Skip to content Skip to navigation

Funded Research Projects

We are pleased to announce the results of our RFP process, focusing on interdisciplinary research relating to cyber-social systems. From the proposals submitted to us, we are funding 32 of the most promising, with $1.1M for 2015-2016, $1.1M in funding for new projects for 2016-2017, and $900K in funding for new projects from 2017-2018. The projects selected are, in alphabetical order:


Advancing AI Research to Help Policymakers Affordably Improve Life’s Starts and Finishes (2016-2017; Fei-Fei Li, Arnold Milstein):

Understanding grows about childhood experiences occurring primarily in lower and middle class homes that limit fulfillment of children’s’ developmental potential. Simultaneously nations and US state governors face rising demand for costly institutional care that many seniors’ dread. In the United States, the cost of long-term care would more than double from 1.3% of US GDP in 2010 to 3% of US GDP in 2050 if the rate of functional limitations among those age 65 and older remains constant (Congressional Budget Office, 2013). These two trends confront policy-makers with painful fiscal trade-offs. The prior watershed decade was the first in which Medicaid funding demands fueled by institutional spending for seniors’ care exceeded state funds available to fund children’s’ education. Rapid advances in the capability and affordability of in-home AI systems may enable policymakers to more affordably and effectively serve these two vulnerable populations during life’s starts and finishes. Effective, scalable uses of prior generations of cyber systems have improved value-for-money in other service sectors such as airlines and banking. However, use of modern AI capabilities to improve the value of more intimate interpersonal human services is fraught with hope and fear for seniors, families, health professionals, educators and policymakers seeking to serve them cost-effectively. Both emotions are well-founded. Stanford faculty, fellows, and students from its schools of engineering and medicine seek to formulate and test psychologically nuanced applications of AI in order to increase policy-makers' and industry's understanding of how modern AI systems can more affordably and effectively (1) enable care planners to select in-home care plans that will generate the largest gains in seniors’ self-care capabilities; and (2) boost non- affluent parents’ contribution to the physical, mental and social development of children.

The Anatomy of Ransomware Attacks (2017-2018; Sharad Goel, Camelia Simoiu):

A new class of malware known as ransomware has emerged and gained popular among cybercriminals over the last decade. Ransomware works by restricting an individual’s access to their computer (e.g., by encrypting their data), and then demanding payment to restore functionality. The first known case of a ransomware attack occurred almost 10 years ago. This attack has since been professionalized and is thought to now be highly profitable, with some estimates placing the damage at hundreds of millions of dollars per year. Despite the harm ransomware can inflict, relatively little is known about the prevalence, characteristics, and circumstances of such attacks. Which segments of the population are most at risk of a ransomware attack? How do users become infected with ransomware? How much ransom is typically demanded, and what proportion of users pay? The aim of this project is twofold: (1) to estimate the prevalence and characteristics of ransomware attacks; and (2) to identify online behaviors that place individuals at risk of experiencing such attacks. To meet these objectives, we will design a comprehensive survey on ransomware experiences and administer it to a representative sample of approximately 1,000 individuals. Through an existing collaboration with thepolling company YouGov, we will be able to pair survey responses with browsing history data for each participant. Aside from questions detailing the attack, the survey will also include questions about general security habits, the technology used (e.g., operating system, web browser, and plug-ins installed), and a test to estimate levels of web-savviness. We will use machine learning techniques to identify online behaviours that are predictive of ransomware attacks. 

Assessing the Impact of Digital Technologies on the Labor Market (2017-2018; Yong Lee, Chuck Eesley, Stephen Zoepf): 

There is increasing concern that the wave of new technologies, such as robotics, digital platform economy, big data analytics, artificial intelligence, and so forth, will disrupt jobs and workers in the near future. While technology can increase productivity of workers and create jobs, it could also directly replace workers. This project examines the net impact of technology on jobs and labor markets. In particular, we examine how digital platform economy in transportation, big data analytics in finance, and robotics in manufacturing will impact workers in each sector. Based on our findings we propose policies that can help alleviate the impact of technology on labor and help prepare future labor for the digital economy. Finally, through this project we will create a platform where scholars and students at Stanford interested in the impact of digital technology on labor markets can exchange ideas, collaborate, and continue to pursue this research objective.

