Year of (reviewing) Books: Spring update

I’ve always read a lot, especially non-fiction. However, I don’t always reflect on a book before moving onto the next one. So, this year, I promised myself that I would rank and write a (coherent) review of every book I read, rather than just a select few. I’m posting the reviews mostly to force myself to keep that promise, though, as the following list suggests, I don’t always succeed. I’m always open to book suggestions, so please send them and I’ll (probably) get to them (at some point).

Without further, ado, my ranking so far:

  1. The New Jim Crow, by Michelle Alexander. Probably the most powerful book I’ve ever read, “The New Jim Crow: Mass Incarceration in the Age of Colorblindness” forcefully makes the case that the American criminal justice system, and specifically the Drug War, has resurrected much of the worst of Jim Crow era laws for black men — criminalization and associated stigma, taking away of voting rights, and few educational and economic opportunities — while prima facie being wholly colorblind and race neutral. 
     
    Michelle Alexander speaks directly to her audience (presumably people open to her ideas but potentially skeptical of the problem’s full extent), and she doesn’t shy away from her claims. The book first lays down the undeniable statistics of the number of black men in jail (or facing the subsequent punishments of a lifetime of stigma that effectively prevents labelled felons from obtaining jobs or housing) for drug crimes. It then systematically hammers away at the most common responses given by defenders of the system.
     
    On the system being ‘race neutral:’ Alexander points out that numerous studies have found that drug use rates are fairly consistent across race, but that the overwhelming percentage of those punished for drug crimes are black men. The discrepancy in punishment is found in discretion throughout the system — police officers who overwhelmingly patrol urban neighborhoods and prosecutors who pursue charges against black defendants at rates far higher than they do against white defendants (just think of the last time a white college kid was labelled a felon for smoking marijuana). Consider the recent “Opioid Crisis,” which overwhelmingly affects white patients and doesn’t discriminate by class. No one is calling for these drug abusers to be thrown into jail, and we recognize that rehab is a much better response. The same understanding hasn’t historically been given to black drug users, and a heavily biased system is the result.
     
    The book (written in 2010, before Black Lives Matter and recent public awareness of some aspects of the problem) also has some valuable thoughts on how bring about change. Alexander underscores the need to argue with the overwhelming statistics, rather than relying on a few poignant examples of police excess. The problem with the latter, she argues, is that it’s too easy to dismiss any given case by casting the victim as a criminal or otherwise blaming them for their punishments. It’s much harder to explain away the fact “75% of people in state prison for drug conviction are people of color although blacks and whites see and use drugs at roughly the same rate,” or that “the proportion of African American men with some sort of criminal record approaches 80% in some major US cities.” The author points out that Civil Rights Leaders intimately understood this phenomenon: Rosa Parks (a churchgoer and moral stalwart in her community) made a difference precisely because she was impossible to dismiss using the standard excuses; unfortunately, such a case cannot exist in today’s system because it operates by criminalizing black men and then punishing criminals, and so by definition has no ‘innocent’ victims. 
     
    I can’t possibly go into the depth and breadth of arguments Michelle Alexander makes in the book — if you read just one book this year, read this one.
  2. Weapons of Math Destruction, by Cathy O’Neil. Weapons of Math Destruction is a shot against ubiquitous, unthinking use of ‘Big Data,’ against the use of unaccountable, opaque algorithms that make decisions ranging from teacher firing to bail decisions. It provides an overview to how seemingly effective algorithms, branded as the solution to biased human decision making, can rather reinforce societal inequities if not held accountable. 
     
    The book is filled with examples, each highlighting a different aspect of the problem. One of the running themes is Goodhart’s law: “When a measure becomes a target, it ceases to be a good measure.” Algorithms operate exactly by turning measures into optimization targets, often leading to excessive optimization of a metric that’s only a proxy for the metrics about which we really care.
     
    In some cases, the metric becomes gamed: the US World News College Rankings algorithm partially ranks departments by the research output of their faculty; one year, a new mathematics department at a Saudi Arabian university rose to become one of the top ranked math departments in the world, above MIT and Cambridge. Why? They paid a group of highly cited researchers a bunch of money to become adjunct faculty, fly in for 3 weeks a year, and change their affiliation on the citations tracker that US World News uses. Nobody would claim that the university is actually better than MIT at mathematics, but neither would anyone claim that in normal circumstances faculty research output does not matter. Unfortunately, universities often game the rankings in far subtler ways, and similar ideas underscore criticisms of teaching measures (“teaching to the test”), and faculty tenure measures. 
     
