system? If so\, which processes potentially lead to thi s instability? If we think of democracy as a complex system\, how can comp lexity science help us understand and mitigate democratic backsliding? Thi s talk is based on these two papers: K. Wiesner et al. (2018) in European Journal of Physics (https://doi.org/10.1088/1361-6404/aaeb4d) and T. Elias si-Rad et al. (2020) in Humanities & Social Sciences Communication (https: //www.nature.com/articles/s41599-020-0518-0).\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/8/ END:VEVENT BEGIN:VEVENT SUMMARY:Thomas Weighill (UNC Greensboro) DTSTART:20211004T213000Z DTEND:20211004T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/9 DESCRIPTION:Title: The Topology of Redistricting\nby Thomas Weighill (UNC Gre ensboro) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nAcro ss the nation\, legislatures and commissions are deciding where the Congr essional districts in their state will be for the next decade. Even under standard constraints such as contiguity and population balance\, they wil l have exponentially many possible maps to choose from. Recent computation al advances have nonetheless made it possible to robustly sample from this vast space of possibilities\, exposing the question of what a typical map looks like to data analysis techniques. In this talk I will show how topo logical data analysis (TDA) can help cut through the complexity and uncov er key political features of redistricting in a given state. This is joint work with Moon Duchin and Tom Needham.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/9/ END:VEVENT BEGIN:VEVENT SUMMARY:Jeanne Clelland and Beth Malmskog (UC Boulder / Colorado College) DTSTART:20211101T213000Z DTEND:20211101T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/10 DESCRIPTION:Title: Colorado in Context: A case study in mathematics and fair red istricting in Colorado\nby Jeanne Clelland and Beth Malmskog (UC Bould er / Colorado College) as part of NYU CDS Math and Democracy Seminar\n\n\n Abstract\nHow do we measure or identify partisan bias in the boundaries of districts for elected representatives? What outcomes are potentially “f air” for a given region depends intimately on its particular human and p olitical geography.  Ensemble analysis is a mathematical/statistical tech nique for putting potential redistricting maps in context of what can be e xpected for maps drawn without partisan data.  This talk will introduce t he basics of ensemble analysis\, describe some recent advances in creating representative ensembles and quantifying mixing\, and discuss how our res earch group has applied the technique in Colorado both in an academic fram ework and as consultants to the 2021 Independent Legislative Redistricting Commission.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/10/ END:VEVENT BEGIN:VEVENT SUMMARY:Ben Blum-Smith (NYU Center for Data Science) DTSTART:20211122T223000Z DTEND:20211122T234500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/11 DESCRIPTION:Title: Fair-division approaches to redistricting\nby Ben Blum-Sm ith (NYU Center for Data Science) as part of NYU CDS Math and Democracy Se minar\n\n\nAbstract\nProminent efforts to fight partisan gerrymandering in the US have sought the help of a (hopefully) neutral arbiter: they have a imed either to elicit intervention from the courts\, or to delegate respon sibility for redistricting to a formally nonpartisan body such as an indep endent commission. In this talk\, we discuss mechanisms to allow partisan actors to produce a fair map without the involvement of such a neutral arb iter. Inspired by the field of game theory\, and more specifically the stu dy of fair-division procedures\, the idea is to use the structured interpl ay of the parties' competing interests to produce a fair map. We survey th e various mechanisms that have been proposed in this fairly young line of research\, propose new mechanisms with novel potentially-desirable feature s\, and analyze them numerically.\n\nThis is joint work with Steven Brams\ , Irfan Jamil\, and Soledad Villar.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/11/ END:VEVENT BEGIN:VEVENT SUMMARY:Greg Herschlag (Duke University) DTSTART:20211213T223000Z DTEND:20211213T234500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/12 DESCRIPTION:Title: Quantifying Gerrymandering: Advances in Sampling Graph Partit ions from Policy-Driven Measures\nby Greg Herschlag (Duke University) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nGerrymanderin g is the process of manipulating political districts either to amplify the power of a political group or suppress the representation of certain demo graphic groups. Although we have seen increasingly precise and effective g errymanders\, a number of mathematicians\, political scientists\, and lawy ers are developing effective methodologies at uncovering and understanding the effects of gerrymandered districts.