Behavioral Biometrics (2015-2017; Russell Poldrack, David Mazieres, Bahman Bahmani):

Authentication is one of the major problems faced by the society in interacting with cyber technology. Passwords, challenge questions, out-of-band text messages, and physiological biometrics create friction with user experience, and yet are increasingly bypassed by hackers. In this project, we will study the use of behavioral biometrics, i.e., the unique traits in user interactions with digital devices and services, for frictionless cyber authentication. We will use a principled approach based on human cognitive psychology, systems security, and data mining to design authentication schemes based on behavioral biometrics. Due to the critical importance of authentication in online and mobile banking and financial services, we will focus on and analyze the effectiveness of our authentication schemes for the use cases in this sector. Further, taking a cyber-social perspective, we will study societal aspects such as data protection legislation and policy, financial regulation compliance, and usability, as well as operational aspects such as personnel and economic impacts, deployment models, and maintenance requirements.

Campaign of the Future (2015-2016; Nate Persily, Bruce Cain):

Technological advances in voting, communication, and fundraising are changing American campaigns and elections in fundamental ways.  Developments in microtargeting, web-based campaign advertisements, and even voting, itself, promise to empower new actors in campaigns and reshape the landscape for political communication.  American democracy's move on-line will have profound implications for the future of traditional intermediary institutions, especially political parties, which have served as the primary avenues for individual participation and representation.  The project on the Campaign of the Future seeks to bring together the relevant actors in the campaign system to analyze these trends in political communication, mobilization, and voting, and to assess their impact on American democracy.  The project will involve conferences of academics and other experts, an edited volume, and a sole-authored book.

Collective Action and Governance in an Online Piecework Economy (2016-2017; Michael Bernstein, Margaret Levi):

The digital gig economy has led to a resurgence of piecework. Without shared factories and water coolers, how do digital pieceworkers coordinate, build solidarity, and take collective action? We will engage in fieldwork with digital pieceworkers who work in data entry, domestic services, and on-­demand driving to understand how they counter algorithmic systems and engage in collective behavior. We will then design and launch a new collectively-governed platform for gig workers. Our goal is to both understand the factors that drive collective behavior in distributed gig platforms, and to introduce new designs and infrastructure that enable the growth of worker collective action in the digital piecework (gig) economy. Our results shed light on the strengths and weaknesses of digital piecework cooperativism, as well as the policies necessary for it to succeed.

Consumer Privacy (2015-2016; Kostas Bimpikis, Yonatan Gur):

Recent advances in information technology have allowed firms to gather detailed data about consumers’ preferences and the structure of their social interactions. Along with the growingly adopted targeting technologies, the wealth of available information benefits firms and holds a lot of promise for individuals. On the other hand, challenges arise with regards to the sensitive nature of the information entities such as firms and government agencies may collect about individuals. In such a context data holders may take advantage of the individuals’ inability to fully comprehend and anticipate the potential uses of their private information with detrimental effects for aggregate social welfare. These challenges are only amplified by the fact that consumer information is a valuable business asset and it is typically infeasible for an individual to retain full control of its informational value. Moreover, individuals interact with one another forming social relationships and preserving one’s privacy in the context of a connected society which presents another set of interesting questions especially when individuals have different views on what may constitute sensitive information. This proposal aims to develop a comprehensive way to think about extracting the significant societal and commercial value of data about individuals while taking their privacy considerations into account.

