> Wow. 1,959 detections out of 10^21 interactions. And they talk about statistical significance. In any other field that is so close to zero you would ignore the results.

1959/1e21 is not the measure of the significance in the study though. That total of 1e21 interactions is split into different types of behaviors (based on theory and previous results). It's deltas on those categorized interactions and comparisons with noise/background models/characterizations that are used to assess the statistical significance of the 1959 events. I am not saying this has to be a true result and not a statistical fluke, but 1959/1e21 is not the "sigma" of this experiment.