WTF is statistical noise?

Statistics are all about generalising data into broad patterns. Taking averages, determining the variance, making trend lines, etc. When determining averages and trends, we have to ignore the occasional outlier. In any data set there is likely to be a few strange data points that do not match the pattern. For statistical purposes, we can simply ignore those as "noise".

For example, in a room full of 100 adult men and 100 adult women, we will find that the men are taller in general. A few of the men will actually be shorter than a few of the women, but on average the men will be taller overall. The outliers (odd-balls) that go the other way are just statistical noise that effects the average result very little.

This is true even more so when your room has 10,000 men and 10,000 women. The averages will barely be effected by the noise data.