The Failure Archive

Every mistake leaves a story.

Real failures. Strange decisions. Unbelievable consequences. Investigating the root mechanisms when systems and human psychology collide.

⚡ The Lead InvestigationIncident Dossier of the Month
The GitLab Accidental Database Deletion: How Five Silent Backup Failures Magnified a Single Typo
corporate⏱️ 5 min read

The GitLab Accidental Database Deletion: How Five Silent Backup Failures Magnified a Single Typo

The complete engineering forensic reconstruction of the 2017 GitLab database deletion. How a single engineer executed an erroneous removal command, exposing five distinct, silent failures in the distributed backup and disaster recovery architecture.

📂 The Ledger Case Files18 Documented Records (17 in Archive)
The $150 Million Halloween: How a Rushed SAP & Manugistics Rollout Paralyzed Hershey
corporate⏱️ 10 min

The $150 Million Halloween: How a Rushed SAP & Manugistics Rollout Paralyzed Hershey

A forensic analysis of the 1999 Hershey ERP implementation disaster, where an aggressive 30-month timeline, skipped testing phases, and a Big Bang rollout left the company unable to ship candy for Halloween.

Read Incident →
Patriot Missile Software Failure: How a 24-Bit Timing Error Caused the Dhahran Disaster
military⏱️ 5 min

Patriot Missile Software Failure: How a 24-Bit Timing Error Caused the Dhahran Disaster

How a 24-bit limited-precision time conversion accumulated over 100+ hours and caused the Patriot system to lose track of an Iraqi Scud at Dhahran in 1991.

Read Incident →
The £330 Million Aftermath: How TSB's Core Banking Migration Failed
corporate⏱️ 5 min

The £330 Million Aftermath: How TSB's Core Banking Migration Failed

When TSB completed a multi-stage migration to Sabadell's Proteo4UK platform in April 2018, the customer data transferred successfully — to the penny. What followed was a failure of assurance: inconsistent data-centre configuration that extensive testing never exposed, thousands of unresolved defects at go-live, and a platform that collapsed when exposed to the scale and conditions of live customer traffic.

Read Incident →
Mars Climate Orbiter: How a Metric–Imperial Software Interface Error Caused a $125 Million Mission Loss
corporate⏱️ 5 min

Mars Climate Orbiter: How a Metric–Imperial Software Interface Error Caused a $125 Million Mission Loss

The complete engineering forensic reconstruction of the 1999 Mars Climate Orbiter loss. How an unverified software interface contract in the SM_FORCES ground system violated metric specifications, corrupted trajectory modeling, and survived nine months of cruise operations.

Read Incident →
Therac-25 Radiation Overdose: How a Race Condition, Counter Overflow, and Missing Hardware Interlocks Caused a Safety-Critical Failure
corporate⏱️ 5 min

Therac-25 Radiation Overdose: How a Race Condition, Counter Overflow, and Missing Hardware Interlocks Caused a Safety-Critical Failure

A forensic reconstruction of the 1985–1987 Therac-25 radiation overdoses. How undocumented code reuse, a concurrent state-consistency race condition, a 1-byte counter rollover, and the removal of physical hardware interlocks combined in a catastrophic systems engineering failure.

Read Incident →
The 16-Bit Integer That Destroyed a $370 Million Rocket
corporate⏱️ 5 min

The 16-Bit Integer That Destroyed a $370 Million Rocket

How an unprotected legacy variable conversion caused the Ariane 5 rocket to self-destruct 39 seconds after launch, exposing the lethal danger of unquestioned software reuse.

Read Incident →
The $6.2 Billion Risk-Model Failure: Inside JPMorgan’s London Whale
corporate⏱️ 16 min

The $6.2 Billion Risk-Model Failure: Inside JPMorgan’s London Whale

In 2012, JPMorgan Chase's Chief Investment Office suffered $6.2 billion in losses on its Synthetic Credit Portfolio amid a breakdown in risk-model governance, manual spreadsheet controls, risk limits, valuation practices, and trading oversight. A flawed new VaR methodology materially understated reported risk.

Read Incident →
When the Database Lied: Inside Target Canada's $2.1 Billion Supply Chain Collapse
corporate⏱️ 14 min

When the Database Lied: Inside Target Canada's $2.1 Billion Supply Chain Collapse

How corrupted product data, an aggressive launch schedule, and tightly coupled warehouse systems turned a digital data-quality problem into a physical logistics crisis.

Read Incident →
The UK Post Office Horizon IT Scandal: When the Computer is Always Right
corporate⏱️ 15 min

The UK Post Office Horizon IT Scandal: When the Computer is Always Right

Fujitsu's buggy Horizon EPOSS software generated massive phantom shortfalls across UK post offices. Armed with the belief that 'the computer is always right,' the Post Office wrongfully prosecuted over 900 subpostmasters.

