James Bridle's writing and art about the complexity of network technologies is often so careful about saying everything succinctly and clinically that it's tempting to believe that he might be part machine. So if anything, this book has proven his humanity, if a little disappointingly.

In content, writing in 2017, Bridle is ahead of his time. His topics range from bias in image machine learning models to secrecy in corporate and government surveillance. However, the structure of the chapters often reads like a Wikipedia dive, leaping between stories and vaguely connected ideas with gleeful abandon. The result is a little chaotic and difficult to connect together. By no means a bad book, but there are better examples that deal with these topics more coherently.