The New Oil

About the Author & the Book

Cathy O’Neil is an American mathematician and data scientist. She got a Ph.D in math from Harvard, and later taught at MIT. In 2007 she left academia to work in the finance industry, an experience she talks about in the book that left her disillusioned with the role of data collection and algorithms and the way that they can harm the outliers. This ultimately led to her publication of Weapons of Math Destruction in 2016.

The saying goes that if you want to cook an omelet, you have to break a few eggs. Weapons of Math Destruction focuses on those eggs who have become casualties on the way to algorithmically modernize the world, using big data to make decisions that are – on the surface – more objective, fair, and accurate. However, O’Neil explores how this is frequently not the case and the flaws with our current approaches to using Big Data to this end.

The Good

After my last brutal book review, this title came like a cool, refreshing breeze on a nice spring day. It is full of information – both factual and anecdotal – but written in plain English. Actual, plain English. The kind I’m writing this blog in, or the kind I would use to explain to my coworkers why I value my privacy. The book is chock full of personal stories which make it easy and entertaining to listen to – who isn’t interested in a story? – but is also full of research, facts, and numbers, again in a totally digestible way. The author never launches off into long explanations of research methodology or bores the reader with elaborate and hard-to-follow math problems, but she’s quick to cite her sources on her claims.

I honestly don’t have much to say in the way of good things, but not in a bad way. It’s a good book, accessibly-written, and full of both examples and figures to back up the information in it. Anything more would you require you to read the book yourself so we could discuss the finer topics she touches on.

The Bad

My biggest complaint with this author is her complete and utter disregard for privacy as even a thought. Literally. Unless I’m forgetting something, I don’t think the topic of privacy ever came up once in her discussion, and if it did it was brushed away very quickly, probably with empty, disproven promises about “anonymizing” data. In fact, I was extremely disappointed in her final argument: “let’s make algorithms better.” At no point did she ever defend or even touch on the idea that maybe algorithms aren’t the way to go, which I suppose makes sense given her background. Instead, she argues that algorithms must simply be better made – proper metrics, correct data, the ability to add feedback to improve the algorithm, etc. To be fair, that is a valid point. If I set my GPS to take me to the movie theater and it instead takes me to city hall, that’s not a very useful GPS is it? If you’re going to build something, you should make sure that it’s accurate and effective.

That said, never once does she address the question of “do we even need algorithms?” Truth be told, I think most of us do want algorithms in some form or fashion: most of us want Netflix to suggest shows we might like or Spotify to show us new bands we might be into. What we also want is agency and modesty. I don’t want Spotify to surreptitiously turn on my microphone because they think that it’ll give them better data, or for Netflix to use the gyroscope in my phone to guess that I’m in a bus when I denied them location permissions. They need to work within the bounds I’ve agreed to. And furthermore, they need to respect if I set a boundary that precludes them entirely. I also wrote about that recently, about how Big Tech likes to follow us around the web and stalk us against our will. It’s all well and good to say that algorithms should be better and more accurate and open to improvement, but we can’t ignore that these algorithms should be optional in the first place. Really optional, not the illusory sandbox I explored in the linked blog post. There needs to be agency to opt out and modesty to collect only the necessary data. Of course, any experienced reader knows that for algorithms, more data improves it, and therefore these demands are at odds. O’Neil never addresses these concerns. She’s quite optimistic, in that regard. If we just all act in good faith, algorithms can change the world. I agree. Too bad that corporations – not just in Big Tech but everywhere – have a long, “set your watch by it” history of acting to increase the profits, even when they know it contradicts “good faith.” These are things that she completely ignores, but to be fair perhaps these are also massive topics that warrant a completely separate book.

Final Verdict

So why am I reviewing a book that barely talks about privacy? Because I think it’s very, very closely related. Privacy is a vast topic that truthfully touches – in some way – nearly every other area of our lives, many of which I very rarely discuss: free speech, economics, bullying, mental health, psychology, philosophy, fake news, cryptocurrency, etc. I purposely keep my focus very narrow in an effort to attempt to increase my effectiveness and – ironically – appeal to as broad an audience as possible, but algorithms don’t just touch privacy, the overlap is so heavy it makes me think of a footballer getting dogpiled in a game. Understanding algorithms is not just a crucial part of making sense of the world we occupy today, it’s also crucial to understanding why your privacy matters and how all this data can be weaponized against you. So while this book may not be directly privacy related – like Extreme Privacy or The Age of Surveillance Capitalism – it’s still so close that I think it would benefit a lot of my readers to give it a read. If nothing else, you’ll walk away more informed about the potential – both good and bad – for this new part of our world.

You can learn more about Weapons of Math Destruction by Cathy O’Neil here.

You can find more recommended services and programs at TheNewOil.org, and you can find our other content across the web here or support our work in a variety of ways here. You can also leave a comment on this post here: Discuss...