Spatio-temporal analysis of fluctuations may reveal a lot of information about a complex system. In particular, it may allow one to quantify the strength of stabilizing mechanisms, and to estimate the distance from a subcritical bifurcation. However, to achieve a reliable analysis, the characteristics of the process noise, and, more important, the statistics of the measurement noise, have to be known with reasonable accuracy. The capabilities of our new techniques will be demonstrated using large-scale social and ecological databases. We will show how drunk are bird population and their observers are, why you cannot see the (rain) forest for the trees, and why parents make wrong choices when they pick a name for their baby.