Answers are grounded in this paper and cite their sources.
Using Le Cam’s notion of deficiency between experiments, one can measure the value of obtaining an additional observation, when one has already been given n n independent and identically distributed observations. We show that for certain types of non-parametric experiments, the value of an additional observation decreases at a rate of 1 / n 1/√n . This is distinct from the typical rate of 1 / n 1/n in the case of parametric experiments, and from the non-decreasing value in the case of very large experiments. In particular, the rate of 1 / n 1/√n holds for the experiment given by observing samples from a density about which we know only that it is bounded from below by some fixed positive constant. For such densities, there exists no estimator that is consistent in total variation distance. We have therefore an example of an experiment which is, in this sense, not estimable, but for which the value of additional observations tends to zero.
No takes yet. Share an insight, caveat, or question.
Tilo Wiklund (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: