I’m finally doing it.
I’m going to keep up with Bryan Alexander’s book club. This one should be easy–I started reading this book over the summer after hearing O’Neil speak at NYSCIO. I got about halfway through before vacation and the ensuing chaos of the start of the semester overwhelmed me; luckily the timing of the book club is such that I should be able to carve out a little time to read it. I believe this is a critically important book for everyone in IT, higher education, or anyone whose life is affected by data (i.e., everyone.) Bryan did an excellent job summarizing the introduction and first chapter, so there’s little point in me reiterating it verbatim; I’ll just add a few of my personal impressions.
O’Neil talks about her career as a number theorist working at Barnard (close to home to me as one of my long-time bosses and mentors was a number theorist by training), who then left for D. E. Shaw (hedge fund), leaving after the financial crash of 2008 to become a “data scientist” and ally with with Occupy Wall Street movement. I appreciate O’Neil’s candidness about her career path, because it underscores her message of algorithmic transparency and understanding the biases that data scientists bring to her work.
I was intrigued by the story of Washington, D.C.’s emphasis on standardized testing to evaluate teachers and students, and fire, “low-performing” teachers; both because I’ve always been suspicious of the methods and intentions of former D.C. schools superintendent Michelle Rhee, and as the spouse of a high school teacher I see that the quantification of her work is more challenging and subtle than many think. Teaching the second half of the AP US History curriculum is great; but if the students don’t do well on the AP test, was it her fault or her partner instructor? What if the elementary and middle schools in her district are doing a worse job of preparing students for her class, and test scores go down because of that? Do we even know how much influence a teacher can actually have on academic performance, or do factors like socioeconomic brackets, racial tensions, or plain hunger outweigh what a teacher does in the classroom?
And that will lead to one of the major messages of this book–what are these algorithms (“Weapons of Math Destruction”) making decisions on our behalf, how do we understand the implicit bias of any algorithm, and what do we do about it?
O’Neil’s writing style is excellent–explaining incredibly complex ideas in simple terms, and making us feel smart. It’s interesting to think this was written before the 2016 election, and all the things that can make us think about data.
This is a vitally important book and one I strongly recommend. It’s not too late to jump on the book club and follow along, and I hope you do.