[ethbr1]

The difference is compounding, unstable systems.

In much simpler form, it's similar to the "Where will this satellite in a decaying orbit land?"

The difficulty is not the approximation: the difficulty is that whether drag begins to increase at t=x or t=x+30sec (based on orientation when it first encounters atmosphere) cascades throughout the rest of the equations and dramatically changes the results.

[deadbabe]

Monte Carlo it

[macksd]

Are you a product manager by any chance? Because I can totally imagine you telling someone to just use some cloud DB because it's infinitely scalable.

[ethbr1]

That doesn't help if the convergence is a bi+ modal distribution, due to the underlying nature of the problem.