\n\nThe basic idea behind these me thods is to compare a given set of districts to a large collection of neut rally drawn plans. The process relies on three distinct components: First\ , we determine rules for compliant redistricting plans along with codifyin g preferences between these plans\; next\, we sample the space of complian t redistricting\nplans (according to our preferences) and generate a large collection of non-partisan alternatives\; finally\, we compare the collec tion of plans to a particular plan of interest. The first step\, though la rgely a legal question of compliance\, provides interesting grounds for ma thematical translation between policies and probability measures\; the sec ond and third points create rich problems in the fields of applied\nmathem atics (sampling theory) and data analysis\, respectively.\n\nIn this talk\ , I will discuss how our research group at Duke has analyzed gerrymanderin g. I will discuss the sampling methods we employ and discuss several recen t algorithmic advances. I will also mention several open problems and chal lenges in this field. These sampling methods provide rich grounds both for mathematical exploration and development and also serve as a practical an d relevant algorithm that can be employed to establish and maintain fair g overnance.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/12/ END:VEVENT BEGIN:VEVENT SUMMARY:Daryl DeFord (Washington State University) DTSTART:20220307T223000Z DTEND:20220307T234500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/13 DESCRIPTION:Title: Partisan Dislocation\, Competitivenes\, and Designing Ensembl es for Redistricting Analysis\nby Daryl DeFord (Washington State Unive rsity) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nComput ational redistricting techniques are playing an increasingly large role in the analysis and design of districting plans for legislative elections. I n this talk I will discuss recent work using Markov chain ensembles to eva luate tradeoffs between redistricting criteria and a new measure\, partisa n dislocation\, that evaluates plans by directly incorporating political g eography. Throughout\, I will demonstrate how examples of these methods in court cases\, reform efforts\, and map construction highlight the importa nt interplay between mathematics\, computational methods\, political scien ce\, and the law.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/13/ END:VEVENT BEGIN:VEVENT SUMMARY:Ana-Andreea Stoica (Columbia University) DTSTART:20220328T213000Z DTEND:20220328T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/14 DESCRIPTION:Title: Diversity and inequality in social networks\nby Ana-Andre ea Stoica (Columbia University) as part of NYU CDS Math and Democracy Semi nar\n\n\nAbstract\nOnline social networks often mirror inequality in real- world networks\, from historical prejudice\, economic or social factors. S uch disparities are often picked up and amplified by algorithms that lever age social data for the purpose of providing recommendations\, diffusing i nformation\, or forming groups. In this talk\, I discuss an overview of my research involving explanations for algorithmic bias in social networks\ , briefly describing my work in information diffusion\, grouping\, and gen eral definitions of inequality. Using network models that reproduce inequa lity seen in online networks\, we'll characterize the relationship between pre-existing bias and algorithms in creating inequality\, discussing diff erent algorithmic solutions for mitigating bias.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/14/ END:VEVENT BEGIN:VEVENT SUMMARY:Jennifer Wilson / David McCune (The New School / William Jewell Co llege) DTSTART:20221107T223000Z DTEND:20221107T233000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/15 DESCRIPTION:Title: Ranked Choice Voting and the Spoiler Effect\nby Jennifer Wilson / David McCune (The New School / William Jewell College) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nOne of the advantages c ommonly cited about Ranked Choice Voting is that it prevents spoilers from affecting the outcome of an election. In this talk we will discuss what a spoiler is and how it can be defined mathematically. Then we will examine how ranked choice voting performs relative to plurality voting based on t his definition. We will approach this theoretically\, assuming impartial\ , anonymous culture and independent culture models\; through simulation us ing both random and single-peaked models\; and empirically\, based on an a nalysis of a large database of American ranked choice elections. All of th ese confirm that ranked choice voting is superior to plurality based on th e likelihood of the spoiler effect occurring.