// DAMAGE_CLOCK — avg rate since 558

By The Numbers

$

estimated total damage since 558 — ticking at $66/sec

Based on 171 documented incidents spanning 1468 years. Rate = total damage ÷ timespan. Recalculates with every new article. Methodology →

01. The Big Picture

171 incidents documented 558 to 2026
1468 years of documented failure oldest: Hagia Sophia: Justinian Declared "I Ha… 558
$3.0T total estimated damage societal cost, not just fines
$18.3B average damage per incident median is lower — outliers pull hard
15.0B people affected documented affected users across all incidents
$203 avg cost per affected person total damage / total affected users

02. The Nerd Stats

// BURN_RATE

$236,940/hour

Money burned per hour since the oldest documented incident (Hagia Sophia, 558). That's $2.1B/year averaged across 1468 years of human folly.

// STACK_HEIGHT

3,323 km high

Stacked in $100 bills (0.109 mm each), the total damage would reach 3,323 km — 8.3× the distance to the ISS (400 km orbit). That's 0.0086× the distance to the Moon, or 378× the height of Everest.

// MOST_EXPENSIVE_INCIDENT

$738.5B

Chernobyl Reactor 4 Explodes During Safety Test That Had Disabled Its Safety Systems — the single most costly incident in the archive.

// MOST_EXPENSIVE_BUG_PER_CHARACTER

$195M / char

Mariner 1 (1962): a missing overbar symbol in the guidance code destroyed a $195M rocket 4 minutes after launch. One character. One rocket. Gone.

// WORLD_GDP_FRACTION

2.8% of world GDP

Total documented damage represents 2.8% of current world GDP (~$110 trillion). In other words, our archive documents enough damage to run the entire global economy for 10 days.

// ISS_FUNDING

1,016 years of ISS

The International Space Station costs ~$3 billion/year to operate. The documented damage in our archive could fund it for 1,016 years — until the year 3042.

03. The Human Cost

Financial figures dominate tech failure coverage. But some of these incidents killed people. This section documents the confirmed death toll and its theoretical long-term impact.

347,116 confirmed deaths across 20 incidents with documented fatalities
45,563,257,800,209 theoretical people who never existed generational model — see methodology below

// GENERATIONAL_LOSS_MODEL

For each incident with documented fatalities: each person who died had a line of descendants that was cut short. We model: generation length = 28 years, children per person = 2.1 (global average). For each incident, we calculate generations elapsed since the incident year, then sum the theoretical descendants across all generations. The result is a conservative estimate of people who theoretically never existed due to each incident.

This is a thought exercise, not a precise demographic projection. It is presented as a sense-of-scale metric, not a moral calculus. The goal is to make the scale of human loss harder to dismiss.

Only incidents with confirmed fatality figures are included. Many more incidents caused indirect deaths not captured here. "Theoretical descendants lost" = sum(2.1n, n = 1 … generations since incident), based on a global average of 2.1 children per person and 28-year generation length.

04. Who's Responsible

Leadership attribution for incidents where the primary decision-maker is publicly identifiable. This is about power and accountability, not blame. The question is whether failure rates match leadership representation.

Of 66 attributable incidents (105 unattributed/institutional of 171 total):

Male-led
93.9% (62 incidents)
Female-led
3.0% (2 incidents)
Mixed
3.0% (2 incidents)
Expected (baseline)
83% male leadership share

93.9% of attributable fails were led by men. Given men hold ~83% of tech leadership roles (Crunchbase baseline), their “fair share” of tech fails would be 83%. They are over-represented by 10.9 percentage points.

Attribution methodology: incidents are tagged based on the primary publicly-named decision-maker at the time of the failure. Institutional/government incidents where individual leadership is unclear are excluded from the attributable count. This covers 66 of 171 incidents.

05. The Calendar of Failure

Based on 74 incidents with known founding dates. Do some months produce more failures than others — or are companies just unlucky when they launch?

// FOUNDING_MONTH — when were these companies born?

Worst: Apr (8 companies). 74 companies with known founding month.

Jan 1
Feb 4
Mar 4
Apr 8
May 5
Jun 4
Jul 4
Aug 2
Sep 5
Oct 3
Nov 2
Dec 3

// FAIL_MONTH — when did they crash?

Worst: Oct (21 incidents). Based on all 171 incidents.

Jan 19
Feb 11
Mar 10
Apr 16
May 12
Jun 11
Jul 17
Aug 15
Sep 11
Oct 21
Nov 17
Dec 11
24.2 years average time from founding to failure median is lower — legacy giants skew the average. Based on 74 incidents with known founding dates.

06. The Acceleration

Tech failure is not evenly distributed across time. As technology penetrated every layer of society, so did the damage when it broke. 19 incidents predate 1960 — the chart below shows the era where scale became systemic.

// INCIDENTS_PER_DECADE

1960s 3
1970s 3
1980s 10
1990s 14
2000s 14
2010s 66
2020s* 42

* 2020s: only 6 years of data. Annualised rate = 70 incidents/decade — on track to be the worst.

// DAMAGE_PER_DECADE

1960s 15B
1970s 5B
1980s 1.3T
1990s 15B
2000s 497B
2010s 517B
2020s 661B

1990s damage dominated by Y2K ($300B) and Black Monday ($528B). 2020s damage includes Chernobyl ($739B, incident 1986, added to archive in current decade).

07. Concentration

Tech failure follows a power law. A handful of catastrophic incidents account for the overwhelming majority of documented damage — the rest are expensive footnotes.

81% of total damage from just 10 incidents out of 171 total — the archive is a long tail of disasters with a very fat head

08. Repeat Offenders

Some names appear in the archive more than once. Attribution is based on title matching — a company appears here if it is the primary subject of the documented incident.

01 Microsoft
$18.0B
02 Google
$2.6B
03 Amazon
$1.7B
04 Yahoo
$5.7B
05 Boeing
$22.2B
06 Tesla
$803M
07 Apple
$500M
08 NASA
$12.5B

Total damage shown is the sum of all documented incidents for that company in the archive. A company appearing many times does not necessarily mean it causes the most damage — small repeated failures can matter more than one large one.

09. Context

World GDP ($110T)

Our archive = 2.8% of world GDP

ISS annual budget ($3B/yr)

Archive damage could fund the ISS for 1,016 years

Most expensive per character

$195M — Mariner 1’s missing overbar, 1962. One symbol. One lost rocket.

Burn rate since 558

$236,940/hour averaged across 1468 years of documented failure