Costs of Cyber Data Breaches in Public Companies (2016-2017; Michael Klausner, George Triantis):

This project addresses the need for quantitative measures of the costs incurred by publicly traded companies that experience cyber data breaches.  The currently available information is focused more on frequency than severity of breach, including the scope and costs of breach.  Most of the publicly available data is the result of surveys and the commercially available cyber-breach data are regarded by many to be inaccurate, incomplete and not organized in such a way as to support analysis.  This project collects a broad set of data relating to the costs that publicly traded firms have incurred as a result of breaches from 2007 to the present.  For its sources, it draws on thorough examination of publicly available sources, including filings with the Securities Exchange Commission, Federal Trade Commission enforcement releases, PACER federal court filings, state breach disclosure websites and national news stories. 

Crowdsourced Democracy (2015-2016; Ashish Goel, Larry Diamond):

YouTube competes with Hollywood as an entertainment channel, and also supplements Hollywood by acting as a distribution mechanism. Twitter has a similar relationship to news media, and Coursera to Universities. But Washington has no such counterpart; there are no online alternatives for making democratic decisions at large scale as a society. As opposed to building consensus and compromise, public discussion boards often devolve into flame wars when dealing with contentious socio-political issues. This project aims to (a) Design systems that are geared towards structured discussion and consensus decisions, as opposed to towards a free-flowing conversation that degenerates into vitriol, (b) Develop an algorithmic and game-theoretic understanding of voting, decision, and incentive mechanisms, and (c) Deploy these systems in real-life crowdsourced democracy processes, and evaluate them rigorously.

Crypto Policy Project (2015-2017; Jennifer Granick, Dan Boneh):

Encryption helps human rights workers, activists, journalists, financial institutions, innovative businesses, and governments protect the confidentiality, integrity, and economic value of their activities. However, strong encryption may mean that governments cannot make sense of data they would otherwise be able to lawfully access in a criminal or intelligence investigation. In the 1970s, and again in the 1990s, U.S. government struggled with tradeoffs between its surveillance/law enforcement missions (potentially thwarted by crypto) and its information assurance/crime prevention missions (furthered by crypto). In the main, these debates were resolved in favor of allowing the proliferation of strong crypto. Today, the crypto policy issue has resurfaced. FBI Director James Comey chides Apple and Google for using cryptography architectures that the companies are unable to decrypt for law enforcement. In secret, the intelligence community is invested in breaking popular encryption schemes, stealing encryption keys, and finding ways to circumvent communications security protocols. The Crypto Policy Project investigates and analyzes the policy and practices of the U.S. and foreign governments for forcing decryption and/or influencing crypto-related design of online platforms and services, devices, and products, both via technical means and through the courts. The project’s interdisciplinary approach includes technical analysis of policy proposals for encryption design, contributed by cryptography researchers in the Stanford Computer Science Department’s Applied Cryptography Group. The project also researches the benefits and detriments of strong encryption on free expression, political engagement, economic development, and other public interests. More information about the Crypto Policy Project can be found at

Cyber-Enabled Information and Influence Warfare and Manipulation: Understanding Problems, Developing Solutions (2017-2018; Amy Zegart, Herb Lin, Tom Fingar, Nate Persily, Lee Ross): 

Hostile cyber operations are characterized as acts of war in academic and policy debates, which in turn has led researchers to apply theories of war to understand actor behavior in this domain. Yet recent events show that many hostile cyber operations fall short of the threshold of war; they are more appropriately in the realm of intelligence operations and covert action. This project examines the psychological, organizational, legal, and international security dimensions of cyber-enabled influence/information warfare and manipulation (IIWAM) operations through this new framing of the problem. The project explores how individuals receive information in a saturated and increasingly manipulated information environment; what geopolitical forces drive state and non-state actors to wage cyber-enabled IIWAM operations; how U.S. national security agencies are structured to deal with information operations; why U.S. democracy is particularly vulnerable; and potential counters to cyber-enabled IIWAM.

Cyber Insurance Policies and Coverage (2016-2017; Ben Lawsky, George Triantis):

As our visiting scholar, Ben Lawsky's research into cyber insurance policies examines how policy language differs across industries as well as between same-industry clients, and introduces scenario planning to evaluate how well coverage matches the scope of potential catastrophic and interconnected cyber incidents.