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/15/ END:VEVENT BEGIN:VEVENT SUMMARY:Jayshree Sarathy (Harvard University) DTSTART:20221121T223000Z DTEND:20221121T233000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/16 DESCRIPTION:Title: Distrust in Noisy Numbers: Epistemic Disconnects Surrounding the Use of Differential Privacy in the 2020 U.S. Census\nby Jayshree S arathy (Harvard University) as part of NYU CDS Math and Democracy Seminar\ n\n\nAbstract\nFor decades\, the U.S. Census Bureau has used disclosure av oidance techniques in order to protect the confidentiality of individuals represented in census data. Yet\, the Census Bureau's modernization of its disclosure avoidance procedures for its 2020 Census triggered a controver sy that is still underway. In this talk\, I argue that the move to differe ntial privacy exposed epistemic disconnects around what we identify as a " statistical imaginary\," destabilizing a network of practitioners that uph olds the legitimacy of census data. I end by raising questions about how w e can—and must—re-imagine our statistical infrastructures going forwar d.\n\nJayshree Sarathy is a 5th year PhD student in Computer Science (and Science & Technology Studies) at Harvard University. She is part of the Th eory of Computation group and OpenDP project\, and is currently a graduate fellow with the Harvard Edmond & Lily Safra Center for Ethics. Her resear ch explores the complexities of privacy and data access within socio-techn ical systems.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/16/ END:VEVENT BEGIN:VEVENT SUMMARY:Sunoo Park (Columbia University) DTSTART:20221205T223000Z DTEND:20221205T233000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/17 DESCRIPTION:Title: CANCELED: Scan\, Shuffle\, Rescan: Two-Prover Election Audits With Untrusted Scanners\nby Sunoo Park (Columbia University) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nWe introduce a new wa y to conduct election audits using untrusted scanners. Post-election audit s perform statistical hypothesis testing to confirm election outcomes. How ever\, existing approaches are costly and laborious for close elections— often the most important cases to audit— requiring extensive hand inspec tion of ballots. We instead propose automated consistency checks\, augment ed by manual checks of only a small number of ballots. Our protocols scan each ballot twice\, shuffling the ballots between scans: a “two-scan” approach inspired by two-prover proof systems. We show that this gives str ong statistical guarantees even for close elections\, provided that (1) th e permutation accomplished by the shuffle is unknown to the scanners and ( 2) the scanners cannot reliably identify a particular ballot among others cast for the same candidate. Our techniques could drastically reduce the t ime\, expense\, and labor of auditing close elections\, which we hope will promote wider deployment.\n\nThis talk has been canceled. It will hopeful ly be rescheduled.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/17/ END:VEVENT BEGIN:VEVENT SUMMARY:Marion Campisi (San Jose State University) DTSTART:20230306T223000Z DTEND:20230306T234500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/18 DESCRIPTION:Title: The Geometry and Election Outcome (GEO) Metric\nby Marion Campisi (San Jose State University) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nWe introduce the Geography and Election Outcome (G EO) metric\, a new method for identifying potential partisan gerrymanders. In contrast with currently popular methods\, the GEO metric uses both geo graphic information about a districting plan as well as district-level par tisan data\, rather than just one or the other. We motivate and define the GEO metric\, which gives a count (a non-negative integer) to each politic al party. The count indicates the number of previously lost districts whic h that party potentially could have had a 50% chance of winning\, without risking any currently won districts\, by making reasonable changes to the input map. We then analyze GEO metric scores for each party in several rec ent elections. We show that this relatively easy to understand and compute metric can encapsulate the results from more elaborate analyses.\n\nMario n Campisi is an Associate Professor in the department of Mathematics and S tatistics at San José State University. Her research interests lie in low dimensional topology\, knot theory and the mathematics of redistricting.