All sorts of biological / ecological factors would come into play like average lifespan, the size of life in relation to the size of the planet, resource scarcity, reproduction rates, mobility, avg. number of offspring, mutation rates, weather patterns, geography, etc.
Then you get into social and technological factors.
We have a sample size of 1, so it’s hard to extrapolate or understand where we are on the probability curve.
The best data point we might have for a self-inflicted civilization ending event would be looking at it from a smaller point view. How many times in history has one group attacked another such that both groups ceased to exist?
One positive sign is that even with all our destruction and violence, someone usually wins or backs down with the ability to continue on living.
Of course there are always long-tail events like mutually assured destruction, but I don’t know if 100 years is the correct timetable for ~100% certainty.
There was an interesting HN discussion about it a few weeks ago here with folks chiming in on the best ways to assess the odds of a long-tail event like MAD:
How would that work? It seems to me the only that matters is that the proper conditions for abiogenesis be maintained for whatever duration it requires. It shouldn't matter how far along the Universe or planet history this period occurs