Read Incident →
You Can't Automate Drywall: Inside Zillow's $408 Million Algorithmic House-Flipping Collapse
money⏱️ 11 min

You Can't Automate Drywall: Inside Zillow's $408 Million Algorithmic House-Flipping Collapse

In late 2021, Zillow shuttered its algorithmic house-flipping division, recording a $304.4 million Q3 inventory write-down ($407.9 million for the year) and reducing its workforce by 25%. Here is the forensic anatomy of a system where prediction throughput outran physical throughput.

Read Incident →
The $900 Million Checkbox: How Three People Looked at the Same Screen and Wired an Entire Bank's Money
money⏱️ 12 min

The $900 Million Checkbox: How Three People Looked at the Same Screen and Wired an Entire Bank's Money

A confusing enterprise interface, an unselected checkbox triad, and three layers of rubber-stamp approval caused Citibank to accidentally wire $893.9 million to hostile hedge funds. Here is what the evidence actually shows.

Read Incident →
He Deleted the Production Database and Let the Intern Take the Fall: Inside a Startup's Midnight Scapegoat Coverup
work⏱️ 12 min

He Deleted the Production Database and Let the Intern Take the Fall: Inside a Startup's Midnight Scapegoat Coverup

At 11:42 PM, a senior lead ran DROP DATABASE on production instead of staging, discovering that 6 months of S3 backups were dead. What followed was an anatomy of workplace terror and scapegoating.

Read Incident →
The $440 Million Trading Glitch That Killed Knight Capital in 45 Minutes
money⏱️ 12 min

The $440 Million Trading Glitch That Killed Knight Capital in 45 Minutes

On August 1, 2012, Knight Capital lost $440 million in 45 minutes when a legacy feature called 'Power Peg' woke from an eight-year sleep and flooded the market with 4 million rogue orders. Here is the forensic post-mortem.

Read Incident →
The 1-Yen Stock Order That Broke Tokyo: How Mizuho Securities Sold 610,000 Shares of a Company That Only Had 14,500
money⏱️ 12 min

The 1-Yen Stock Order That Broke Tokyo: How Mizuho Securities Sold 610,000 Shares of a Company That Only Had 14,500

On December 8, 2005, a broker typed 'sell 610,000 shares at 1 yen' instead of 'sell 1 share at 610,000 yen'. Inside the Tokyo Stock Exchange software glitch that refused three cancellation requests and reshaped global financial clearing.

Read Incident →
The $105 Billion Typo: How Samsung Securities Issued 2.8 Billion Ghost Shares and Triggered a 37-Minute Trading Free-for-All
money⏱️ 12 min

The $105 Billion Typo: How Samsung Securities Issued 2.8 Billion Ghost Shares and Triggered a 37-Minute Trading Free-for-All

On April 6, 2018, a clerk entered 1,000 shares instead of 1,000 won. Inside the internal ledger glitch that minted $105 billion in unbacked phantom equity, sparking a 37-minute insider selling frenzy.

Read Incident →
The Bedroom Trader Who Helped Trigger the $1 Trillion Flash Crash: Navinder Sarao and the 2010 Market Fracture
money⏱️ 12 min

The Bedroom Trader Who Helped Trigger the $1 Trillion Flash Crash: Navinder Sarao and the 2010 Market Fracture

On May 6, 2010, the US financial market plunged nearly 1,000 points in thirty-six minutes. Five years later, federal investigators traced a key catalyst to a modest bedroom in suburban London.

Read Incident →
The Friday Morning Update That Took Down 8.5 Million Computers: The Anatomy of CrowdStrike's Channel File 291
corporate⏱️ 12 min

The Friday Morning Update That Took Down 8.5 Million Computers: The Anatomy of CrowdStrike's Channel File 291

At 04:09 UTC on July 19, 2024, a 40KB configuration file was pushed to 8.5 million Windows machines. Within minutes, the global economy froze. Here is the forensic systems post-mortem.

Read Incident →
ErrorLedger Editorial Ethos

AI can answer questions. ErrorLedger tells you what happened.

Every entry in ErrorLedger investigates the deeper mechanics of human and corporate failure. We dissect the flawed incentives, systemic pressures, and organizational blind spots that turn minor compromises into catastrophic disasters.

Strict Claim Firewall: Verifiable court records and primary disclosures.
The Archivist's Verdict: Systemic root causes over simple scapegoating.
Live Reader Verdicts: Genuine community debates and blame distribution.