\ n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/18/ END:VEVENT BEGIN:VEVENT SUMMARY:Sunoo Park (Columbia University) DTSTART:20230424T213000Z DTEND:20230424T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/19 DESCRIPTION:Title: Scan\, Shuffle\, Rescan: Two-Prover Election Audits With Untr usted Scanners\nby Sunoo Park (Columbia University) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nWe introduce a new way to condu ct election audits using untrusted scanners. Post-election audits perform statistical hypothesis testing to confirm election outcomes. However\, exi sting approaches are costly and laborious for close elections—often the most important cases to audit— requiring extensive hand inspection of ba llots. We instead propose automated consistency checks\, augmented by manu al checks of only a small number of ballots. Our protocols scan each ballo t twice\, shuffling the ballots between scans: a “two-scan” approach i nspired by two-prover proof systems. We show that this gives strong statis tical guarantees even for close elections\, provided that (1) the permutat ion accomplished by the shuffle is unknown to the scanners and (2) the sca nners cannot reliably identify a particular ballot among others cast for t he same candidate. Our techniques could drastically reduce the time\, expe nse\, and labor of auditing close elections\, which we hope will promote w ider deployment. Joint work with Douglas W. Jones\, Ronald L. Rivest\, and Adam Sealfon.\n\nSunoo Park is a postdoctoral fellow at Columbia Universi ty and visiting fellow at Columbia Law School. Her research is in security \, cryptography\, privacy\, and related law/policy issues. She received he r J.D. at Harvard Law School\, her Ph.D. in computer science at MIT\, and her B.A. in computer science from the University of Cambridge.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/19/ END:VEVENT BEGIN:VEVENT SUMMARY:Ranthony Edmonds and Parker Edwards (MSRI/Duke University and Flor ida Atlantic University) DTSTART:20231106T224500Z DTEND:20231106T234500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/20 DESCRIPTION:Title: Quantifying Communities of Interest in Electoral Redistrictin g\nby Ranthony Edmonds and Parker Edwards (MSRI/Duke University and Fl orida Atlantic University) as part of NYU CDS Math and Democracy Seminar\n \n\nAbstract\nCommunities of interest are groups of people\, such as ethni c\, racial\, and economic groups\, with common sets of concerns that may b e affected by legislation. Many states have requirements to preserve commu nities of interest as part of their redistricting process. While some stat es collect data about communities of interest in the form of public testim ony\, there are no states to our knowledge which systematically collect\, aggregate\, and summarize spatialized testimony on communities of interest when drawing new districting plans.\n\nDuring the 2021 redistricting cycl e\, our team worked to quantify communities of interest by collecting and synthesizing thousands of community maps in partnership with grassroots or ganizations and/or government offices. In most cases\, the spatialized tes timony collected included both geographic and semantic data–a spatial re presentation of a community as a polygon\, as well as a written narrative description of that community. In this talk\, we outline our aggregation p ipeline that started with spatialized testimony as input\, and output proc essed community clusters for a given state with geographic and semantic co hesion.\n\nBios: Dr. Ranthony A.C. Edmonds is a Berlekamp Postdoctoral Res earcher at the Simons Laufer Mathematical Sciences Institute affiliated w ith the Department of Mathematics at Duke University. She earned a PhD i n Mathematics in 2018 from the University of Iowa\, an MS in Mathematical Sciences from Eastern Kentucky University in 2013\, and a BA in English an d a BS in Mathematics from the University of Kentucky in 2011. Her researc h interests include applied algebraic topology\, data science\, commutativ e ring theory\, and mathematics education. She is deeply invested in quant itative justice\, that is\, using mathematical tools to address societal i ssues rooted in inequity. Her current work in quantitative justice involve s applications of mathematics and statistics to electoral redistricting. \n\nDr. Parker Edwards is an Assistant Professor in the Department of Math ematical Sciences at Florida Atlantic University. His research focuses on both theory and applications for combining machine learning with tools fro m computational algebraic topology and geometry to analyze complex and hig h-dimensional data sets. He received a PhD in Mathematics from the Univers ity of Florida in 2020 and MSc in Mathematics and the Foundations of Compu ter Science from the University of Oxford in 2016.