Several experiments [1-3] show significant deviations from predictions of the statistical model of nuclear reactions. We summarize unsuccessful recent theoretical efforts to account for such disagreement in terms of a violation of orthogonal invariance caused by the Thomas-Ehrman shift. We report on numerical simulations involving a large number of gamma decay channels that also give rise to violation of orthogonal invariance but likewise do not account for the discrepancies. We discuss the statistical model in the light of these results.