What Happened Every Other Time

580 years of technology disruptions, 20 case studies, and what actually happened to the workers. History doesn't predict the future — but it narrows the range of plausible outcomes.

By Philipp Maul · Nexalps · Early 2026 · Part 3 of 7 in the European AI Labour Market suite

Based on 52 primary sources including Acemoglu & Restrepo, Dittmar, Feigenbaum & Gross, David and Autor.

Time Span
580 years
Printing press (1440) to AI (2022)
Median Adoption Lag
20–50 yrs
Market entry to measurable productivity
Disruption Patterns
5 types
Create, Expand, Absorb, Restructure, Eliminate
Worst-Case Job Loss
90%
Containerisation of dock work

1. The Adoption Timeline Every general-purpose technology follows a multi-decade arc from market entry to restructuring

10 general-purpose technologies (9 historic + AI), plotted across 4 phases. AI is in the earliest stages of adoption — the historical pattern predicts 15–40 years to full restructuring.

2. Five Disruption Types Not all disruptions work the same way — the type determines who wins and who loses

Every occupation-level disruption in our dataset follows one of five types. The diagnostic framework below helps predict which one.

Net job creation or preservation
Jobs persist but transform or degrade
Jobs eliminated

Diagnostic Framework

Six variables predict which disruption type an occupation will follow. The single most important question: does the technology automate some tasks (leaving high-value human tasks) or all core tasks?

Diagnostic framework showing how six variables predict disruption type

3. European Divergence Same technology, different outcomes

The most consistent finding across 580 years: institutional frameworks shape who bears the cost, not whether disruption occurs. Hall & Soskice’s Varieties of Capitalism (2001) explains why.

Three Varieties of Capitalism

Three institutional logics produce structurally different responses to the same technology. Neither is superior — but they specialise. LMEs (Liberal Market Economies — US, UK, Australia) coordinate through competitive markets, flexible labour, and equity capital. CMEs (Coordinated Market Economies — Germany, Nordics, Japan, Switzerland) coordinate through works councils, vocational training, and patient capital. SMEs (State-led Market Economies — China) coordinate through industrial policy, state-directed credit, and administrative control.

Comparison of Liberal, Coordinated, and State-led Market Economy institutions

Robot Density vs Manufacturing Job Loss

Germany has ~1.5× the robot density of the US but lost only 5% of manufacturing jobs (vs 36% in the US). The magnitude matters less than the pattern — CMEs redirect automation, LMEs displace labour.

Four Institutional Mechanisms That Redirect Disruption

Author’s Note

The academic framework for this divergence is Hall & Soskice’s Varieties of Capitalism (2001): Liberal Market Economies (US, UK) coordinate through markets and produce radical innovation; Coordinated Market Economies (Germany, Nordics, Japan) coordinate through institutions and produce incremental innovation; State-led Market Economies (China) coordinate through industrial policy and produce directed leapfrog innovation. Acemoglu & Johnson call the critical variable enabling vs. replacing technology: does AI augment workers or substitute them? Europe’s institutional architecture gives it a structural advantage in steering toward the enabling path — if it chooses to use it.

— Philipp Maul

Go deeper

Explore the 20 case studies in detail, or see what actually happened to the displaced workers.

20 Case Studies → Findings & Predictions → Worker Outcomes → AI Exposure ↗ Job Market ↗