4.1 Why extinction would be different

In August 1939, Albert Einstein signed a letter to President Franklin Roosevelt. It warned that a nuclear chain reaction could be used to build "extremely powerful bombs of a new type". Six years later, two such bombs destroyed Hiroshima and Nagasaki. For the first time, a handful of decisions by a handful of people could plausibly end civilisation.

Since then, the list of ways humanity could seriously damage its own future has grown: nuclear war, engineered pandemics and, some argue, advanced AI. People who study these risks call the worst of them existential risks. These are events that would cause human extinction, or would permanently and drastically curtail humanity's potential.

Why treat these as a separate category? The philosopher Derek Parfit offered a thought experiment. Compare three outcomes:

1. Peace.
2. A war that kills 99% of the world's population.
3. A war that kills 100%.

Most people say the biggest difference is between 1 and 2. Parfit thought the difference between 2 and 3 was far greater. After the second, humanity could recover. After the third, there is no one left to recover, and no future generations at all.

How likely is any of this? Make your own estimate first.

There is one more reason some people put extra weight on these risks: the future could be very large. If humanity survives as long as a typical mammal species, and the population stays at about 11 billion, Max Roser estimates that around 100 trillion people are yet to be born. Will MacAskill argues in What We Owe the Future that people far away in time matter as much as people far away in space. If that is right, an extinction event would destroy not only the lives of everyone alive, but all the lives that would otherwise have followed.

You don't need that argument for this unit to matter. A pandemic or nuclear war that killed billions would be the worst event in history, whatever you think about future generations.