-
The middle manager's decade: delayering from the 1990s to 2026
Computers were supposed to end middle management in 1958, again in the 1990s, and now with AI. Amazon, Meta and Google are flattening. Gallup's data show where wide spans work and where they fail.
-
Will they hire us back? What the 1990s downsizing wave teaches
The 1990s re-engineering wave and the busts of 2001 and 2008 answer the question every laid-off worker asks: some people are rehired, late, and often on a contract. Here is what that history predicts for 2027–28.
-
Ten to twenty years: why technology's payoff always comes late
Electricity took decades to transform factories, and computers about twenty years to show up in productivity statistics. The productivity J-curve explains why, and why cutting staff during the dip may delay AI's payoff.
-
Europe's layoffs are not about AI: cars, telecoms and the state
Europe's November 2026 layoffs come from autos, telecoms and Romanian state companies, not AI. Consultation law means cuts are announced early and carried out late, often in 2027 and 2028.
-
The first rung: what happens when entry-level jobs disappear
Employment of 22-to-25-year-olds in the most AI-exposed jobs is now 19% below trend, and Germany shows a similar drop. Guilds, the B-2 bomber and simple arithmetic show what firms lose when the first rung disappears.
-
Outages are getting rarer and bigger: ten years of global failures
Data-centre outage rates have fallen five years running, yet major cloud incidents rose from 7 in 2024 to 18 in 2026. A decade of global failures, the causes that repeat, and my forecast: a 90–95% chance of another global outage within a year.
-
Normal accidents in the cloud: what Perrow, Cook and Dekker predicted
Perrow, Cook, Dekker and Geer predicted how cloud outages unfold: small failures interacting in tightly coupled systems, no single root cause, slow drift, and monoculture. Their words mapped onto CrowdStrike, AWS October 2025 and Microsoft November 2026.
-
Voluntary first, compulsory later: how a buyout really works
A buyout runs in four steps: the window, the count, the gap and the quiet part. November's cases at Microsoft, JLR, Michelin and Disney show the pattern, and the research says the most mobile people take the money first. General information, not advice.