Adrian Vermeule (Harvard Law School) has posted Optimal Child Abuse (Northwestern University Law Review, Forthcoming) on SSRN. Here is the abstract:
I will argue that under contemporary social conditions, in contrast to those that characterize traditional pre-modern forms of social organization, child abuse is not something to be strictly minimized, but rather optimized. A system of child protection, including criminal prosecutions and child protective services, should tolerate a predictable level of child abuse as the inevitable byproduct of attaining other ends that are desirable overall.
There are three principal grounds for this claim. First, the architects of the contemporary regime that combines criminal law with the case-worker model of child protective services were not only worried about the abuse of children by their patents. They were equally concerned about the harms to children that result from removal of children from parents and siblings and mistreatment by government alternatives, including in traditional orphanages, group homes, and foster care. Second, the rate of child abuse in the contemporary era is greater than in the early twentieth century — so much greater that the the system of child protection has been forced to tolerate a relatively high level of child abuse. Third, the costs of government placement of abused children are necessarily positive and plausibly large, in part because any institutional monitors created to detect and punish abuses must themselves be monitored for abuse.
The architects of the modern system of child protection believed that the optimal system would inevitably involve tradeoffs between the benefits of removing children from abusive environments and the costs of a government system for the care of children removed from such environments. In that sense, the modern system constantly gropes towards an institutional package solution that embodies an optimal level of child abuse.
Recommended, although this framing of the issue may not be the most helpful in policy debates outside the academic context.