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/20/ END:VEVENT BEGIN:VEVENT SUMMARY:Sam Wang (Princeton and Electoral Innovation Lab) DTSTART:20231213T223000Z DTEND:20231213T233000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/21 DESCRIPTION:Title: Dimensionality reduction reveals dependence of voter polariza tion on political context\nby Sam Wang (Princeton and Electoral Innova tion Lab) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nPol itical dynamics in the U.S. have become highly polarized\, despite the fac t that individual voters can have complex views. To capture the dimensiona lity of how voters express their preferences\, we are analyzing over 400 r anked-choice elections\, in which voters rank candidates in order of prefe rence. We find that most voters act as if they share a representation of c andidates on a single axis. Voters each have a place on that axis\, in the aggregate defining a spectrum of simple political behavior. Voter spectra are more bimodal for executive and federal offices than for local offices \, suggesting that polarization of voter behavior is strongly shaped by av ailable choices. We are now investigating how candidate and voter behavior may be shaped by incentives arising from ranked-choice voting and other r eforms. Such shaping would suggest practical strategies to reduce politica l polarization.\n\nSam Wang has been a professor at Princeton University s ince 2000\, and director of the Electoral Innovation Lab since 2020. He ho lds a B.S. in physics from the California Institute of Technology and a Ph .D. in neuroscience from Stanford University. He has published over 100 ar ticles spanning neuroscience\, elections\, and democracy reform. A central feature of his research is the use and development of statistical tools f or dealing with large\, complex data sets. In 2004\, he pioneered methods for the aggregation of state polls to predict U.S. presidential elections. In 2012 he recognized new\, systematic distortions in representation in t he U.S. House\, leading to the creation of the Princeton Gerrymandering Pr oject. In 2020 these projects were subsumed into the Electoral Innovation Lab\, whose mission is to create and apply a practical science of democrac y reform.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/21/ END:VEVENT BEGIN:VEVENT SUMMARY:Abigail Hickok and Mason Porter (Columbia and UCLA) DTSTART:20240506T213000Z DTEND:20240506T223000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/22 DESCRIPTION:Title: Topological Data Analysis of Voting-Site Coverage\nby Abi gail Hickok and Mason Porter (Columbia and UCLA) as part of NYU CDS Math a nd Democracy Seminar\n\n\nAbstract\nIn many cities in the United States\, it can take a very long time to go to a polling site to cast a vote in an election. To find such "voting deserts" in an algorithmic way\, we use per sistent homology (PH)\, which is a type of topological data analysis (TDA) that allows one to detect "holes" in data. In this talk\, we'll give an i ntroduction to TDA and PH. We will then discuss our recent work on PH to d etect voting deserts and in the coverage of other resources.\n\n(Use inter active livestream for Q&A but the view-only livestream should have better sound.)\n\nAbigail Hickok is an NSF postdoctoral fellow in the Department of Mathematics at Columbia University. Prior to joining Columbia\, she com pleted a PhD in applied mathematics at UCLA in 2023\, and she received her undergraduate degree in mathematics at Princeton in 2018. Her research is on the theory and applications of geometric and topological data analysis .\n\nMason Porter is a professor in the Department of Mathematics at Unive rsity of California\, Los Angeles (UCLA). He earned a B.S. in Applied Math ematics from Caltech in 1998 and a Ph.D. from the Center for Applied Mathe matics at Cornell University in 2002. Mason held postdoctoral positions at Georgia Tech\, the Mathematical Sciences Research Institute\, and Califor nia Institute of Technology (Caltech). He joined as faculty at University of Oxford in 2007 and moved to UCLA in 2016. Mason is a Fellow of the Amer ican Mathematical Society\, the American Physical Society\, and the Societ y for Industrial and Applied Mathematics. In recognition of his mentoring of undergraduate researchers\, Mason won the 2017 Council on Undergraduate Research (CUR) Faculty Mentoring Award in the Advanced Career Category in the Mathematics and Computer Science Division. To date\, 26 students have completed their Ph.D. degrees under Mason's mentorship\, and Mason has al so mentored several postdocs\, more than 30 masters students\, and more th an 100 undergraduate students on various research projects. Mason's resear ch interests lie in theory and (rather diverse) applications of networks\, complex systems\, and nonlinear systems.