A second world should be built at prevention pace, not race pace. By Toby Ord's estimates for this century (degrees of belief, not measurements), natural risk is about 1 in 10,000 and anthropogenic risk about 1 in 6, and he says a second planet partly decorrelates war, disease and tyranny but almost not at all unaligned AI, his largest single entry at about 1 in 10. So racing to arrive a century earlier pays only if it raises anthropogenic risk by less than a break-even of about 0.06% (if the second world removes only natural risk) to about 6% (assuming, beyond Ord, that it also absorbs half of nuclear, climate and other environmental risk and a quarter of engineered pandemics); and a race is most likely to raise the AI and weapons risks a second world cannot absorb. Jared Diamond's Easter Island shows the race form: with no emigration as an escape valve, clan competition for bigger statues spent wood, rope and food that could not be replaced. A redundancy programme run as competitive display draws down the shared stock it was meant to insure.

groundsSource document

the impossibility of emigration as an escape valve because of Easter's isolation, a focus on statue construction for reasons already discussed, and competition between clans and chiefs driving the erection of bigger statues requiring more wood, rope, and food.

groundsSource document

Easter Island was as isolated in the Pacific Ocean as the earth is today in space. When the Easter Islanders got into difficulties, there was nowhere to which they could flee, nor to which they could turn for help. Nor shall we modern earthlings have recourse elsewhere if our troubles increase.

groundsThe Precipice, Toby Ord

the relative loss each year is actually rather slow, about one part in five billion, and it is this relative reduction in our potential that matters. The pressures towards prudence and wisdom are larger in relative terms, and counsel care over haste. If rushing to acquire these technologies a century earlier diminished our chance of survival by even one part in fifty million, it would be counterproductive

groundsThe Precipice, Toby Ord

And the total natural risk, around 1 in 10,000. Nuclear war, around 1 in 1,000. Climate change, around 1 in 1,000. Other environmental damage, around 1 in 1,000. Naturally arising pandemics, around 1 in 10,000. Engineered pandemics, around 1 in 30. Unaligned artificial intelligence, around 1 in 10. Unforeseen anthropogenic risks, around 1 in 30. Other anthropogenic risks, around 1 in 50. And the total anthropogenic risk, around 1 in 6. And the total existential risk, around 1 in 6.

groundsThe Precipice, Toby Ord

risks, such as disease, war, tyranny, and permanently locking in bad values, are correlated across different planets. If they affect one, they are somewhat more likely to affect the others too. A few risks, such as unaligned AGI and vacuum collapse, are almost completely correlated.