Periodic Reporting for period 1 - DataStories (DataStories: Making Use of Interpretive Judgments of Data Creators)
Reporting period: 2019-06-01 to 2021-05-31
The DataStories project has asked: What can we learn from understanding the range of interpretive judgments that appear in a dataset?
In responding to this question, DataStories demonstrates the potential of structural data analysis to a diverse audience of humanist scholars and students.
To inspire and engage the broadest possible audience, Dr. Melanie Feinberg, the DataStories researcher, developed a unique, innovative approach based on narrative accounts of concrete, everyday data situations. These engaging narratives examine and interrogate common experiences such as stepping on a scale, buying groceries in a foreign supermarket, or reacting to a misspelling of one’s name. Through these narratives, the DataStories project provides an account of data as a form of human expression. The DataStories project establishes that, if we hope to act responsibly with data—whether we are collecting it, aggregating it, manipulating it, interpreting it, or making decisions with it—we need to appreciate this human character.
In responding to this question, DataStories demonstrates the potential of structural data analysis to a diverse audience of humanist scholars and students.
To inspire and engage the broadest possible audience, Dr. Melanie Feinberg, the DataStories researcher, developed a unique, innovative approach based on narrative accounts of concrete, everyday data situations. These engaging narratives examine and interrogate common experiences such as stepping on a scale, buying groceries in a foreign supermarket, or reacting to a misspelling of one’s name. Through these narratives, the DataStories project provides an account of data as a form of human expression. The DataStories project establishes that, if we hope to act responsibly with data—whether we are collecting it, aggregating it, manipulating it, interpreting it, or making decisions with it—we need to appreciate this human character.
As the primary output of the DataStories project, Dr. Feinberg has completed a book manuscript. The book, tentatively titled Rambles Through Everyday Data: A Travelogue and Field Guide, is under contract with MIT Press. The book demonstrates a practical, critical, and generative mode of thinking about data: its creation, management, aggregation, and use. Concurrently, the book establishes the widespread applicability of information science concepts for a diverse array of readers, across varied disciplinary backgrounds.
Each of the book’s seven chapters pairs an engaging, self-contained main essay—the travelogue—with a scholarly companion essay—the field guide. Each main essay begins with an anecdote, such as a visit to a library, running out of butter, or cooking rice on a different stove. Through sustained reflection on these everyday stories, the book argues that, to understand the power and pitfalls of data science, we must attend to the data itself, and not just the algorithms that we use to manipulate the data.
The book’s stories coalesce around the following themes:
• Data—whether collected and processed by machines or collected and processed by people—is the product of many acts of human interpretation. No system, manual or automated, can fully contain the dynamic creativity of all these aggregated data creators. Data, therefore, will always retain some ambiguity.
• By paying attention to how data works structurally, we can better understand this ambiguity: where it arises, how it manifests, and what we might want to do about it.
• We can, sometimes, constrain data ambiguity. These efforts always fail in some respects, and time undoes them, but we may want to employ them nonetheless.
• Like poetry, or music, or any form of human communication, data is reinterpreted—in a sense, remade—every time that it is used. Our engagement with its ambiguity is, therefore, an ongoing process. Data is a story that we read and reread; it is not an equation to be solved.
With its uniquely accessible, readable approach, the book aims to reach a wide audience of scholars, students, and the general public.
Each of the book’s seven chapters pairs an engaging, self-contained main essay—the travelogue—with a scholarly companion essay—the field guide. Each main essay begins with an anecdote, such as a visit to a library, running out of butter, or cooking rice on a different stove. Through sustained reflection on these everyday stories, the book argues that, to understand the power and pitfalls of data science, we must attend to the data itself, and not just the algorithms that we use to manipulate the data.
The book’s stories coalesce around the following themes:
• Data—whether collected and processed by machines or collected and processed by people—is the product of many acts of human interpretation. No system, manual or automated, can fully contain the dynamic creativity of all these aggregated data creators. Data, therefore, will always retain some ambiguity.
• By paying attention to how data works structurally, we can better understand this ambiguity: where it arises, how it manifests, and what we might want to do about it.
• We can, sometimes, constrain data ambiguity. These efforts always fail in some respects, and time undoes them, but we may want to employ them nonetheless.
• Like poetry, or music, or any form of human communication, data is reinterpreted—in a sense, remade—every time that it is used. Our engagement with its ambiguity is, therefore, an ongoing process. Data is a story that we read and reread; it is not an equation to be solved.
With its uniquely accessible, readable approach, the book aims to reach a wide audience of scholars, students, and the general public.
The book will be of particular interest to humanities scholars who seek to explain, critique, and design data and its accompanying infrastructures.
The book is notable in its equal appeal to scholars, students, and the general public.
The book’s emphasis on meticulous explication of everyday examples and its approachable style makes it especially appropriate for university students. It can be used for introductory courses in disciplines such as data science, digital humanities, media studies, and information science.
The book is accessible to the non-scholar and should engage the general reader. As stories of algorithmic bias and misinformation appear in the popular press—for instance, male applicants for the AppleCard being given higher credit limits than their wives, bail being set lower for white defendants than black ones, dangerous posts allowed to proliferate on social media — a growing audience seeks to understand how data is being used to make decisions. The intense focus on testing data to comprehend and manage the coronavirus pandemic only magnifies these existing concerns. The challenges that we see with understanding what coronavirus tests mean are not unusual; they are typical. But we don’t need to be epidemiologists to understand the basic contours of this situation. A similar story arises when we go looking for baguettes in Austin, Texas, and wind up at a bakery called Tous Les Jours, where the bread is authentically French—just not in the way that we might expect (it is French by way of Korea).
This book demonstrates how the same ambiguities and uncertainties that we encounter in our journeys through everyday data persist—and are even exacerbated—in conditions of algorithmically mediated prediction. We’ve all been to a bakery; we’ve all looked at a scale; we’ve all struggled to understand whether two blenders on Amazon are actually “the same” product. By thinking through these common experiences with depth and precision, the book will help general readers make similar use of their own everyday data knowledge.
The book is notable in its equal appeal to scholars, students, and the general public.
The book’s emphasis on meticulous explication of everyday examples and its approachable style makes it especially appropriate for university students. It can be used for introductory courses in disciplines such as data science, digital humanities, media studies, and information science.
The book is accessible to the non-scholar and should engage the general reader. As stories of algorithmic bias and misinformation appear in the popular press—for instance, male applicants for the AppleCard being given higher credit limits than their wives, bail being set lower for white defendants than black ones, dangerous posts allowed to proliferate on social media — a growing audience seeks to understand how data is being used to make decisions. The intense focus on testing data to comprehend and manage the coronavirus pandemic only magnifies these existing concerns. The challenges that we see with understanding what coronavirus tests mean are not unusual; they are typical. But we don’t need to be epidemiologists to understand the basic contours of this situation. A similar story arises when we go looking for baguettes in Austin, Texas, and wind up at a bakery called Tous Les Jours, where the bread is authentically French—just not in the way that we might expect (it is French by way of Korea).
This book demonstrates how the same ambiguities and uncertainties that we encounter in our journeys through everyday data persist—and are even exacerbated—in conditions of algorithmically mediated prediction. We’ve all been to a bakery; we’ve all looked at a scale; we’ve all struggled to understand whether two blenders on Amazon are actually “the same” product. By thinking through these common experiences with depth and precision, the book will help general readers make similar use of their own everyday data knowledge.