CrashOS / museum edition ALL SYSTEMS: QUESTIONABLE

am I the bug?ORIGINAL CONCEPT ART / CLICK THE LITTLE WEIRDO
readme_before_it_breaks.txt
THE MUSEUM OF DIGITAL CATASTROPHES

Everything is
fine. on fire.

Real failures. Unbelievable consequences. Meet the bugs, bad calls, and broken assumptions that changed our digital world.

Open the archive ↗Not sure where to start? Follow a trail →
expensive_mistake.exe
$460 million.
45 minutes.Knight Capital / open the case file ↗
disaster_lab.exePractice a bad day ↗system_check.exeFind the weak points ↗
16 REAL CASES / ZERO PERFECT SYSTEMSthat’s all folks
16source-linked case files
5curated reading trails
5hands-on rehearsals
Facts first. Lessons that last. How we build the archive ↗
ARTIFACT 03 / FAULT LINE · CONCEPT ART

CASE FILE 002 / KNIGHT CAPITAL

45 minutes.
$460 million gone.

Dormant code. An incomplete deployment. A trading system that wouldn’t stop. Follow the chain of failure—and the safeguards that were missing.

Read the case file ↗Explore the interactive timelineYou’re on call: play the incident

NEW IN THE ARCHIVE / AI CONTAINMENT

When the task crosses the boundary.

View all 16 case files ↗
You’re on call: contain the collective

YOUR NEXT CALL

The incident campaign is open.

Investigate evidence, earn response distinctions, and build a personal field manual.

Start the campaignThis week’s mysteryRun a team tabletop

THE DISASTER LAB / YOUR TURN

History repeats.
Practice first.

Choose your kind of trouble.
Rehearse a fictional incident, or examine the safeguards in your own system.

01 / disaster_simulator.exe
FICTIONAL INCIDENTS3–5 MIN

A safe place
for a very bad day.

The release is live. Something is wrong. What do you do next?

SCENARIO PREVIEW

09:41 Orders start behaving strangely.

09:43 The alerts keep coming.

Your next decision matters.

Choose a scenarioMake the callsGet a debrief
Choose a rehearsal

5 scenarios · AI containment, releases, recovery, commerce & dependencies

02 / disaster_check.exe
YOUR SYSTEM, YOUR EVIDENCE4 SAFEGUARDS

Find the cracks.
Make a plan.

Look beyond “we should be fine.” What have you actually tested?

  1. 01
    RecoveryCan you get your data back?
  2. 02
    ReleasesCan you contain a bad change?
  3. 03
    DependenciesWill your fallback really work?
  4. 04
    InterfacesDo both sides agree?
Review my safeguards

Leave with evidence gaps, next steps, and an editable action plan.

NO PRODUCTION ACCESS

Authored rehearsals and self-reported reviews. No scanning, no predictions, no live AI.

KEEP INVESTIGATING

Follow a recurring failure.

All 5 reading trails
READING TRAIL 013 CASES · ABOUT 10 MIN

The change that spread.

When a small update gets a very large audience.

Follow the trail ↗
READING TRAIL 023 CASES · ABOUT 10 MIN

The way back.

Having a recovery plan is different from proving one.

Follow the trail ↗