civlab·failure-museum.psyverse.fun·memorial

⊘ Failure Museum

A serious archive of how things actually broke — companies, missions, treaties, theories — and what was learned.

Survivorship bias rules our histories. We learn the names of the companies that succeeded, the missions that landed, the treaties that held. We learn nothing about the dozens that didn't — and so we keep making the same mistakes. The Failure Museum is the missing wing. Each entry is a single failure, with the structural decisions that produced it, the warning signs that were ignored, the public narrative at the time, and the autopsy as the field eventually wrote it. Curated; sources required.

The goal is not schadenfreude. The goal is the thing autopsies do for medicine: convert losses into knowledge that future practitioners can use.

Interactive demo

Use it — don't read about it.

1961 · geo
Plan inherited from Eisenhower team; dissenting analyses suppressed.
1986 · eng
O-ring brittleness in cold launch; engineers warned and were overruled.
1998 · finance
Nobel-laureate-led; assumed Russia couldn't default. It defaulted.
2003 · eng
Engineering triumph; market for routine supersonic civil flight didn't materialize.
2010 · eng
BP cement-job decisions + missing safety review compounded into blowout.
2010 · tech
Successful national-scale terminal network couldn't migrate to open Internet protocols.
2011 · eng
Tsunami exceeded design; backup generators in basement; cooling lost.
2012 · gov
Top-down hard caps without binding enforcement; major emitters opted out.
2016 · biotech
Edison machine never worked; secrecy + celebrity board enabled long deception.
2018 · tech
Belief reinforcement learning was the path. Foundation models came from elsewhere.
2019 · eng
Single-sensor MCAS without crew training; FAA delegated review back to Boeing.
2019 · finance
S-1 disclosure exposed governance and metrics that private rounds had concealed.

Features

What this platform promises.

Browse by domain

Engineering, biotech, geopolitics, finance, governance.

Filter by failure mode

'Single point of failure', 'denied warnings', 'goal substitution'.

Random failure of the day

Forced confrontation with one historically expensive mistake.

Modules

Modules that compose this platform.

01 · case-files

Case files

100+ failures across business, science, engineering, governance.

02 · structural

Structural decisions

What was decided that made failure likely, often years before.

03 · warnings

Warning signs

Which signals would have flagged the trajectory? Why were they missed?

04 · narrative

Narrative-at-the-time

What everyone said before the failure was visible.

05 · autopsy

Autopsy

What the field eventually concluded, with sources.

Data model

Failure

field

type

note

uuid

Failure id

domain

enum

{eng, biotech, geo, finance, gov, social}

year

Year of culmination

mode

string[]

Failure mode tags

loss

string

Quantified loss (lives, dollars, missions)

sources

url[]

Primary sources