Cyber Systems in Healthcare Organizations (2015-2016; Melissa Valentine, Mohsen Bayati):

Advanced cyber-systems hold tremendous promise for transforming modern hospitals, potentially improving their capacity, safety, and operational efficiency by extending limited human ability for memory, judgement, and situational awareness.  Yet technologies are not exogenous “interventions” into organizational systems.  Instead, they are shaped by and then in turn shape the social system, as people interpret and enact the technologies based on their professional identities or the power dynamics activated by the technology implementation process. Also, at present these systems are technically limited in their ability to support situational awareness and data-driven decision-making.  This research project aims to advance understanding of cyber-technology enactment and to advance the frontier of dynamic learning and decision-making in health care organizations. We plan to use ethnographic field research methods and operations research analytics to study and improve two health-care cyber-systems: real-time locating services (RTLS) at the Stanford South Bay Cancer Center and a new Hospital Operations Center (HOC) at the Stanford Hospital.

Cyber-Text Technologies: Presenting the Future of the Past (2016-2017; Elaine Treharne, Ronald Egan):

Cyber Text Technologies will ascertain from a small but detailed set of case studies to what extent all forms of human communication might be not only systematic, but also effectively skeumorphic, unconsciously emulative, and to an extent formulaically replicative. We aim to investigate through machine learning that employs refined and targeted modeling whether or not specific conventions and identifiable trends characterize every text technology from Cuneiform to Snapchat. Since text technological transformation is constant and cyclical, it is a definitive goal of this project to attempt to project into the future the ways in which these trends will manifest themselves in the development of new information systems and forms of communication from Virtual Reality to Incorporeal Technologies.

Cyber Tech in Patient-Physician Relationships (2015-2016; Abraham Verghese, Jeff Chi, Sonoo Thadaney):

We believe the 20th century was one of technology development, and the 21st century is focused on appropriate and relevant application of technology. The technology in cyber-social health systems promised increased efficiency, improved effectiveness, fewer errors, and global connectivity. However, we’re now aware of the unintended consequences. In modern medicine, particularly in the hospital, the bed-ridden patient has become an icon for computerized patient data - an entity we’ve termed the “iPatient”. The iPatient receives the undivided attention of the medical team, is the subject of discussions on cyber-social medical systems about their medical images and more. The integration of human patients and doctors with medical cyber technologies, such as the electronic patient record and networked tools recording medical data, form a cyber-social medical system. In this system, often the human patient experiences abandonment and wonders - where are the doctors? What is going on? What is the plan? Who is in charge? The disjunction between the patient’s experiences of inattentiveness is at odds with the physicians’ perception of being very attentive (to the patient’s data), and of providing good medical care. We believe this disconnect stems from the idea that knowledge and data don’t translate to wisdom, a key component in the art of medicine.

Cyber Work: The Future of Networked Labor (2015-2018; Michael Bernstein, Ramesh Johari, Margaret Levi, Melissa Valentine):

Technology has transformed from a tool that supports work into a comprehensive infrastructure that connects workers to employers. Platforms such as Uber and Amazon Mechanical Turk, which announce themselves as the “gig economy” and “paid crowdsourcing”, signal a shift where workers and employers connect ad-hoc, at large scale, to accomplish complex tasks. This shift to online networked labor has the potential to dramatically reconfigure how we shape our careers, organizations, and market platforms, and in turn shifts how those careers, organizations and platforms shape our society. Inspired by this transformation and its risks, our project addresses challenges facing the entire span of the networked labor ecosystem: individuals, organizations, and the work platform itself. We study three fundamental questions: first, how will people manage their work lives online? Second, how might organizations look in a future of networked labor? Third, how do networked labor platforms succeed? To address these challenges, we propose a combination of social scientific, design and engineering endeavors. Our efforts aim to envision the future of digital work, and to inform and create the technological platforms that enable it.

