Written by one of the more articulate practitioners of Data Science, "Weapons of Math Destruction" questions the use of algorithms and analytics in a range of domains including education, the justice system and politics. The essential issue seems to be that some of these models have two flaws: (1) the results are to some extent self-fulfilling (e.g. Recidivism models that inflict longer sentences on some offenders, when time in jail makes people more likely to reoffend) (2) the models are not adjusted in light of incorrect predictions (e.g. Models of teacher performance that don't respond to huge year-to-year swings in scores and incorporate no feedbacks from students' later life experience).
Couple this with a peculiarly American willingness to screw its citizens over with unpredictable work schedules and docking of pay when "wellness" targets are not met, etc, and you have a recipe for so-called innovation leaving us much worse off.
This book is a great introduction to the moral dimensions of science. In many people’s mind, science is independent, objective and dispassionate. This book succinctly challenges that naive view of science. It demonstrates, sometimes too often, how science is a very human activity and inarguably exists within a very powerful cultural matrix that determines what we choose to measure, how we interpret data and the subtle forces of self-interest. As someone who develops WMD’s for industry, I found the book extremely useful in articulating our moral responsibilities in the field. My hope is this book and it’s cautionary tales becomes a compulsory text for those engaged in the field and that it provides a moral mirror in which we can reflect, often.
This book offers the readers a unique and balanced point of view on how big data can be used to create inequality for many people in our society to benefit a very few groups (ie. the 1%, the technocrats and government agencies)