    In other cases, no gaming is necessary; the problem lies in metric misspecification, i.e. things other than the optimization metric matter in reality. For example, many fast-food and retail establishments use a program to schedule their employees’ shifts; this program minimizes labor cost by scheduling (hourly) employees based on predicted customer traffic, using weather and historical data. This program is apparently extremely effective at its primary objective of minimizing labor cost while having enough employees to meet demand. However, it wreaks havoc on employees’ schedules and lives, as employees have to contend with unpredictable hours when scheduling day-care, appointments, or other jobs. Such a problem may be fixed by constraining the program, but adding such constraints must become a business priority. 
     
    Unfortunately, even if a metric is generally well thought out, the scale of its application may harm diversity. For example, even if the metric of the US World News College Rankings were perfect, the fact that the rankings affect every school (through the number of student applications and federal funding) leads to a standardization that might not be good. Rather than each college being able to cater to different types of students, all get ranked on the same metric. 
     
    Overall, O’Neil provides a warning call against the ever-increasing use of algorithms in society, even when the algorithm seems to be succeeding. The above examples shed only part of the problem; e.g. questions about whether ML techniques pick up bias present in the training data are especially tricky to solve. She is not a Luddite, and does not encourage the readers to be; such algorithms undoubtedly can do a lot of good in the world, but maybe we should think about these potential problems before blindly listening to algorithmic outputs.
  3. Shoe Dog, by Phil Knight. Shoe Dog is the memoir of Phil Knight, the founder of Nike; the next book on this list, Elon Musk, is a biography of the leader of PayPal, Tesla, SpaceX, and SolarCity. They are both excellently written stories that present compelling examples of companies built not for high-growth in anticipation of a mind-numbing acquisition figure, but rather to fulfill life-long ambitions of their creators. 
     
    In Shoe Dog, Phil Knight gives a year-by-year account of the first few decades of the company that eventually became Nike. We now all know his company as the epitome of a American “brand,” but the memoir focuses on the years before brand marketing was a thing and before Nike turned sneakers into something most people wear every day. He describes, in excruciating detail, the many years of revenue in the low thousands, when he was importing Japanese shoes and before he could even afford giving himself a salary. The story is a classic one of struggle and perseverance in service of a dream, and Knight does not spare the reader from all the times he almost failed — or the luck and borderline illegal acts that saved him. As a memoir, it of course defends some of the author’s controversial actions, but it doesn’t seem to hide them. Overall a great read on the creation of the modern American brand.
  4. Elon Musk, by Ashlee Vance. The book is a similarly balanced view of the enigmatic figure behind some of the world’s most ambitious companies. What strikes the reader, immediately, is how Musk differs from the other most wildly successful entrepreneurs. For example, Bill Gates, Mark Zuckerberg, and Knight all succeeded because they had special skill-sets or interests, honed in adolescence — programming for the first two and running for last — and then, when they identified a matching opportunity, they leveraged those skills, their insane work ethic, and business aptitude to build lasting companies. Only once they succeeded did they seriously adopt well-meaning missions like wanting to “connect the world.” Musk, post-Paypal, proceeds in the opposite order. He starts with what he views as humanity’s greatest threats — being a single planet species (SpaceX), pollution (Tesla and SolarCity), and aggressive AI (OpenAI) — and then leverages his memory, work ethic, and technical aptitude to develop the skills and knowledge he needs to build lasting companies to address those challenges. The book counter-balances this idyllic view of his businesses, however, with numerous examples of Musk’s infamous impatience and temper. The same mission that drives him to tackle the biggest challenges prevents him from tolerating any weakness that hinders his progress. Overall a well-recommended book on a uniquely consequential personality.
  5. Rise of the Rocket Girls, by Nathalia Holt. A good read, one that presents the history of JPL and NASA as a whole through the lens of the women who, from the outset, participated in every facet of design, testing, and launching of rockets. The book begins its story before the equally wonderful Hidden Figures does, in the 1930's when JPL was simply a dream of a few rogue students at Caltech dubbed the “Suicide Squad.” It proceeds by telling the stories of the women of many cultural and racial backgrounds. I especially enjoyed the book because it did not limit itself to telling just the personal stories, as Hidden Figures did — it weaved the personal within a broader context of NASA history through the 2000s, from before its inception when JPL was reluctantly building military rockets to the Space Race to almost the modern day — and it was not afraid to get technical. This narrative style is effective in underscoring that the personal stories are not just tangentially related to NASA’s awe-inspiring feats, but rather that the individual, highly technical, contributions by the women the book profiles were essential to those feats.
  6. Being Mortal, by Atul Gawande
  7. Hillbilly Elegy, by JD Vance
  8. Underground Airlines, by Ben H. Winters. The only fiction book I’ve read so far this year, Underground Airlines imagines an alternate history in which slavery has continued in the United States to modern times — Abraham Lincoln was assassinated before the Civil War, ‘compromise’ amendments were passed forever protecting the institution, and four deep-south states stubbornly maintain their ways. The author is a good world-builder, and the book shines when exploring the moral compromises made by all the characters in such a world. For example, the Northern liberal who convinces themselves that they’ve done enough with “Clean Hand Laws” reminds us of ourselves, satisfied by the labor promises made by the companies from whom we buy our phones and clothes — their rationalizations seem lacking only because we can live in a world where “that’s just the way it is” doesn’t apply to the things they are applying it to. That isn’t to say that one should ignore economic and political constraints (and I am as strong a critic of Sanders and his supporters as most), but the book does underscore that the conversation can’t end at listing the constraints.
  9. Wright Brothers, by David McCullough. This book tells Wilbur and Orville Wright’s story as the brothers who were the first to build and fly a heavier-than-air plane that could take off in its own right, level off, then land on the same or a higher level as takeoff. It stands apart from others about innovators by spending the entire book on their lives and work, and virtually none of it waxing poetic on the broader implications. Also unusual for such a book, it explores the life and contributions of Catherine Wright, Wilbur and Orville’s sister, primary caretaker, and supporter. 
     