Cybersecurity of Health Care Data in Population-Based Health Information Exchanges (2016-2017; Lorene Nelson, Mark Cullen, Dan Boneh, Ling Yu, Michelle Mello, Nigam Shah):

Health information exchanges (HIEs) integrate electronic health records from multiple healthcare facilities in a geographic region, and secondary use by researchers of the data from these large generalizable samples have tremendous potential for benefitting population health and informing public policy.  Although state and federal laws provide some architecture for protecting healthcare information privacy, most of these laws predate and do not contemplate the recent explosion in the number of online information sources, leaps in computing capabilities, and rise in hacking incidents.  The absence of standard data sharing guidelines with rigorous privacy guarantees has made health care organizations afraid to share data, posing significant hurdles for the population health research use of HIE data.  An interdisciplinary group of faculty with expertise in population health, epidemiology, computer science, statistics, health policy, bioethics and law will address these challenges from two perspectives:  (1) societal, legal and health policy aspects, assessing the public’s views regarding HIE privacy risks and proposing legal and policy measures that would better achieve the balance of risks and benefits the public desires; and (2) cybersecurity aspects, addressing the key technological challenges for designing secure encryption systems and statistical techniques and identifying critical gaps where novel solutions are needed.

The Deteriorating Health of the Digital Information Ecosystem and Its Deleterious Effects on Democracy & Human Rights (2017-2018; Eileen Donahoe, Larry Diamond, Russell Berman, Daphne Keller):

While appreciating the benefits of digital platforms for society, this project addresses deleterious effects of digital technologies on the global information ecosystem and negative consequences for democracy. It will identify a spectrum of negative dynamics (e.g., spread of fake news, creation of echo chambers, normalization of hate, reinforcement of stereotypes/bias, facilitation of violent extremism, weaponization of information/doxing, psychographic targeting, information operations) and describe how the combined effect of these dynamics presents an ominous threat to democracy. It will address conceptual challenges posed by globalized digital technology to democratic governance; and provide practical analysis of a range of proposed policies, regulations, and voluntary codes of conduct put forth by governments, private sector platforms and civil society to address these challenges. Key practical concerns include movement by democratic governments toward illiberal regulation and erosion of the core concept of platform immunity from liability which has been essential to free expression. We will convene global governance actors to generate policies that optimize for protection of freedom, security, and democratic discourse. This project will produce educational materials, and help policymakers to identify optimal roles for private sector platforms and democratic governments in the global digitized context, so society can reap the benefits of technology.

Documenting Combined Capabilities for Internet Security (2017-2018; Amy Zegart, Jesse Sowell, Herb Lin, Harold Trinkunas): 

Transnational network operator communities, which comprise roughly a few thou- sand individuals from volunteer organizations, non-profits, and for-profit firms, are little known but critical actors ensuring Internet security. Operator community members collaborate informally with both industry peers and law enforcement to keep nefarious activities such as botnets and phishing out of cyberspace. For instance, approximately 90% of e-mail that traverses the Internet is spam. These actors ensure that spam does not reach users’ inboxes. Yet, precisely because these communities operate ad hoc, through personal relationships based on trust, they remain vastly understudied. This project proposes to fill the research and policy design gap in this space. Through interviews, fieldwork, and Track II style workshops, researchers will engage directly with parties involved in operational incident response to better document and understand (1) how operator communities function, (2) how they could improve cyber incident response by developing more systematic combined capabilities with law enforcement, and (3) how to better design cybersecurity policy in light of these dynamics. The end goal is to produce both academic and policy oriented documents that culminate in a book designed to provide state actors with a much needed roadmap to developing reliable combined capabilities for incident response.

