[puzzledobserver]

I suspect that it isn't the unpredictability, rather it is the acceptable margin of error.

Say you have a 1500 kg car with a 100 hp engine, which starts from a standstill and runs for 5 seconds. How fast will the car be moving at the end of this run? This depends on how much friction the vehicle experiences, and is difficult to accurately predict. Try a similar calculation with a rocket in space, and you can calculate its final speed with a very high degree of certainty.

But now consider the potential impact of a malfunction: if the car engine fails to start, you might be inconvenienced, but you will be fine. On the other hand, if the rocket engine fails to ignite in space, you can't perform necessary orbital maneuvers and you might be in deep trouble.

It's something about environment on Earth being fundamentally hospitable to human life, and environment in space being similarly inhospitable. Because of this, malfunctions and unpredictable system dynamics on Earth are most often just minor inconveniences, while malfunctions in space / underwater / deep underground are disastrous events.

[cesarb]

> But now consider the potential impact of a malfunction: if the car engine fails to start, you might be inconvenienced, but you will be fine. On the other hand, if the rocket engine fails to ignite in space, you can't perform necessary orbital maneuvers and you might be in deep trouble.

On a smaller but more relatable scale: it's like the difference between a car or a train, and an airplane. When you have a malfunction on a car or a train, an appropriate response is just to stop; when you have a malfunction on an airplane, you can't just stop, you have to keep things working long enough to get to land before you can stop.