Modern statistical applications often require researchers to make inferences about large numbers of variables from relatively small datasets. One therefore often has to ask: what is the minimum number of datapoints required to fit a particular model? This question has surprising deep connections to rigidity theory and that is the subject of this talk.