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/22/ END:VEVENT BEGIN:VEVENT SUMMARY:Cory McCartan (NYU Center for Data Science) DTSTART:20240513T213000Z DTEND:20240513T223000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/23 DESCRIPTION:Title: Estimating Racial Disparities When Race is Not Observed\n by Cory McCartan (NYU Center for Data Science) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nDiscovering and quantifying racial dispa rities is critical to ensuring equitable distribution of public goods and services\, and building fair decision-making algorithms and processes.  But in many important contexts\, data about race is not available at the i ndividual level.  Methods exist to predict individuals' race from attribu tes like their name and location\, but these tools create their own set of   statistical challenges\, which if not addressed can significantly unde rstate or overstate the size of racial disparities.  This talk will disc uss these challenges and introduce new methodology to address them\, allo wing for accurate inference of racial disparities in datasets without raci al information.  The authors have worked with the U.S. Treasury Departmen t to apply the new method to millions of individual tax returns to estimat e disparities in who claims the home mortgage interest deduction\, the mo st expensive individual deduction in the federal tax code.\n\nCory McCarta n is a Faculty Fellow at CDS and will join the Penn State Department of St atistics in July.  He works on methodological and applied problems in th e social sciences\, including gerrymandering\, electoral reform\, privacy of public data\, and racial disparities.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/23/ END:VEVENT BEGIN:VEVENT SUMMARY:Lalitha Sankar (Arizona State University) DTSTART:20250224T223000Z DTEND:20250224T234500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/24 DESCRIPTION:Title: Understanding Last Layer Retraining Methods for Fair Classifi cation: Theory and Algorithms\nby Lalitha Sankar (Arizona State Univer sity) as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nLast-la yer retraining (LLR) methods have emerged as an efficient framework for en suring fairness and robustness in deep models. In this talk\, we present a n overview of existing methods and provide theoretical guarantees for seve ral prominent methods. Under the threat of label noise\, either in the cla ss or domain annotations\, we show that these naive methods fail. To addre ss these issues\, we present a new robust LLR method in the framework of t wo-stage corrections and demonstrate that it achieves state-of-the-art per formance under domain label noise with minimal data overhead. Finally\, we demonstrate that class label noise causes catastrophic failures even with robust two-stage methods\, and propose a drop-in label correction which o utperforms existing methods with very low computational and data cost.\n\n Lalitha Sankar is a Professor in the School of Electrical\, Computer and E nergy Engineering at Arizona State University. She joined ASU as an assist ant professor in fall of 2012\, and was an associate professor from 2018-2 023. She received a bachelor's degree from the Indian Institute of Techno logy\, Bombay\, a master's degree from the University of Maryland\, and a doctorate from Rutgers University in 2007. Following her doctorate\, Sank ar was a recipient of a three-year Science and Technology Teaching Postdoc toral Fellowship from the Council on Science and Technology at Princeton U niversity\, following which she was an associate research scholar at Princ eton. Prior to her doctoral studies\, she was a senior member of technical staff at AT&T Shannon Laboratories.