Review: Talking to Strangers

The Good

Gladwell says that the audiobook is like a "well polished podcast." He's right, he narrates it well and the music choice is good. Where possible, he includes actual audio from interviewees and other quotes. This does make the audio book more engaging and entertaining.

The core of Gladwell's thesis is that strangers are difficult to understand. Whether they're from different cultures, trying to deceive us, or just drunk. We should communicate tentatively and with humility. We make things a lot worse with miscommunications and misunderstandings.

The Bad

Gladwell makes several crucial errors. He graphically describes rape and pedophilia without any warning, he presents a misleading view of rape and alcohol, and his examples miss important context and nuance.

He graphically describes one instance of rape and two instances of pedophilia with no warning. This is enough to warn strongly against the book. Anyone who has suffered with through these may find the book brings up traumatic memories. Save yourself the hassle and avoid it for this reason alone.

Gladwell misses out on important context and nuance. Early in the book he compares judge's performance to an algorithm. He claims the algorithm is better than the judges, but ignores the associated baggage of predictive policing; the racially-biased data given to these algorithms, the increased policing of BIPOC, and so on. These are important because they help the reader understand that the algorithms are not magic. They do make systematic mistakes, just like people. Their accuracy and precision should be scrutinised and not uncritically presented as "better."

Similarly, Gladwell gives two ideas of how we interpret others' statements:

1. "Default to true" -- accept what the other person is saying uncritically until there is overwhelming evidence of deceit.
2. The "holy fool" -- assume everyone is lying all the time until there is overwhelming evidence there is no deceit.

We don't need to keep this dichotomy. We can be better Bayesians and update our beliefs when we get new information. If there are roughly 200 active serial killers in the US at any one time and you find out that someone you know had blood in their car you can update your beliefs about them. You don't have to assume everyone's totally fine or totally out to get you. Nuance exists!

Finally, Gladwell opines on alcohol and rape. It didn't sit right with me, but I am not qualified to properly critique it. I defer to people more qualified than me on this one. I found it creepy if nothing else.

Overall

I do not recommend this book.

While Gladwell's thesis and presentation are good, his use of existing studies and ideas about how to apply them are not great. His graphic descriptions of rape and pedophilia are unnecessary and do not have content warnings.

Gladwell seriously weakens his argument by missing nuance. For example, uncritically comparing judge's decisions to algorithms, or presenting the false dichotomy of "default to true" and "holy fool."