The Brown Institute of Stanford and Columbia universities announces its Magic Grant winners

The Brown Institute for Media Innovation, a collaboration between Stanford University’s School of Engineering and Columbia Journalism School, is awarding $1 million in funding for 15 projects as part of the 2020-21 Magic Grant program. Each year, the Brown Institute awards grants to foster new tools and modes of expression, and to create stories that escape the bounds of page and screen.

This year’s awards include 10 Magic Grants and four seed grants. Each project addresses an important contemporary issue, be it political, cultural or technical.

For example, one project will undertake a first of its kind look at the interplay between media coverage of the criminal justice system, public opinion and legislation – a study that provides historical framing for reporters covering the demonstrations and events following the killing of George Floyd. Another grant will forge a new partnership to support local data journalism, compiling new local data sets and experimenting with peer mentoring.

The institute was established in 2012 by a gift from Helen Gurley Brown. David and Helen Gurley Brown believed that magic happens when innovative technology is combined with great content and talented people are given the opportunity to explore and create new ways to inform and entertain.

The following is a complete list of Magic Grants funded by the Brown Institute for 2020‑21:

Local Live(s)

Lauren Peace and Carissa Quiambao, MS ’20 at Columbia Journalism School, and McArdle Hankin, masters candidate in communication at Stanford

Local Live(s) aims to create a dialogue between journalists in local news organizations and the communities they serve. Through live events (held virtually) with journalists telling the stories behind their reporting, they invite their community into “a world that they have only observed from above the fold.” By establishing a rapport between journalists and community members, the team is hoping to help restore trust in local news. Complementary to the live events, the team will produce a handbook for local newsrooms on how to curate, design and launch live journalism events.

COVID Local News Collaboration

Members from Big Local News and OpenNews

The COVID Local News Collaboration is a partnership between OpenNews and Big Local News. Its aim is to help journalists tell deeper, data-driven stories that assist communities responding to COVID-19. The Big Local News platform, funded as a Magic Grant in 2018-19, has become an important platform for data related to COVID-19, offering information not easily accessible to journalists otherwise. OpenNews has been exploring new kinds of peer mentoring for data journalists and the joint creation of story guides. As a team, these groups will identify COVID-19 stories that are most needed by communities across the country and help local newsrooms tell them through data and visualizations.

Wolf Pack: How Media Coverage of Criminal Justice Enabled Mass Incarceration

Carroll Bogert, president of The Marshall Project and visiting scholar at Columbia, LynNell Hancock, professor of journalism at Columbia Journalism School, and Shanto Iyengar, professor in political science at Stanford

Wolf Pack examines for the first time how media coverage of criminal justice helped turn the United States into the most incarcerated country in the world. In collaboration with data scientists at both Stanford and Columbia, the team will prepare a database of national and local media coverage of criminal justice and, through natural language processing techniques, reveal the framing and narrative structure, and especially the racial and ethnic clues, embedded in this coverage. The lessons from this project will inevitably provide historical framing for contemporary events.

Self-Moderating Online Focus Groups and Deliberation

Lodewijk Gelauff and Sukolsak Sakshuwong, both doctoral candidates in management science and engineering at Stanford

Due to the precautions we are taking around COVID-19, our society has rapidly shifted from traditional in-person meetings to online meetings. Building a tool for better online deliberation through an automated moderator that provides equitable, respectful, and constructive conversation is the focus of Self-Moderating Online Focus Groups and Deliberation. To support deliberative democracy, the team will develop a scalable online platform that addresses challenges, such as a small group dominating the conversation, a single topic that absorbs too much time, and a biased moderator.

Improving Remote Learning via Hierarchical Decomposition of Instructional Videos

Anh Truong and Chien-Yi Chang, both doctoral candidates in computer science at Stanford

Increasingly, we turn to instructional videos for accomplishing everyday tasks and learning new skills such as sewing a face mask, cooking and household repairs. Improving Remote Learning via Hierarchical Decomposition of Instructional Videos will facilitate the creation of instructional video to allow better navigation by hierarchically segmenting tasks into steps and providing voice-based navigation commands for accessing the steps. This will be accomplished with algorithms that can automatically learn shared action steps from videos across different tasks by explicitly leveraging the conjugate constraints between actions and states.

Sports Illustrated: Enabling Machines to Understand and Describe Tennis Matches

Sumith Kulal and Haotian Zhang, both doctoral candidates in computer science at Stanford

Major sports events like the World Cup and Wimbledon attract millions of viewers. Sports Illustrated: Enabling Machines to Understand and Describe Tennis Matches aims to build a higher-level abstraction for understanding and describing sports videos (e.g. soccer, tennis) by enabling three types of applications (1) Retrieval: retrieve similar action from a video database (2) Text: synthesize text/speech to describe details of the action (3) Edit: edit particular micro-movement and re-render the video. This team hopes to provide a better viewing experience for audiences and powerful game analysis for athletes.

Leitmotif: Location-Driven Audio Storytelling

Jacob Ritchie, doctoral candidate in computer science at Stanford, and Jean Costa, postdoctoral researcher in computer science at Stanford

Delivering dynamic audio storytelling, using geolocation, to connect the user to stories of people, places and things that they walk by, is the Leitmotif: Location-Driven Audio Storytelling system this team is undertaking. The team will create software to enable the generation of location-specific audio stories, and a paired smartphone application to allow users to consume audio content. Users will be able to preview and select audio stories of interest to listen to as they move through their physical world.