\n\nSankar's research interests are at the intersection of information and data sciences including a background in signal processing\, learning theory\, and control theory with applicati ons to the design of machine learning algorithms with algorithmic fairness \, privacy\, and robustness guarantees. Her research also applies such met hods to complex networks including the electric power grid and healthcare systems. \n\nFor her doctoral work\, she received the 2007-2008 Electrical Engineering Academic Achievement Award from Rutgers University. She recei ved the IEEE Globecom 2011 Best Paper Award for her work on privacy of sid e-information in multi-user data systems. She was awarded the National Sci ence Foundation CAREER award in 2014 for her project on privacy-guaranteed distributed interactions in critical infrastructure networks such as the Smart Grid. She has led an NSF Institute on Data-intensive Research in Sci ence and Engineering (I-DIRSE)\, is a recipient of an NSF SCALE MoDL (Math ematics of Deep Learning) grant\, and a Google AI for Social Good grant. S ankar was a distinguished lecturer for the IEEE Information Theory Society from 2020-2022. She serves as an Associate Editor for the IEEE Transactio ns on Information Forensics and Security\, IEEE Information Theory Transac tions\, and was an AE for the IEEE BITS Magazine until August 2024.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/24/ END:VEVENT BEGIN:VEVENT SUMMARY:Matthew Dahl DTSTART:20250331T213000Z DTEND:20250331T224500Z DTSTAMP:20260404T092654Z UID:MathandDemoc/25 DESCRIPTION:Title: Large Legal Fictions: Detecting Legal Hallucinations in Large Language Models\nby Matthew Dahl as part of NYU CDS Math and Democrac y Seminar\n\n\nAbstract\nDo large language models (LLMs) know the law? LLM s are increasingly being used to augment legal practice\, but their revolu tionary potential is threatened by the presence of legal "hallucinations" -- textual output that is not consistent with the content of the law. In t his talk\, I theorize the provenance and nature of these hallucinations an d discuss methods for detecting them in LLM outputs. I then share results from three experiments auditing off-the-shelf LLMs and industry retrieval- augmented generation (RAG) models\, showing that legal errors remain wides pread. I conclude by emphasizing the need for empirical evidence in an age of ever-increasing hype about AI's ability to replace lawyers and expand access to justice.\n\nMatthew Dahl is a JD/PhD student at Yale Law School and Yale Department of Political Science. His research on AI\, judicial be havior\, and legal citation analysis has been published in the Journal of Empirical Legal Studies and the Journal of Legal Analysis. Before coming t o Yale\, he was a Fair Housing Fellow at Brancart & Brancart and received his BA from Pomona College.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/25/ END:VEVENT BEGIN:VEVENT SUMMARY:Bailey Passmore DTSTART:20250428T213000Z DTEND:20250428T223000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/26 DESCRIPTION:Title: Public data & human rights\nby Bailey Passmore as part of NYU CDS Math and Democracy Seminar\n\n\nAbstract\nFor the last 30 years\, the Human Rights Data Analysis Group (HRDAG) has been using statistics an d data science to support human rights advocacy campaigns around the world . While our international work typically involves casualty counts and usin g advanced statistical techniques to estimate undocumented victims\, our U .S. work often involves using public data to report human rights violation s experienced by those who are still living. Now\, the California Racial J ustice Act of 2020 has opened up even more opportunities for us to contrib ute to campaigns for racial justice\, particularly for those affected by r acial bias in criminal legal proceedings. Since early 2024\, HRDAG has bee n collaborating with public defenders to provide support for RJA claims fo r their clients. This led to our first time providing expert testimony in a U.S. courtroom in February 2025\, when we presented and defended 3 simpl e statistics based on the District Attorney's case data and county census data. Bailey will discuss the model HRDAG uses for obtaining and analyzing public data to address local human rights concerns\, as well as their exp erience working on an RJA case.\n\nBio: Bailey Passmore has been a data sc ientist at the Human Rights Data Analysis Group (“HRDAG”) since Januar y 2022. While at HRDAG\, they have designed reproducible and transparent d ata processing streams that include a variety of tasks\, such as scraping data from public transparency platforms\, extracting structured data from unstructured document collections\, extracting key information from text d ata using LLMs\, database deduplication and entity resolution\, version re solution\, and producing statistical analyses that speak to patterns of ra cial bias. Prior to their position at HRDAG\, Bailey worked as an undergra duate Data Science and Research Consultant for the San Diego Supercomputer Center\, where they mined\, cleaned\, and analyzed system performance dat a and prepared the findings for the Practice and Experience in Advanced Re search Computing (PEARC) conferences. Bailey graduated from the University of California\, San Diego with a bachelor of science degree in Cognitive Science\, after transferring with a background in Mathematics and Computer Science.