Book Review - The Efficiency Paradox: What Big Data Can't Do
This article is part of Aigora's "Review" series, in which we review resources, such as books, to help our readers find valuable content!
Review Score: 5 stars out of 5
As I've detailed in earlier blog posts, my study of data science to support market research consulting has led me back to the roots I grew when I completed my Ph.D. in mathematics and postdoctoral work in computational neuroscience. Especially exciting to me has been the proliferation of machine learning and AI tools through seemingly every aspect of society. And it was excitement over these tools together with the need for support in the use of these tools that led me to branch out from the Institute for Perception and to found Aigora earlier this year.
It was with this excitement that I came across "The Efficiency Paradox: What Big Data Can't Do," by Edward Tenner (provided to me by Amazon's recommender system with unintentional irony). The title intrigued me as I typically welcome countervailing opinions, so I decided to invest the time in reading the book. Having obtained the book from my local library, I began reading at my first opportunity and was pleasantly surprised by both the quality of the writing and the depth of the thinking.
In short, the book provides a much-needed counterbalance to the current AI buzz that seems almost as ubiquitous as machine learning algorithms themselves. Chief among the insights provided is the near constant operation of Campbell's Law - when we use a metric as an objective measure of performance and attempt to improve performance by optimizing that metric, the risk that the optimization process will affect the metric increases with the scale on which we implement the optimization. For example, when Standards of Learning (SOLs) are used to assess teacher performance, teachers begin to "teach to the test," and the SOLs themselves cease to reflect what they were initially intended to measure. In many cases, as has been the case with SOLs and education quality, the impact of the optimization process is to damage the very performance the optimization was intended to improve.
Dr. Tenner discusses the impact of data-driven approaches to educational optimization in Chapter 3 of his book, but he also discusses the unintentional consequences of data-driven strategies on our social fabric, on navigation, and on medicine. In every case, Dr. Tenner provides compelling arguments and examples supporting a "not-so-fast" attitude towards the implementation of data-driven schemes. While Dr. Tenner is ultimately not a neo-Luddite - in many cases he provides personal examples of his use of data-driven tools - his message is instead that we should be wary with, and perhaps even skeptical of, an embrace of data-driven efficiency. As Dr. Tenner convincingly argues, we may be forfeiting value without realizing it, with the ironic effect that we lose long-term effectiveness even as we gain efficiency in the short-term.
In his conclusion, Dr. Tenner argues the best strategy is a hybrid one in which we blend data-driven tools with more classically human approaches, such as following one's instincts and opening oneself to serendipity by placing oneself in stimulating environments. This last point seemed so similar to Jeff Bezos' recent praise of wandering, in fact, I wondered whether Dr. Tenner had influenced Bezos's thinking. Finally, throughout the book, Dr. Tenner provides helpful suggestions on how to use data-driven tools in a manner that accommodates domain expertise and allows for discovery instead of merely allowing oneself to be the passive receiver of information delivered by an algorithm.
In all, this book upgraded my thinking about data-processing, about science, and about our enjoyment of life in general. I also learned that Dr. Tenner and I don't actually disagree on the value of data-driven approaches - I also believe that human and machine intelligences together are better than either intelligence individually. Even though we ended in agreement, this book improved me, and that's ultimately the only reason I ever read anything. Thus I am grateful to Dr. Tenner for a job well done.
That's it for now. If you'd like to receive email updates from Aigora, including weekly video recaps of our blog activity, click on the button below to join our email list. Thanks for stopping by!
