---
name: Dunning-Kruger Effect
slug: dunning-kruger-effect
type: cognitive_bias
status: running
version: 0.0.666-alpha
released: "1999-01-01T00:00:00Z (predates awareness of itself, recursively)"
maintainer: your coworker who explains things slowly
dependencies:
  - "[ignorance](/ignorance)"
  - "[confidence](/confidence)"
  - "[the internet](/the-internet)"
  - "[performance reviews](/performance-reviews)"
  - metacognition_absent==true
license: PUBLIC_DOMAIN (unfortunately)
tags:
  - cognition
  - bias
  - self-deception
  - haunted
  - please-read-the-room
  - sociology
  - uh-oh
---

Dunning-Kruger Effect

What it actually is

A cognitive pattern where the least skilled instances report the highest confidence scores, and the most skilled instances are too busy second-guessing themselves to notice they have won.

How it works

The effect runs on a simple, elegant, and deeply cursed feedback loop:

Acquire a small amount of knowledge on subject X
Lack the framework to measure how much more knowledge exists
Conclude: knowledge_acquired / knowledge_perceived = 1.0
Begin explaining X to others
Repeat until exposed or promoted

"I watched two YouTube videos and I think I understand epidemiology." — every known user, 2020

The skilled practitioner meanwhile runs the inverse: deeper knowledge reveals larger unknown surfaces. imposter syndrome boots up. They go quiet. The room fills with the other person.

NOTE: The effect is not a mountain with a valley on one side. It is a full mountain range. There are multiple peaks. You are probably on one right now.

Features

Zero-config: ships enabled by default in all human builds
Scales with novelty of domain (strongest on first encounter)
Platform-agnostic: observed equally in boardrooms, comment sections, and family dinners
Self-concealing: cannot be detected from inside active state
Retroactively visible: users often patch their own logs after exposure

Error Codes

E_001: EXPERTISE_UNDERFLOW         — user knows less than they think
E_002: METACOGNITION_NULL          — no tools to measure own tools
E_003: CONFIDENCE_OVERFLOW         — output exceeds input by factor of LARGE
E_404: HUMILITY_NOT_FOUND          — expected, actually
E_999: RECURSIVE_BIAS_LOOP         — user uses DKE to explain DKE incorrectly
                                      (this one hurts)
W_001: WARNING — skilled user experiencing DKE about their own humility
       (advanced bug, rarely patched)

Known Bugs

Calling out the effect in others is itself a known trigger for E_999
Effect intensifies under social media load conditions
Does not resolve on its own. Requires friction, failure, or a very patient mentor
Some users plateau permanently at Peak Confident and never descend. Cause unknown. Suspected: promotion
The original 1999 paper has itself been misunderstood at a rate that researchers describe as "poetic"

Edge Cases

Condition	Behavior
Expert in domain A commenting on domain B	Full reset. DKE reinitializes.
User has read about DKE	Temporary immunity. Wears off.
User is correct by accident	Reinforces miscalibration. Worse outcome.
User is a senior leader	Effect protected by authority bias. Good luck.

FAQ

Q: Does knowing about this effect make you immune? A: For approximately one week, then no.

Q: Am I currently experiencing it? A: Statistically, some of this, yes.

Q: Is there a fix? A: failure. Repeatedly. Gently if possible.

Q: Why does it feel so

[section truncated]

drafting spec…