Influence - Robert Cialdini

3 minute read

Following my new year new me resolution, I’ve finished reading the first book of the year (started in December 2017 anyway).

Very brief summary

A book written by a psychologist originally in 1984, then revised a few times. It’s about 6 major factors that affect people’s decisions, namely reciprocity, consistency, social proof, liking, authority and scarcity. Analyses of scientific evidence made up the majority of the book.

General feelings

Most ideas are not new to me, probably heard them from introduction to psychology podcasts from MIT MOOC. But there are quite a few interesting examples that I’ve never heard of. 7/10 not a waste of time reading it (higher the better), 3/10 difficulty of reading (lower the better).

For those who won’t read it

  • Reciprocity is when people give you something, you feel obliged to return favor. Sometimes it is given in a form of a concession, and you also feel the urge to make a concession. To fight against it, frame the gifts from merchants as promotional campaigns, not social interactions.

  • Consistency is the tendency to act in accordance to one’s previous acts. Sometimes compliance agents ask you for a small favor, then ask for a bigger one to trigger more responses for the latter. Avoid exploitation by recognizing the tricks and stop being consistent.

  • Social proof is when people look for others’ reactions to decide how they should act. Could lead to disastrous results. Fight back by realizing when others are also uncertain of how to react, and think instead of copy.

  • Liking is when we associate better characteristics to people with some good features, often in an unrelated way. Avoid exploitation by separating requests with requestors.

  • Authority is obviously listening to experts, or whoever appear to be experts. Be cautious when people pretend to be experts, and realize that even experts make mistakes too.

  • Scarcity is also intuitive, people value scarce goods more, especially when there are competitors for it. Avoid exploitation by separating the utility of the thing and the economic value of it due to scarcity.

For those who read it

There are a few points that came to my mind reading this book. First, there is a built in dilemma to all these psychological exploitations, as pointed out in the book. We rely on these psychological cues because we need them to simplify thinking, but they expose us to manipulations. The author suggests to wage a war against manipulators, but I think that’s asking for too much. However, I did find one tactic that he pointed out useful. It is to imagine, at the moment of deciding, if what had happened between the requestor and myself didn’t really happen, would I have made a different decision? This reminded me of 手割, a technique employed by Go players. This is because in both the process of buying a new car and playing a game of Go, it really doesn’t matter how you got to that situation, only the current state matters. In Go, it is the current game board, and in a dealership it is the deal you’re getting. It does not matter whether the dealer is likeable, whether there are ten other buyers in line, or whether he made a discount for you - only the current price tag and the car quality. Therefore in both cases, it is logical to disregard the history of events and consider only the present. (In Go, the technique is similar to asking the question of, if we started at a different route, would I have made the same choice? Anyway I don’t claim to be a good player or a good explainer of Go strategies.)

Second, it is mentioned quite a few times in the book that the psychological shortcuts we have are quite useful, but sometimes simplistic that they misfire. This reminded me on some recent research done in the field of machine learning, to show the possibility of exploiting a black box image classifier. In the research findings, a trained model that tags an image is given a picture X that has tag A, then it is modified pixel by pixel until it looks like picture Y to humans which is supposed to have tag B instead, yet the model still tags the generated image as A. When I saw this, I thought it was a shortcoming of machine learning in general since adversarial inputs are hard to avoid without a theoretically proven generalization bound on deep learning. But with all the evidence laid out in the book, it seems like humans are not much better after all.

The next book on the list: Capital in the 21st Century. To write down my next book is to exploit the consistency psychological urge, so as to maximize the likelihood of actually finishing the book.