Nonfiction November book review:
Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy by Cathy O'Neil, (6h 23 min, read by the author)
This book reads like a crazy conspiracy theory - hidden algorithms that you have no idea about are controlling your life. Your credit score, your college admissions, your insurance rates. Does it seem reasonable for companies to base rates on mathematical models? How about if you get to keep your job?
The author was a math professor who joined Wall Street and began working for hedge funds, developing algorithms for making money. After the financial crash, she recognized the damage the algorithms had wrought. Looking further, she concluded that these algorithms or models are biased against the poorest and contribute to keeping them poor.
The story of the teacher, who by all accounts was an excellent teacher (parents and principal and colleagues) who was fired based on the hidden criteria intended to weed out the poor teachers is an example O'Neil provides to support her concern about these Weapons of Math Destruction. (lol, clever title). Any time an algorithm has characteristics of opacity (those affected can't see the criteria), scale (how widely it is applied) and damage (when factors contribute to incarceration or poverty cycles) she calls them WMD.
The examples were fascinating and scary. These algorithms are why 17 year old males have crazy expensive car insurance rates, why your Facebook feed can drive you batty, and can be as extreme as deciding who to hire based on your social media followers. Deciding things by proxy - basing the individual on the group characteristics, can be what makes assumptions dangerous.
Real life examples being used to make the point, from the small to the large, make this an enjoyable read, much in the style of Malcolm Gladwell. Plus, getting to read a book written by a smart, math female was awesome.
No comments:
Post a Comment
Thanks for commenting, so nice of you to visit.
(I'll try without the letters for a while - so please dont be a spammer! Let's try no anonymous users)