First Principles for Governing Academic Records in the Digital Era (2015-2016; Mitchell Stevens, Dan Boneh, Tom Black):

Digital learning environments and data analytics have dramatically expanded what might count as academic records, raising questions about the viability of inherited record systems predicated on paper or paper-equivalent documents and institutionally based verification systems.  Engineers and student services professionals at Stanford and worldwide are actively developing academic record systems more appropriate for a digital era.  Because academic credentials are increasingly fateful for people’s life chances, all of those who produce and purvey them must do so with careful attention to the privacy and discretion of learners and to the integrity of the human relationships inherent in any instructional process.  Our project specifies first principles for the ethical governance of these new technologies.

Folk Theories of Cyber-Social Systems and their Implications for Privacy (2016-2018; Jeff Hancock, Michael Bernstein):

As people interact with complex cyber-social systems such as Facebook’s ranked news feed and Uber’s hiring algorithms, they build up folk theories of how these systems work. These theories, however, can often be wrong. For example, many people believed the Facebook news feed to be an unfiltered window of their friends’ behavior, leading to widespread surprise and news coverage when a Facebook experiment on emotional contagion highlighted that Facebook manipulates the content of users’ feeds. We propose to investigate the folk theories that people hold about complex cyber-social systems, and determine whether users’ privacy behaviors on these systems are direct reflections of their folk theories. We then propose targeted design interventions to nudge users’ folk theories. This research highlights how systems and algorithms impact society not only through their direct outputs, but also through the (potentially problematic) understandings that people form of them.

How Intermediaries Affect User Choice in News and Commerce (2016-2018; Susan Athey, David Blei):

Access to digital information involves intermediaries. For online news, these are web pages and apps provided by news organizations, search engines, news aggregators, portals, and social media. For shopping, intermediaries are primarily e-commerce websites and apps. In such settings, users choose from a large set of alternatives, but the effective alternatives at a point in time are limited by the options presented on a web page or mobile screen. This project develops new methods for estimating user preferences as well as product characteristics from data about user choices, taking into account what options were presented (e.g. the links on the web page they visit). We build on recently developed computational methods for large‐scale Bayesian models, adapting the methods to incorporate the specific features of web browsing. We further incorporate insights from economics about how to ensure that estimates reflect fundamental preferences on the part of the users (causality versus correlation). We use the estimates to make counterfactual predictions about questions such as, what would happen if intermediaries changed the way they select news. This helps uncover the forces that shape news consumption today as well as incentives faced by news organizations in the future. We also consider applications to e-commerce.

The Impact of Digitization on Labor Markets, Product Quality, and Information (2017-2018; Susan Athey, Paul Oyer, Markus Mobius): 

Automation through both robotics and software is quickly diffusing, and digital platforms continue to expand, with profound implications for labor markets and democracy. First, the effect of automation on labor markets has been the subject of much debate, but little rigorous analysis. This project builds an economic model and uses it to evaluate conditions under which automation causes large increases in inequality or poverty. The project will develop empirical estimates of factors such as those that relate to preferences around transportation (which will be impacted by automation), including workers’ willingness to commute and move. Second, marketplaces and platforms serve as intermediaries between consumers and the goods and services they consume; they also determine who supplies those goods and services, and indirectly, their quality and composition. Platforms affect labor market opportunities of individuals, as well as labor market equilibrium. Platforms often provide flexibility for workers, and enable new approaches to incentivize quality. This project measures these effects in several industries, including services (ridesharing, dog-sitting, and rooms) and markets for news, which is fundamental to democracy. New algorithms for improving the performance of marketplaces as well as the efficiency of government will be developed and deployed, and their performance evaluated.

