The Castle Bravo test is a good example of that. More than double the expected yield, simply because our model of Lithium 7 was inaccurate. https://en.m.wikipe

The Castle Bravo test is a good example of that. More than double the expected yield, simply because our model of Lithium 7 was inaccurate. https://en.m.wikipedia.org/wiki/Castle_Bravo https://en.m.wikipedia.org/wiki/Castle_Bravo Big errors like this have all been tested and modeled away, but the problem is exceedingly complex, and I am just as certain enough uncertainty remains in the current models to be a genuine concern.