Informational influence mistakes seen from the repair side

Trusting other people is not the mistake. Almost everything you know arrived that way. The mistakes are all about losing track of how much of it came from one place.

What informational influence is

Informational influence mistakes have a single shape: you go along with what others say or do because you are treating their response as evidence about how things actually are. You accept the information, and your belief changes. Normative influence is the other half of the pair — going along in order to be accepted, or to avoid standing out, without believing anything new.

The distinction is standard in social psychology and is usually credited to Deutsch and Gerard, who separated the two in their work on conformity. Informational influence produces private acceptance: you really do come to believe the thing. Normative influence produces public compliance: you say or do the thing while privately unconvinced. Only the first changes what you think you know, which is why every mistake below is a failure of evidence rather than a failure of courage.

The classic demonstrations are Sherif's autokinetic experiments, where people in a darkened room converged on each other's estimates of a stationary point of light, and Asch's line-judgment experiments, where people gave plainly wrong answers to match a unanimous group. Cialdini's Influence covers both under the principle of social proof, the habit of reading other people's behavior as information. Borrowing from other minds is not a flaw to be removed; the mistakes are in what gets borrowed and in whether the borrowing is recorded.

What to take away

The mistakes

Counting voices instead of sources. Three people say it, so it must be right. Trace each one back a step and they land on the same origin. Nothing was invented and the agreement carries no additional evidence. Run the same mechanism in public and at speed and you get an availability cascade: a claim's repeated circulation makes it feel more familiar, more important, and eventually more true. This is the one to fix first, because it is common, cheap to check, and it silently inflates confidence in almost every field.

Treating familiarity as evidence — the illusory truth effect. A claim you have encountered many times feels more likely than one you have met once. The effect has a name, and it is the reason a statement repeated often enough starts to feel like background knowledge. Repetition does not add support and it does change how the claim feels, which is why the number of times you have seen something is worth ignoring as deliberately as you can manage.

Reading a summary and citing the source. You read the article about the report. Then you say what the report found. Summaries drop conditions, and the conditions are usually the part that mattered. If you did not open it, say you did not.

Taking silence as verification. Nobody objected, so it is presumably right. In most rooms nobody objected because objecting was expensive, and absence of correction is compatible with almost anything.

Mistaking confidence for information. The person who states things without hedging is easier to follow and no more likely to be right. This is authority bias in its everyday form: the claim borrows weight from the standing of whoever said it rather than from the grounds behind it. Calibrated uncertainty is a skill and it sounds like weakness, which is a bad arrangement for everyone.

Forgetting how you learned it — source-monitoring error and source amnesia. Seen, told, read, or worked out are four different kinds of knowing, and after a few weeks they all feel the same. Source-monitoring error is attributing a memory to the wrong origin; source amnesia is keeping the content while losing the origin entirely. Most overconfidence is not about the claim; it is about the provenance having quietly gone missing.

Updating on the person rather than on what they said. You changed your view because somebody impressive said it. Sometimes that is correct, and the Stanford Encyclopedia of Philosophy entry "The Epistemology of Testimony" sets out the conditions under which it is (problems of testimony). What makes it a mistake is not noticing that the source was doing all the work.

Assuming the crowd had your question. Many people chose this and none of them had your constraint. The choice was real and its relevance was assumed, which is the same error in a different costume and is treated under social proof.

Passing on a claim at a higher confidence than you received it. Each retelling smooths the hedges. "I think, though I have not checked" becomes "apparently" becomes a flat statement in three steps, and nobody in the chain did anything wrong. What a speaker implicitly represents about their own grounds when they simply state something is the subject of the Stanford Encyclopedia of Philosophy entry titled "Assertion" (Stanford Encyclopedia entry on assertion), and the norms it describes are the ones quietly broken here.

What went wrong in each case

Read down the right-hand column and one pattern runs through it. In each row, something that transmits easily has been mistaken for something that is true, and transmission is much cheaper than verification. That gap is the permanent condition rather than a temporary problem, which is why the checks have to be habits rather than decisions.

The checks

Repairing a claim you have already passed on at the wrong confidence takes longer than the four checks, because the people you told now hold the inflated version. Go back to them, name the claim, say what you actually had when you passed it on, restate it at the confidence your grounds support, and hand over the source you can point to. Inoculation theory — building a weakened version of a misleading claim into what you pass on, so that people who meet the strong version later recognize it, an approach now usually called prebunking — is the preventive version of the same move.

None of this requires distrusting people, and a person who checked nothing and trusted nobody would know almost nothing at all. The point is to keep the receipts attached, which is the habit described under informational influence.

When it is deliberate

Most of these mistakes happen with nobody at fault. A few are engineered: a manufactured chorus, which is astroturfing when the apparent grassroots agreement is staged or bought; a claim laundered through a friendly summary so the conditions that mattered disappear; and repetition arranged so that a single origin looks like independent agreement.

The tell is the same in each case: tracing back one step is discouraged or made difficult, and an honest source is happy to say where it came from. Someone who changes the subject has answered you.

The field guide for deliberate versions is under manipulation resistance, and the standard separating them from honest influence is under social influence basics. The case where a claim's weight comes from a credential rather than a source is under authority.

Common questions

Is it wrong to rely on a summary?

No. Almost all reading is summaries and the alternative is knowing very little. The mistake is citing the original when you read the summary, which costs you nothing to fix.

How do I ask where something came from without sounding hostile?

Ask it as curiosity about the source rather than doubt about the person. "Where did you see that? I want to read it" is a request rather than a challenge, and it usually gets a link.

What if I cannot trace something back?

Then hold it at the confidence a rumor deserves and say so when you repeat it. Untraceable claims are not automatically false; they just should not be passed on with weight they cannot carry.

Does any of this apply to my own memory?

Yes, and more than people expect. The provenance of your own beliefs decays faster than the beliefs do, which is the reason to write down where something came from at the time.