It turns out that Columbia law student Philip Ellenbogen developed the same idea in a 2009 article for the Columbia Journal of Law and Social Problems (see esp. pp. 363-68). As he puts it (footnotes omitted):
[A]victim of prison rape is often much smaller and weaker than the perpetrator of the violence. Therefore, while prior violence should certainly be taken into account, the classification system should incorporate the strength and size of the inmate as well. This Note encourages prison officials to continue to use degree of violence as a factor but within that violence classification, to check each inmate for height and weight, and classify each according to a sliding scale….
Without a size advantage or superior strength, the act of raping someone becomes quite difficult.
Ellenbogen’s article also summarizes evidence showing that prison rape is a serious problem in Canadian prisons as well as in the US, despite the less harsh reputation of the Canadian criminal justice system. This reinforces my point that the failure to control prison rape is in large part the result of structural flaws of government rather than idiosyncratic policy failures. [...]