The Straight Line Instinct

Core idea

The straight line instinct is the urge to extend any visible trend with a ruler. Show people three points that rise, and they imagine the fourth rises by the same amount — even when the underlying process is a doubling, a saturation, or a one-time bulge that is already flattening. The instinct served our ancestors well for predicting where a flying stone would land. It serves us poorly for predicting population, disease outbreaks, or the cost of vaccines.

Rosling's argument: real-world curves come in at least five shapes — straight, S-bend, slide, hump, doubling line. Treating every line as straight is the source of both alarmist forecasts (the population will explode) and dangerous underestimates (Ebola will be contained).

The rule of thumb: lines might bend. Before extrapolating, ask which shape this curve is, and what you would expect it to look like beyond the data you have.

Why it matters

Forecasting in straight lines distorts both panic and complacency

In the same topic Rosling tells two stories. One is Ebola in 2014, where he and most observers initially read a doubling line as a straight one and were slow to react. The other is global population, where billions of people read a straight line where the experts see an S-curve flattening at 10–12 billion. Both errors come from the same instinctive shortcut. One produces fatal delay, the other produces panicked overreaction. The instinct is not biased in one direction — it is biased toward whichever shape happens to look like the segment of curve you can see.

A child's growth chart proves you already know this

Rosling's grandson Mino grew seven inches in his first six months. Extrapolate straight: thirteen-foot toddler. Everyone laughs because we have firsthand experience of growing bodies and we know the line bends. We do not have firsthand experience of national fertility transitions, viral epidemics, or technology adoption curves — so the same naive extrapolation gets through unchallenged. The instinct is not stupid; it is just untrained on unfamiliar terrain.

Peak child has already happened

The single most important fact in the topic, and the one most people get wrong: the number of children in the world is no longer increasing. World population will keep growing for several decades, but not because more children are being born — because the children already alive are growing into adults. This 'fill-up effect' has three generations of momentum and then it stops. Reading the population curve as a straight line misses this completely.

Key takeaways

Mental model

Copy sourceZoom

Practical application

Example

A startup is signing up 200 new users per week. After eight weeks of steady growth the team plans for the next year by drawing a straight line: 200 a week, 10,000 by year-end. Hiring, infrastructure, and burn-rate forecasts all assume this trajectory.

But the underlying mechanism is referral, not advertising. Each new user brings in 0.3 more users on average within their first month. That is not a straight line — it is the early arm of a doubling line. The team will spend Q1 noticing that growth is exceeding plan, Q2 noticing that the gap is widening, and Q3 discovering that the database falls over because their capacity plan was off by an order of magnitude. The straight-line read is not 'pessimistic' or 'optimistic'; it is wrong in a specific way that matters.

The opposite error shows up elsewhere. A team running a vaccination campaign in a low-Level-2 region reaches 60 percent coverage in two years and the funder, reading a straight line, demands 100 percent in another year. But vaccination follows an S-bend — the last 20 percent (vaccine-hesitant households, hard-to-reach geography) is harder than the first 80, and the curve flattens. Forecasting it as straight produces unrealistic targets, missed milestones, and political pressure that misallocates resources at exactly the point where the work gets harder.

In both cases the underlying instinct is the same: extending three points with a ruler. The corrective in both cases is the same: ask which curve family this is before you draw the next segment.

Related lessons