COVID-19 FOIA Repository

Derek Kravitz, adjunct assistant professor at Columbia Journalism School

The COVID-19 FOIA Repository began as a 2019-2020 Magic Grant that shifted its focus to reporting on how local governments were responding to the COVID-19 pandemic. The team will continue its work, issuing targeted Freedom of Information Act requests to build a nationwide repository of COVID-19 related emails between city, county and state officials. So far, the project has requested records from more than 200 agencies in 44 states, and received 16 substantive responses from 10 states totaling more than 50,000 pages and hundreds of attachments, in data and PDF forms. The COVID-19 FOIA Repository will make the full document sets searchable and available to news organizations, academics and the public.

A Data ‘Concierge Service’ for Climate and Resilience Journalism

Francesco Fiondella, Rémi Cousin, Ashley Curtis and Weston Anderson, research and communications staff at International Research Institute for Climate and Society at Columbia University

The Data ‘Concierge Service’ for Climate and Resilience Journalism seeks to create a quick-response, concierge-style data service to help journalists access Columbia’s vast climate and environmental data repositories and connect them to its climate scientists. This service addresses two key challenges: first, many journalists aren’t aware of the massive quantities of climate data and analytical tools available to enhance their reporting; and second, it facilitates expert assistance to identify the most relevant and reliable data sets for their stories as well as help downloading the data.

The COVID Financial Crisis

Nick Thieme, Emily Merwin DiRico, Kenneth Foskett, Jennifer Peebles and John Perry, staff at the Atlanta Journal Constitution

The COVID Financial Crisis project will create an investigative series into the economic impacts of the COVID-19 crisis on Georgia’s citizens, companies, and municipal governments. Focusing on personal and corporate bankruptcies and municipal defaults, the project will measure the financial damage to the people of Georgia, allowing a comparison between this crisis and previous ones. Are black Americans more likely to file for bankruptcy because of COVID-19 than white Americans? Are rural municipalities more likely to default on bond obligations than urban municipalities? What industry factors are most associated with bankruptcy in the COVID-19 era?

Seed grants

In addition to Magic Grants, the Brown Institute is providing seed funds to the following initiatives to assist in prototyping and early project development:

Project Immerse

Lance Weiler, The Digital Storytelling Lab at Columbia University and Nicholas Fortugno, Playmatics

In a time of deep fakes, conspiracy theories, and AI-driven writing and social networking scams, we are surrounded by manipulative and deceptive technologies that, in the wrong hands, could constitute an existential threat to our society. Project Immerse will apply a learning methodology the team has called “attract and educate” to create an immersive educational experience using existing technologies, specifically the Miro collaborative platform and Zoom, to tell stories featuring these malicious techniques. Building an anthology series akin to Black Mirror, the project will create “episodes” that deal with the ways misinformation manifests on the internet, drawn out to their potential dramatic and powerful conclusions. Each episode will be a standalone piece of entertainment built-in web-native technologies, but at the end of each episode, Project Immerse takes users behind the scenes and gives them access to the actual tools to play with on their own.

The Right to be Forgotten in U.S. Newsrooms

Sarah Collins, MS ’20 at Columbia Journalism School

The Right to be Forgotten in U.S. Newsrooms will research and create guidelines for newsrooms to “unpublish” obsolete content – such as stories about crimes that are now sealed by courts, reports about arrests that never led to charges, and factually inaccurate information. Through consultation with experts in philosophy, ethics, journalism, civic nonprofits, international relations, law, government, internet technology and big data, the project will consider not just situations when content should be unpublished, but also examine concrete technical steps newsrooms can follow to securely and correctly remove the articles in question.

Community Networking for Community Storytelling in Appalachia

Houman Saberi, Greta Byrum, Raul Enriquez and April Jarocki, Community Tech NY

The Community Networking for Community Storytelling in Appalachia will build on their existing efforts to develop community-owned communications infrastructure and use it as a platform for storytelling. Through this, the project will highlight the ongoing efforts of the residents of the Clearfork Valley in rural Eastern Tennessee to build transformational resilience in the face of the impacts of the coal industry and a changing climate.

Automatic Identification of Online Harassment of Women Journalists

Julia Hirschberg, professor of computer science at Columbia University, and Sarah Iva Levitan, postdoctoral research scientist at Columbia University

The Automatic Identification of Online Harassment of Women Journalists will develop methods to identify abusive and hateful speech targeting female journalists on social media. The project will work with journalists to collect a large-scale corpus of private harassment messages received by journalists on Twitter, and to develop an easily-employed annotation method for labeling messages by degree of observed harassment, collecting self-labeled data from journalists. They will use automated natural language processing methods to extract features from these messages and to build machine learning classifiers to distinguish between harassing or abusive and neutral messages. These classifiers will be integrated into a tool for journalists to use to manage their Twitter feeds to identify and segregate these messages.

Editor’s note: In an earlier version of this story, Lodewijk Gelauff and Sukolsak Sakshuwong were listed as doctoral candidates in computer science; however they are doctoral candidates in management science and engineering. The article has been updated with this correction.