He also makes a comparison with an earlier geological period,  snowball Earth, about 650 million years ago, when possibly nearly all or most of Earth was covered in ice and snow. The concentrations of carbon dioxide built up through this period, over millions of years, as volcanoes continued to emit it into the atmosphere. This is the slow carbon cycle and normally the CO2 would dissolve back into the ocean, taken out of the atmosphere from rain or directly absorbed into the sea. But during a snowball Earth, there's no open water, or almost none, to absorb the CO2. Meanwhile the snow and ice has no effect at all on volcanoes, which continue to add it to the atmosphere as usual, and that's why the carbon dioxide builds up. Eventually enough forms to warm up the world again and the ice melts.

He remarks in the paper that when the ice melted finally, after snowball Earth, there could have been as much as 10% of CO2 in the atmosphere - compared to 0.04% right now. The sun was also 6% cooler than it is today. This didn’t lead to a runaway greenhouse either which is pretty conclusive evidence that it can't happen easily.

He mentions in the paper that if we had 10% CO2 in the future - after a snowball Earth, and with our 6% warmer sun, then perhaps it would trigger a runaway greenhouse when the ice melted. However, that’s hardly likely to happen, so it is a very theoretical idea, perhaps relevant to planets orbiting other stars?

- a bit unsure about this because of slushball Earth, so the 10% CO2 is only one hypothesis.