Epistemic Books to Spot Informational Influence

Informational influence is the ordinary kind: you copy people because you think they know something you do not. Almost everything you believe arrived this way, including most of what you would describe as your own opinions.

So the reading list for this topic is not about persuasion technique at all. The useful question is narrower and harder: how do you tell who actually knows? Books that help with that come from five different places, and none of them is the influence section.

What to take away

Calibration: can you score yourself?

The most valuable thing a book here can give you is a way to find out how often you are right, because intuition about your own accuracy is close to useless.

Superforecasting by Philip Tetlock and Dan Gardner is the standard entry point: it is about people making explicit, dated, checkable predictions and then being scored. Thinking in Bets by Annie Duke covers similar ground from a poker player's angle, and is better on the emotional part: separating a bad decision from a bad outcome.

Verdict 5/5. Philip Tetlock and Dan Gardner, 2015. Read this if you want a scoreboard for your own forecasts.

What to take from either: the habit of attaching a number and a date to a belief. "I think this will happen" is unscoreable. "I think this is about seventy per cent likely by March" can be wrong in a way you will notice.

What to leave: the implication that becoming well-calibrated is mostly a matter of wanting to be. It is a practice with a scorecard, and without the scorecard you are reading a book about exercise.

Reasoning: why argument works the way it does

The Enigma of Reason by Hugo Mercier and Dan Sperber is the most interesting book on this shelf, because it takes seriously a fact everyone notices and nobody explains: people are terrible at finding flaws in their own reasoning and rather good at finding them in someone else's.

Verdict 4/5. Hugo Mercier and Dan Sperber, 2017. Read this if you argue for a living and want to know why arguing moves people.

That asymmetry has a practical consequence. If your reasoning improves mainly under challenge, then reading alone will not do it. You need someone who disagrees and is not afraid of you, which is an argument for a colleague rather than a book, and worth knowing before you buy the book.

Evidence literacy: reading the numbers you are handed

Calling Bullshit by Carl Bergstrom and Jevin West is the practical entry here. It is about spotting bad quantitative claims without needing statistics training: selection effects, misleading axes, numbers that could not have been measured the way they are described.

Verdict 4/5. Carl Bergstrom and Jevin West, 2021. Read this if numbers reach you with no method attached.

This matters more than the influence literature for one reason. Most bad information reaches you already wearing evidence, and the failure mode is not "I was persuaded by a charming person" but "I read a chart".

Pair it with the discipline described in authority research: before believing a striking finding, ask whether anyone who wanted a different answer has repeated it. The same test applies to informational influence research, where a striking finding is only as good as its failed replications.

Expertise and its limits

The Death of Expertise by Tom Nichols is worth reading precisely because it is one-sided, and reading it that way is the exercise. It argues that non-experts systematically overrate their own judgment. It is much weaker on the reverse failure: expert overreach past the boundary of the credential, which is the failure you are more likely to meet in a professional setting.

Verdict 3/5. Tom Nichols, 2017. Read this if you work with credentialled people and need to see where the credential stops.

Read it against your own experience and note where you disagree. That is a more useful use of the book than agreeing with it.

Disclosure: who paid for this

A large share of what looks like informational influence is purchased. A review, a recommendation, a comparison table, an enthusiastic mention: any of these can be advertising with the label removed. Paid placements are one of the informational influence signs that a source was chosen for you rather than by you.

You do not need a book. The Federal Trade Commission's Guides Concerning the Use of Endorsements and Testimonials in Advertising (16 CFR Part 255) explains how disclosure should look. It also shows what missing disclosure looks like.

Read it once. For years after, you will spot the phrasing that stops short of saying who paid, the review that never mentions a downside, and the comparison where one option was never really in contention.

An exercise worth more than any of them

Pick something you believe about a subject you have not personally studied. Anything: a health claim, an economic one, something about a technology.

Know which of your beliefs are load-bearing enough to be worth tracing. That is the whole aim of the exercise.

Now answer three questions in writing. Who told you? When? And what would that person have had to observe, personally, to know it?

The third question does most of the work. Often the honest answer is they could not have observed it either, and you are three or four handoffs from anyone who checked anything.

That is not a reason to disbelieve it. The chain of transmission is how knowledge works.

What each shelf actually delivers

The right-hand column is the part booksellers leave out. Each shelf does one job well, and reading three books from the same slot mostly produces restatement.

Related habits of checking who is answerable for a claim are under authority. The version that runs on published counts rather than individual sources is under social proof.

What a good book here has in common

It makes you worse at trusting yourself, not just better at doubting others. The failure mode of this whole genre is the reader who finishes it convinced they are the rare clear-thinking person in a world of the credulous. That reader has acquired a new way to dismiss disagreement and nothing else.

The test is simple: name one belief a chapter changed.

More on the mechanism itself in informational influence, and on how it interacts with going along to belong in social influence basics. Where a claim is being carried by how many people repeat it, the questions to ask are contested rather than obvious, and the Stanford Encyclopedia entry on social epistemology sets out why.

Common questions

How many books from this list do I actually need?

One from the calibration shelf and one from the evidence shelf covers most of the ground. Adding a third from either slot tends to repeat the second.

Is a popular science book a reasonable source here?

As a map, yes. As a citation, no. Read them to learn what questions exist, then go to the source when a specific claim matters to a decision.

What if a book I liked turns out to rest on revised evidence?

Keep the framing and drop the numbers. The descriptions of what people do have generally held up better than the explanations of why.

How do I stop reading this genre as a spectator?

At the end of each chapter, name the last time you did the thing being described. If you cannot, you were reading about other people.