Abstract
A recently discovered probability estimation technique based on reconstructability analysis of an initial relative frequency estimate is examined. Analytical and experimental results suggest that this technique does not differ markedly from previously developed methods of probability estimation.