Welcome back to AAI Reads!
Welcome to our first The Alexandria Archives Reads! (AAI Reads!). The book that we’re highlighting is Weapons of Math Destruction: How Big Data Increases Inequality And Threatens Democracy. Written by Cathy O’Neil in 2016, it’s a great introduction to the way that algorithms impact humanity. Specifically, it redefines the “m” in WMD from “mass” to “math,” although the impacts of many of the algorithmic examples she explores could probably be categorized as both.
To structure our thoughts about the book, we’re going to capture the spirit of data through the lens of file formats. This time, Meghan will be our ‘TIFF’ reviewer. She (re)read the book in the last few weeks and can act as an uncompressed approach, with no loss of memory in her review. Paulina will be taking the “JPEG” perspective. A more compressed and slightly lossy take on the work. And lastly, Eric will be our ‘thumbnail’ review. An approach that focuses more on a general impression rather than details.
Meghan – TIFF review
Halfway through reading Weapons of Math Destruction, I realized…I’d read this book before. It came out during the first year of my PhD. This was a time when I was reading a LOT, but it was also a time of intense pressure and stress, so the specifics of what I read have–like the early drafts of my thesis–faded into a gentle haze. That said, on re-read, I still find a lot of value in this text, and some things that O’Neil speculated about have become entrenched in our day-to-day lives, despite her warnings.
Specifically, reading O’Neil’s concerns about digital advertising and data collection in light of the Facebook-Cambridge Analytica scandal (that occurred two years after publication) shows the text’s prescience and continuing importance. It also shows how little has changed for the positive concerning the ethics of collecting big data sets through social media.
O’Neil notes, concerning the application of algorithmic models to personal data,
‘…they [algorithms] send some people to Harvard, line others up for cheap loans or good jobs, and reduce jail sentences for certain lucky felons. But the point is not whether some people benefit. It’s that so many suffer.’
O’Neil illustrates how the people who benefit from WMDs do so at the expense of others. Those who suffer are, as she says, ‘collateral damage.’
Looking around over the past two years, I couldn’t help but connect these concepts of help and harm, benefit and ‘collateral damage’, to how data was used during the ongoing COVID-19 pandemic. Specifically, people used data to prevent many of the most vulnerable from receiving the same levels of medical care and access to help that those of privilege receive. Early on, O’Neil notes that,
‘The privileged, we’ll see time and again, are processed more by people, the masses by machines.`
While the technical aspects of O’Neil’s text may be starting to show their age, the conceptual issues are still relevant. This book made me angry, in the best possible way, about how people manipulate data to do harm. Here’s hoping I can hold on to that anger, and not let it go like I did on first read.
Paulina – JPEG review
I read this after consuming a number of other books about data, and it was really cool to see how the ideas from newer works drew from this book. You can see the inspiration in Data Feminism and Living in Data from O’Neil’s examples and, as books written after 2016, both take the ideas even further with regards to data as social justice.
Overall, I enjoyed the book and think it is still worth reading, though aspects of its initial publication date, 2016, are starting to show. For example, O’Neil uses the term “tribe” or “tribal” in a way that could probably be removed.
Were O’Neil interested in addressing that issue, in an update to the book or or in a new one, I’d love to see her discuss: Indigenous Data Sovereignty and how it intersects with her WMD concept; the ways that WMDs specifically harm Indigenous communities; or how historical non-algorithmic data collection in such communities set the stage for modern WMDs.
However, an extremely relevant point from the book is her way of assessing what is and is not a WMD. The checklist she uses is perfect for assessing any data set, big or small, in archaeology or in any discipline.
Specifically, the focus on opacity, scale, and damage are something we should be thinking about when we, as archaeologists, consider understanding the past. We should always spend time examining our proxies and mathematical models to see if they are actually testing what we think they are.
Eric – Thumbnail review
As illustrated by the low-quality image of the book’s cover, my recollections of Weapons of Math Destruction are pretty hazy. My mind has done some heavy and “lossy data compression” about the contents. All this is to say that I’ve forgotten the specifics, and am left with a pretty general and distorted understanding of the book!
Mainly, I remember it as one of the first major critiques (that I read) of the social impacts of “Big Data.” It explored systemic (often racist) biases in data collection and organization and the design of algorithms that make use of such data. It incorporated examples from domains as varied as policing and parole, credit ratings, home loans, employment resume review, health insurance, and more that impact people’s daily lives.
These points still resonate, and still shape my sense of needs and priorities, even though I have forgotten the details of O’Neils’ arguments and examples. As time passes, I’m sure I’m increasingly mixing the points made by Weapons of Math Destruction with points made in another key book on this topic, Automating Inequality by Virginia Eubanks.
Nevertheless, these books highlight the social need to treat data and algorithms with appropriate skepticism. Data and algorithms only have an appearance of neutral objectivity, but they are in fact very human products based on contestable assumptions, implicit (and sometimes explicit) biases, and can be used and abused in ways that cause tremendous harm. As educators, we have a huge social responsibility to raise awareness of how to evaluate and critique data and the (mis)uses of data. These powerfully argued points made by these books led us to respond by launching the “data literacy” program.