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/26/ END:VEVENT BEGIN:VEVENT SUMMARY:Daphne Skipper DTSTART:20251110T223000Z DTEND:20251110T233000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/27 DESCRIPTION:Title: Geographic Access to Polling\nby Daphne Skipper as part o f NYU CDS Math and Democracy Seminar\n\n\nAbstract\nLonger travel distance s to polling places can discourage people from voting\, and these effects tend to fall hardest on minority communities. In this talk\, I will share a new approach for selecting polling sites that promote more equitable geo graphic access to voting. Our method does two things: it assesses how fair a given set of polling sites is\, and it identifies the optimal set of si tes to open from a list of possible locations. The key idea is to borrow a concept from the environmental justice literature\, the Kolm–Pollak Equ ally Distributed Equivalent (EDE)\, which is designed to compare distribut ions of disamenities such as exposure to air pollution. By adapting this m easure\, we can strike a balance between minimizing the average distance t o polls and improving access for residents who live farthest away. I will introduce the intuition behind the Kolm–Pollak EDE\, show how it fits in to an optimization model that scales to city- and county-level problems\, and demonstrate its use through a case study of early voting sites in DeKa lb County\, Georgia\, during the 2020\, 2022\, and 2024 elections.\n\nDaph ne Skipper is a mathematician and operations researcher specializing in co mbinatorial and global optimization. Her theoretical work examines nonline ar modeling structures that arise across a wide range of optimization prob lems\, with the goal of providing practical insight into how these structu res are handled in models and algorithms. She applies these insights to la rge\, complex systems where better modeling translates into real-world imp act. Some examples of her applied projects include maximizing the impact o f pollution-mitigation efforts in the Chesapeake Bay watershed\, optimizin g gas mixing and network operations to better meet demand\, and designing equitable facility-location models that balance efficiency with fairness. In this latter area\, her work spans methodological development\, equitabl e selection of election polling sites\, and improving access to grocery st ores in food deserts. Her research has appeared in leading journals such a s Nature Communications\, the Election Law Journal\, and Mathematical Prog ramming\, reflecting her commitment to applying mathematical rigor to prob lems of societal importance. Daphne lives and works in Annapolis\, Marylan d.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/27/ END:VEVENT BEGIN:VEVENT SUMMARY:Zarina Dhillon DTSTART:20251124T223000Z DTEND:20251124T233000Z DTSTAMP:20260404T092654Z UID:MathandDemoc/28 DESCRIPTION:Title: Evaluating Methods Used to Quantify Racial Segregation\nb y Zarina Dhillon as part of NYU CDS Math and Democracy Seminar\n\n\nAbstra ct\nRacial segregation has long been a problem in communities across the U nited States\, and in understanding how it is quantified we enhance our ab ility to offer proposals for eradication. Many metrics have been developed for measurement\, but none fully capture the nuances of this complicated issue: This work provides an overview of four mathematical approaches that have been developed to study segregation\, explains how they function\, a nd compares/contrasts their effectiveness in various situations in order t o determine which best succeeds. An additional focus lies in a case study of Los Angeles (LA) County. It was found that attempts to further standard ize outputs erases crucial data\, and compressing this issue into one scor e is not representative of its complexity. This suggests that future explo ration should attempt to study segregation more comprehensively rather tha n distilling an incredibly complicated and important issue into a single s tatistic.\n\nZarina Dhillon is earning her Masters in Applied Statistics a t NYU Steinhart with a concentration on data science for social impact. Za rina is also a Parke Research Fellow in the Brennan Center for Justice's D emocracy Program\, where she focuses on voter turnout and redistricting. S he earned her Bachelors in Mathematics from Claremont McKenna College as a proud transfer student from Santa Barbara City College\, where she earned nine associates degrees spanning economics\, philosophy\, communications\ , psychology\, and mathematics.\n LOCATION:https://stable.researchseminars.org/talk/MathandDemoc/28/ END:VEVENT END:VCALENDAR