    The book shines when describing the brothers’ tenaciousness. Though they didn’t have any formal education, they obsessed with reading books about flying, spent months studying and observing birds, and built a wind tunnel because of doubt about published aerodynamically tables. They carefully chose Kitty Hawk to test their planes, contacting weather bureaus and towns across the country to find the perfect combination of weather, wind, and sand (for safe crash landings). They were fearless in building and flying test gliders and planes. Due to these qualities, they succeeded where others with government funding and formal training had failed. 
     
    Even when they achieved unimaginable fame — with kings and presidents among the thousands who’d come to observe each flight — they were dedicated to improving their craft. Early on during their testing, they rejected an introduction to Andrew Carnegie to fund their work, preferring to continue running their bicycle shop. One journalist described a day in which a parade was held in their honor in their hometown, Dayton, Ohio. In between meeting the luminaries and handling their various ceremonial duties, they would sneak back to their shop to prepare for an upcoming test flight. To the end, it was “their patient perseverance, their calm faith in ultimate success, their mutual consideration in each other” that took them to new heights.
  10. Algorithms to Live By, by Brian Christian and Tom Griffiths. The authors promise to recast common algorithms from a variety of fields in computer science to the problems that we face every-day. If you’re familiar with the assumptions often baked into the optimality proofs of such algorithms, you know that in one sense this promise is impossible: nothing that would be considered an ‘algorithm’ could possibly be taught in a book and then uniformly applied. 
     
    Instead, this delightful book describes lessons learned from algorithmic concepts (including standard sorting/searching, caching in architecture, and multi-armed bandits and the secretary problem in probability) and suggests unexpected ways one could apply them in our lives and broader society. For example, a messy desk with unsorted stacks of papers may seem suboptimal for finding things. However, computer scientists know that if one isn’t querying a list enough to amortize the sorting cost, then one shouldn’t sort. When lessons from caching are added, this notion is even stronger (even a messy cache is better than an organized but expensive-to-access storage). At an organizational level, he also suggests that success of the Least Recently Used (LRU) cache algorithm suggests that libraries shouldn’t be in a rush to put returned books back in their ‘proper’ place, but should instead just have a ‘recently returned’ bookshelf out front that patrons could browse. In addition to saving countless middle school ‘volunteers’ from the bane of stacking books likely to be checked out again soon, the book-shelf would create an ad-hoc societal reading list. 
     