The Interdependence and Fragmentation of Life Experiences across Cyber-social Systems (2016-2018; Byron Reeves, Nilam Ram, Laura Carstensen):

Smartphones and laptop computers now allow multitasking among a greater range of experiences than has ever been possible. People switch between radically different content – from work to play to social relationships – and often within seconds. This means that understanding interchanges with cyber-social systems (e.g., health, social relationships, finances, shopping, transportation, work productivity, learning) will depend as much on stitching together experiences across domains as on examining experiences scattered within any single system. Health interactions influence social relationships, which influence financial decisions, which influence work productivity, and so on. Our project is about the interdependence of cyber-social systems across time domains (i.e., from seconds to years) and spatial locations. Many behaviors now emerge through digital experiences, producing data that can be accessed through portals and accumulated into a cohesive picture of individual life. In Phase 1 we will design, borrow and build software tools that gather, store, search and visualize experiences over time, paying attention to embedded security and privacy issues. In Phase 2 we will use those tools to test hypotheses and explore the interdependences among domains. We will emphasize the communication of data back to individuals and organizations so that each might use insights from cross-domain tracking to improve policies and practices in learning, commerce and social relationships.

Policy-Friendly Remote Access to Computer Resources: The Successor to SSH (2017-2018; Keith Winstein, David Mazieres, Ben Calvert, Erwin Lopez, Ashley Tolbert):

From its introduction in 1995, the Secure Shell (SSH) has become a ubiquitous tool for users to connect securely with networked and “cloud” servers. But as the importance of cybersecurity has increased in the last two decades, and as systems like the Secure Web and TLS have seen considerable evolution, SSH has yet to realize commensurate improvements in its manageability, auditability, or support of prudent security policies. This project is a collaboration between PIs at the Stanford Computer Science Department and in the Cyber Security organization of SLAC, a national laboratory that Stanford operates for the U.S. Department of Energy. The Computer Science PIs have prior experience developing SSH-like systems that have been deployed to millions of users. The SLAC PIs create cybersecurity policies that govern SLAC’s use of SSH, and implement these policies subject to federal cybersecurity regulations and oversight. Together, we will develop a successor to SSH that is responsive to today’s real-world cybersecurity concerns and deployable at security-conscious and policy-driven organizations. Relative to today’s SSH, the improvements will focus on three areas: policy-friendliness, allowing the implementation and analysis (including “what if?” questions) of cybersecurity policies governing authentication and authorization, security improvements, and usability improvements to encourage consistent and appropriate use. We will use SLAC as a motivating “launch customer”: if the new system can be welcomed and deployed in a security-conscious and regulated environment like SLAC, we believe it will see widespread use. 

Political Framing and its Propagation in Media (2017-2018; Dan Jurafsky, Matthew Gentzkow, Jure Leskovec, Jennifer Pan):

We propose to investigate political framing in digital media, using a novel combination of computational linguistics and machine learning tools to investigate key areas crucial for preserving democracy in the post-industrial world. These include the ability of governments or non-state actors to influence or undermine the democratic process through propaganda or agenda-setting, the way new media distinguish (or don’t distinguish) subjective opinions from objective data, how minority and majority groups are portrayed, and the way partisan frames emerge and diffuse. Our research makes use of data spanning different media, time periods, and content creators (e.g., journalists, ordinary citizens, politicians), including our own corpus of 50 billion news media articles and social media posts, as well as historical collections of media in multiple languages. Our project has the potential to significantly advance our understanding of the effect of cyber media on the political landscape as well as developing novel computational tools to help detect these latent influences on media.

Resilient and Robust Connectivity for Medical Devices in the Developing World (2016-2017; Keith Winstein, Leonore Herzenberg):

The practice of medicine increasingly relies on large quantities of data—often, gigabytes from imaging, genomics, or blood-based analysis. But the medical devices that collect this data are generally designed with the assumption that Internet access is robust and always on. In the developing world, this assumption breaks down. From Addis Ababa to Zomba, cellular networks are typically the best available connectivity, and these networks experience chronic brownouts and fluctuations. The result: blood-pressure monitors, flow cytometers, and MRI scanners all function poorly, unable to send data where it can be analyzed. We have developed a relationship with medical clinics operated by the U.S.-based Clinton Health Access Initiative (CHAI) in Kenya, Ethiopia, Zimbabwe, Malawi, and Uganda. We propose to kick-start a project to design and deploy a series of reliable Internet gateways boxes at these sites. We will use two main approaches: (1) a “robust tunnel” that masks brownouts with computer-networking techniques (e.g. aggressive retransmission and error-correction coding) to construct more-reliable connectivity out of multiple less-reliable cellular links, and (2) a “self-incentivizing network” that rewards anybody who successfully carries uploads from the medical facility to the Internet, by publicly offering to pay money in exchange for carrying an encrypted dataset to the cloud.

