Pooling, double descent and classical emergence

In statistics, multiple data points are often pooled together such that it is as if they come from identical random variables - one example of heavy pooling is a fixed effects model. As the number of data pooled successfully increases, the law of large numbers kicks in as to eliminate stochastic variations in the sample average of pooled data. This eliminates stochastic variations such that a model effectively becomes deterministic, with N data points aggregated into one data point. In the big-data regime then, the bias-variance tradeoff becomes largely meaningless - more the model complexity, better the performance. The example of how classical mechanics emerges from quantum mechanics demonstrates such a breakdown.