    One danger in such a book is that, if written poorly, it’ll either be boring to those who are aware of the technical concepts or intimidating jargon-filled-nonsense for the uninitiated (or, worse — both, while simultaneously managing to butcher the technical ideas). The first half mostly avoids both. The authors take the time to explain various algorithms, and computer scientists may want to skim/listen at triple speed at such parts. However, they do so without being pedantic and offer enough nuggets of insight on how the concepts can be cleverly adapted to unfamiliar contexts, such as the examples above. The second half, on the other hand, seems to explicitly target non-CS majors. The authors go through a variety of technical ideas — regularization in learning, NP-hardness and approximation algorithms, game theory, and optimization — describing the theory in lay terms but without adding much new insight for those familiar with the concepts. Instead, they map the concepts to the same canonical real-world problems often used as teaching tools in technical classes. The result is a book that is better than almost any other I’ve read in transferring lessons often derived with tedious math to real-world situations, without losing much of the theoretical insights, but one that will often feel overly familiar to a technical audience.
  11. Undoing Project, by Michael Lewis. Book about the friendship of Daniel Kahneman and Amos Tversky, who together exposed the everyday irrationalities of human decision making and upended economic thinking in the process. Not a bad book, and entertaining at times, but not the one to read if you want to delve into their theories. 
     
    If you haven’t read Kahneman’s Thinking Fast and Slow, I’d recommend that book over this one (and most any other book), as it walks the reader through the cognitive biases that they uncovered and that are built into human decision making. His articulation of “System 1” — the instant, intuitive, subconscious, often irrational ‘gut-reaction,’ versus the “System 2” — the slow, deliberative, rational consciousness, is still one of my favorite mental models.
     
    This book is more historical than scientific — though it discusses their results enough to convince the reader of their importance, the book is more a biography of the friendship of Amos Tversky and Daniel Kahneman. It sheds light on issues of ownership, credit, and reputation in science. They did their most important work together in a room for years, with no notion of one of them getting more credit than the other; however, Tversky ended up getting a faculty position at Stanford and myriad rewards during their lifetime, and Kahneman much less so, because Tversky was more outwardly brilliant and aggressive, whereas Kahneman was introspective and shy. The award discrepancy created much tension and eventually destroyed their relationship.
  12. Moonwalking with Einstein, by Joshua Foer. This book follows the author’s journey from journalist covering the memory championships to becoming the American memory champion, in a year. He describes the tricks, all variants of the ‘memory palace,’ that champions use to memorize tens of thousands of digits of pi or a deck of card in under two minutes. But the book is more about the culture around memory, including the famous Ancient Greek orators, the oral passing down of scripture and epics, and various synesthetes who seem to effortlessly perform miraculous memory feats. 
     
    It’s an entertaining read, and the author senses that most readers will view his methods from a heavily skeptical vantage point; he himself brings up the numerous self-help books that claim to teach quick tricks to unlock one’s inner genius, changing one’s life forever. He instead chooses to describe the hard work, focused training, and feedback from researchers he put in over a year to accomplish his feat. One leaves the book with the notion that the tricks he describes may be worth it (for me, improving my ability to remember details from papers I read would be invaluable), but that the effort involved (and the resulting opportunity cost) does not make it an obvious decision.
  13. Thinking, ed. John Brockman. My thoughts on this book are similar to those expressed in my previous reviews of Edge.com/John Brockman books — it’s a clever and entertaining concept, but the lack of depth is frustrating. This book covers many perspectives on ‘thinking’, but it often feels as if I’m getting a single side of a theory in each of 20 different fields, rather than a true academic debate.
  14. The Tycoons: How Andrew Carnegie, John D. Rockefeller, Jay Gould, and J. P. Morgan Invented the American Supereconomy, by Charles R. Morris. I’d been excited to read this book for quite a while — though I have read many books on the history of computing and information, I had not read their counterparts about more traditional American industries. This book covers far more than just the life stories of the titular names; it is a history of the American economic development after the civil war through the rest of the 19th century, also known as the Gilded Age. It often reads like a textbook, with far more numeric precision than would interest a casual reader. The book is effective in presenting a complex picture through multiple lenses, rather than championing one side of a contentious debate. For example, in discussing the roles of trusts and trustbusting, it discusses both the ways trusts were beneficial or at least not the bogey-men they’re made out to be, while identifying a few instances where they were anti-competitive and anti-consumer.
     
    I was hoping this would be a solid prequel to Breaking Rockefeller, which I read last year and which is focused on the rivals that challenged Standard Oil. That it places the lives of the tycoons in the context of their age becomes a hindrance here; the book simply tries to do too much, and I finished the book still unable to articulate a condensed summary of each tycoon’s most important actions and moments. 
     
    Overall a decent read, though one that I would not advise to tackle as an audiobook due to its textbook-y nature.
  15. Throwing Rocks at the Google Bus, by Douglas Rushkoff.