Secrecy of Sequential Decision-Making (2016-2017; Kuang Xu):

The increasing prevalence of large-scale ​surveillance and data collection infrastructures deployed by government agencies and private companies has brought global attention to the astonishing power enabled by modern cyber technologies. While such information appears to be revealing (e.g., a consumer's past browsing behavior may be indicative of the final purchase decision), we still lack a satisfactory understanding of the true value of the data collected, in terms of the extent to which it allows one to predict an individual's intention or future behavior using his or her past actions. ​This project aims to create a new mathematical framework to quantify the fundamental degree of information leakage asociated with an individual's sequential decision-making process, as well as design intelligent algorithms and decision-making policies that are capable of obfuscating an individual's future actions, even against a powerful data collector.

Self-Incentivizing Networks (2015-2016; Keith Winstein, Ramesh Johari):

We are developing the engineering and economic tools to enable self-incentivizing enclaves on the Internet, where entrepreneurs can add incrementally to the network's capacity and be rewarded for their contribution, however small, to encourage the buildout of connectivity in under-served areas. This problem cuts across the domains of congestion-control, traffic engineering, and wide-area routing and settlement on the Internet.

Social Media and Democracy (2017-2018; Frank Fukuyama, Nate Persily, Ashish Goel): 

Advances in technology and the rise of the Internet are upsetting the longstanding western political balance. Democracies around the world face growing threats from demagogues and populists who use social media to spread xenophobia, misinformation, and fear in furtherance of their agendas. The news media find their traditional role as a source of reliable information and a forum for reasoned debate diminished by changing business models and consumer preferences. And autocratic regimes exploit global information networks to undermine and disrupt democratic institutions. This project will undertake a program of research that will provide an empirical and algorithmic base for policy innovations to preserve democratic institutions in the face of these new challenges. It will work in conjunction with the new Global Digital Policy Incubator (GDPi) at the Center on Democracy, Development and the Rule of Law by producing a series of studies about both the nature of digital threats to democratic institutions, as well as comparative analysis of existing and emerging responses to this challenge. It will also leverage the school of engineering to explore technological responses where appropriate.

Stanford-China Track 2 Cyber Diplomacy (2015-2016; Herb Lin, David Relman):

It is well known that the cyber relationship between China and the United States is a source of great friction between the two nations.  As a step towards improving the cyber relationship, the CISAC/Hoover cyber program will seek to build a sustainable and ongoing dialogue with China, through an initial effort to conduct cooperative research on cybersecurity-related issues.  This cooperative effort involves two workshops, one held at a Chinese university and another at Stanford.  These workshops (and the resulting publications) should be regarded as a proof of concept that a productive dialogue between Stanford and the Chinese is possible, and funding from a variety of sources would be sought after this project to expand the dialogue and to make it more sustainable.

Uncovering Authoritarian Rule with Cyber Technology: Estimating the Prevalence of Collective Action and Repression in Authoritarian Regimes with Unstructured Digital Data (2016-2018; Jennifer Pan, John Duchi):

We aim to develop a methodology to generate the first rigorous scientific measure of a variable of paramount importance to academics and public policy makers worldwide ­­ the prevalence, location, and scale of collective action events and repression of these events in authoritarian regimes. There is on­going debate over whether cyber technologies threaten the survival of authoritarian regimes by facilitating collective action or whether authoritarian regimes are using cyber technologies to strengthen their rule. We propose to develop an independent measure of collective action and the regime’s repressive response to social mobilization by developing algorithms to detect these activities by using unstructured digital data, including images and text, generated by individuals who witness these events and publicly shared on social media platforms.