Reviews
Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor, by Virginia Eubanks
tuesday_evening's review
dark
emotional
informative
sad
medium-paced
5.0
Content Warnings
Moderate: Racism and Classism
phire's review
4.0
This was more of a 3.5 or maybe a 3.75.
Things this book was good at: tracing the history of data discrimination from scientific charity to modern algorithms, painting a vivid picture of the human costs of poorly implemented automated systems, and contextualizing the three case studies in their respective political environments.
Things this book struggled with: occasionally getting caught up in folksy flourishes, lacking in-depth exploration of intersectional concerns, and lacking in-depth analysis of the "so what" component of the issue. The last chapter provides an overview of the examples discussed and attempts to make a case for why both liberals and conservatives should care that data-driven social services hurt the most marginalized. I found this the weakest chapter, especially the section tying data justice to so-called national values, and while reading it I couldn't help thinking of that "I Don't Know How To Explain To You That You Should Care About Other People" article. I think framing this conversation around how it fits into existing US partisan paradigms cedes more ground than is necessary, as it precludes considering an entirely new paradigm around this transnational issue. The conclusion is much sharper and more incisive in highlighting our collective responsibility to undo a problem that was deliberately caused by conscious choices.
You should read it if: you are fascinated (if repulsed) by detailed explorations of how politicians screw with the lives of ordinary people. I especially appreciated the discussion of LA's Skid Row and the housing matching system.
Things this book was good at: tracing the history of data discrimination from scientific charity to modern algorithms, painting a vivid picture of the human costs of poorly implemented automated systems, and contextualizing the three case studies in their respective political environments.
Things this book struggled with: occasionally getting caught up in folksy flourishes, lacking in-depth exploration of intersectional concerns, and lacking in-depth analysis of the "so what" component of the issue. The last chapter provides an overview of the examples discussed and attempts to make a case for why both liberals and conservatives should care that data-driven social services hurt the most marginalized. I found this the weakest chapter, especially the section tying data justice to so-called national values, and while reading it I couldn't help thinking of that "I Don't Know How To Explain To You That You Should Care About Other People" article. I think framing this conversation around how it fits into existing US partisan paradigms cedes more ground than is necessary, as it precludes considering an entirely new paradigm around this transnational issue. The conclusion is much sharper and more incisive in highlighting our collective responsibility to undo a problem that was deliberately caused by conscious choices.
You should read it if: you are fascinated (if repulsed) by detailed explorations of how politicians screw with the lives of ordinary people. I especially appreciated the discussion of LA's Skid Row and the housing matching system.
ifeustel's review
5.0
Programmers, Public Health officials, social workers, lawmakers - if you have not read this book or engaged with this topic you are late to the party.
Beginning with an anecdote about a personal experience on the trails of accessing healthcare in today’s systems, Eubanks then walks the reader through three case studies of the relationship between technology and public services.
With engaging prose and analytical clarity, Eubanks draws a clear line between the history of surveillance technologies that we would consider commonplace in everyone’s lives and the trajectory of these technologies in the future, by examining how algorithms are currently being used to police the poor.
Eubanks’ well researched and engaging book makes the case for keeping the human element as an important part of our public services. Algorithms are not pure logic - they betray the biases and prejudices of their creators.
Do algorithms divert people from services that they need? Do they ask the right questions to include the whole population in decisions about allocating resources? Do they ask people to give up their privacy and agency unduly in exchange for public services? Do they protect people’s information?
All important questions which you can expect to examine in all their complexity from both angles in #Automating Inequality by Virginia Eubanks
Beginning with an anecdote about a personal experience on the trails of accessing healthcare in today’s systems, Eubanks then walks the reader through three case studies of the relationship between technology and public services.
With engaging prose and analytical clarity, Eubanks draws a clear line between the history of surveillance technologies that we would consider commonplace in everyone’s lives and the trajectory of these technologies in the future, by examining how algorithms are currently being used to police the poor.
Eubanks’ well researched and engaging book makes the case for keeping the human element as an important part of our public services. Algorithms are not pure logic - they betray the biases and prejudices of their creators.
