[in_vestor]

Sounds like we’re on the same page. As I’ve thought about this, I can’t escape the disorienting feeling that many more filters are in the past than in the future (and ones with worse probabilities are in the past too). Do you perceive the same? And does the cumulative probability of future filters seem smaller than the cumulative probability of past filters?

[ben_w]

Given all the exoplanets with no signs of life, either abiogenesis is probably very hard or life is very fragile (~1e-3 or fewer planets will both develop life and retain it for long enough to pass through an oxygenation catastrophe, but that's squished several filters together).

Difficult to do more than guess past filters beyond oxygenation given the sample size of n=1.

Future? All unknown-unknowns. Even if you rule out paper-clipping-AI-gone-wrong scenarios by assuming only weak and narrow AI slightly less than we have today, the mere ability to get a million colonists to Mars, even with just SpaceX's Starship, requires enough space industry to be a direct military threat to Earth.