Hannah Fry describes her book Hello World quite nicely in the introduction.
“It’s about asking if an algorithm is having a net benefit on society. About when you should trust a machine over your own judgement, and when you should resist the temptation to leave machines in control. It’s about breaking open the algorithms and finding their limits; and about looking hard at ourselves and finding our own. About separating the harm from the good and deciding what kind of world we live in.” (pages 3 - 4)
My own blurb: it’s a general overview of the dawn of an algorithmic era, an exploration of our first steps (and missteps) into a digitized tomorrow. The book is divvied up into chapters by topic — medicine, cars, and justice to name a few. Each one roughly follows the same pattern: a success story of algorithmic thinking and a cautionary tale of where it could (and often does) go wrong.
The set of case studies are very good. They serve as springboards into the tough discussions at the intersection of data, technology, and society. We hear of risky false positives in automated diagnoses, criminal statistics serving eager police, and the challenge of making backup plans for self-driving cars that go awry. Fry’s way of describing complicated ideas like random forests, image detection, and geospatial analysis is inspiring and a nice model of how science journalism should be. Her exposition of true positives versus false positives was spot on — Darth Vaders as people who first seem low risk but are actually high risk, and Luke Skywalkers as vice versa (pages 67 - 70). Makes me wonder why metrics like this are not front and center in the pop-technology articles raving about machine learning. Too nerdy, I guess.
As a practitioner of data science, I particularly liked the bit about IBM Watson, and why its creators had a hard time generalizing trivia show prowess into grand-sweeping medical wizardry. Which brings me to Fry’s clever test for algorithms in the news.
“Swap out any of the buzzwords, like ‘machine learning’, ‘artificial intelligence’, and ‘neural network’ and swap in the word ‘magic’. Does everything still make grammatical sense? Is any of the meaning lost? If not, I’d be worried that it’s all nonsense.”
The media hype of algorithms and technology has been all too real. My sense, however, is that people and businesses are slowly coming to terms with the differences between their prior expectations and reality. We’re all starting to witness the unintended consequences of the technologies that were formerly trumpeted to us as progress. Well written books like Hello World — optimistic, yet grounded in realism — are a much needed step forward in the discussion.
That said, I have a few minor concerns. Fry admits several times that open investigation of algorithms is necessary to improve them from their biases and inaccuracies. But there’s little talk on how to do this, or the complexities therein (open code vs open data, right to be forgotten, the EFF in general). And while the “Art” chapter was well done, I found its central message — that defining quality, in a math-y way — is not merely a problem of aesthetics. It is perhaps the core theme of the book, and one that perhaps deserves longer (and earlier) attention. But hey, I like nitpicking sometimes, only because I care a lot about this stuff.
The book is overall a great success, though. It’s a quick read, with an easygoing and sometimes humorous style. The reference section is impressive, too, giving me lots of follow-up reading to do. Fry communicates clearly on a central set of topics for the digital age, and reading the book made me think about my own career as a data scientist through a bigger, more societal lens. The world will be a much better place if we continue to talk about technology with Hello World’s lessons in mind.