Do algorithms divert people from services that they need? Do they ask the right questions to include the whole population in decisions about allocating resources? Do they ask people to give up their privacy and agency unduly in exchange for public services? Do they protect people’s information?
All important questions which you can expect to examine in all their complexity from both angles in #Automating Inequality by Virginia Eubanks
quigonchuy's review
5.0
This is a phenomenal read. Shines a light on how we tend to assume that algorithms and other high tech tools are "blind" and don't have the bias that humans can. Except they can, and they do. They are programmed to, whether intentionally or not. But they are not the neutral arbiters we often take them to be. The automating of public benefit eligibility, policing, and other government programs has negative results that are often overlooked because they are being applied or determined by technology instead of people. Very illuminating and informative.
ameliaholcomb's review
4.0
I appreciated the three areas of focus and the stories that Eubanks chose and how she tied them all together. This is another really important book in discussing the injustices of the US welfare system.
vladco's review
2.0
I spent my time reading this mediating between the angel on one shoulder, telling me "she's not wrong, and that was a very good point, and well made!" and the devil on my other shoulder telling me, "Like, duh. This book could be summarized as 'the bureaucracy is expanding to meet the expanding needs of the bureaucracy,' starting from a foundation of 'the nature of man is evil; what is good in him is artificial.'"
There are wonderful parts of this book (particularly the first and last chapters). The rest of it feels like tiresome examples to show how the bureaucracies humans design can be inhumane, and how software systems can deliver those painful payloads at scale. I feel like the book is missing the deeper questions: how do we design humane systems that scale to the needs of the governed? How do we balance between the often-conflicting goals of eliminating waste, inefficiency, and fraud while still delivering high-touch service that treats people with dignity, respect, mercy, and kindness? What are the principles of good design? What are the anti-patterns to be avoided? The book fails to look at this from a satellite perspective, missing the big patterns and leverage opportunities.
There are wonderful parts of this book (particularly the first and last chapters). The rest of it feels like tiresome examples to show how the bureaucracies humans design can be inhumane, and how software systems can deliver those painful payloads at scale. I feel like the book is missing the deeper questions: how do we design humane systems that scale to the needs of the governed? How do we balance between the often-conflicting goals of eliminating waste, inefficiency, and fraud while still delivering high-touch service that treats people with dignity, respect, mercy, and kindness? What are the principles of good design? What are the anti-patterns to be avoided? The book fails to look at this from a satellite perspective, missing the big patterns and leverage opportunities.
leaton01's review
5.0
The supposed quest to create more efficient systems within government programs through automation and algorithms--particularly those that focus on social welfare--is not so much a cost-saving, efficient, and effective approach to caring for society's most vulnerable but rather a means of making the process harder, more-complicated, and near-impossible to challenge for those who suffer from the many bugs (or features as the case may be) of the technologies being used. Starting first with a look at the rise of poorhouses and the ways in which the poor have been penalized and punished in US history, Eubanks then moves into looking at the introduction of more technological solutions that are often purported to improve services, save money, and reduce fraud and abuse. So often, as Eubanks shows in case after case, the reduction in fraud and abuse more or less comes at the cost of many more people being accused of fraud and abuse (false positives) that come to have many long-term ramifications, including destroying progress in people's lives or limiting their ability to get future help when needed. As she emphasizes, the many different tools used to police the poor in many ways perpetuate their status and keep them from escaping poverty. The introduction of algorithms and programs that in many ways perpetuate negative stereotypes about the poor, only exacerbate it. What I found most powerful about Eubanks works was the ways in which she helps the reader understand how the technology continues to appear neutral but that the assumptions of the programmers and the ease with which technology moves from a "recommendation" to double-check a potential problem to "evidence" of a problem creates an system that produces bias just by the mere process of producing data and reports on that data. The best example of this is the ways in which in many systems, reports of abuse (real or otherwise) increase the chances in which future reports of abuse are deemed more serious--even though, the reporting mechanism might be entirely arbitrary. Anyone interested in social justice, social work, or understanding how being poor subjects one to increasing criminalization should check this book out.