New complications for old
In Mahomed the Supreme Court split 3-2 over what items of evidence were included as propensity evidence. The majority focused narrowly on one incident, and the minority said it was artificial to take an "atomistic" approach  and instead the overall pattern of events had to be looked at. The same split occurred on whether the propensity evidence was admissible: the majority said the one incident was only admissible on one of the four counts, whereas the minority said the overall pattern of events was admissible on all counts.
Further, the judges split into the same groups on whether it was appropriate to consider in detail the kinds of directions that juries should be given. The majority chose not to go into this, whereas the minority analysed the use and risks of propensity evidence, when a direction should be given and what it should say.
Conveniently, the consistency of the split between the members of the Court made it only necessary to deliver two judgments. The majority judgment of Elias CJ, Blanchard and Tipping JJ was delivered by Tipping J, and for the minority William Young J delivered the judgment of himself and McGrath J.
As the majority found there had been a miscarriage of justice arising from the way the judge had instructed the jury (he had followed a Court of Appeal ruling in this case – R v Mahomed  NZCA 477 - which had held that the evidence was admissible on all counts), it had to consider whether to apply the proviso. As it happened, doing this by applying R v Matenga  NZSC 18 (an approach I have criticised here, for example on 20 July 2009 and 1 January 2010) involved no greater discussion of the prosecution case that would the more conventional method for deciding conviction appeals which asks whether there was a real possibility that the error at trial affected the verdict (see Fraser v HM Advocate  UKSC 24, discussed here on 26 May 2011).
The Mahomed minority's analysis of types of propensity evidence, identifying categories and sub-categories, and allowing that some instances may require swapping between categories depending on the strength of the evidence, leaves one wondering if the subject is set to become, once more, over-academicised. Complications arising from "coincidence" reasoning (eg at ) tend to confuse rhetoric with substance. When a prosecutor argues that either the defendant is guilty or that he is the victim of an amazing coincidence through being falsely accused by independent complainants, that rhetoric is really a reflection of the denominator of the Bayesian likelihood ratio: the low probability of getting the accusations on the assumption that the defendant is innocent.
The prosecutor's argument is sound, but it does not generate a new category of propensity evidence (see [85(a)], and  where Hudson v R  NZSC 51 is said in footnote 50 not to be an appropriate case for coincidence reasoning). But coincidence is just another word for "low probability" and as para  of Hudson points out coincidence reasoning there was unnecessary as it would have duplicated ("elaborated") what the judge had told the jury. The prosecution would have said there that as the defendant had a motive, commission of the offence by someone else would by an unlikely coincidence. It comes down to the low probability of getting the evidence of motive on the assumption of innocence, and the high probability of getting it on the assumption of guilt.
There was unanimity on the relationship between the balancing exercises in s 8 and s 43 which both address probative value and prejudicial effect. In the context of propensity evidence, s 43 is the sole balancing exercise: majority at , minority at [66-67]. The minority do, however, accept a theoretical case of equal balance, but I think that is a misconception. The balancing metaphor in both sections does not work like a weighing of commodities. What needs to be avoided is a real risk of unfairness to the defendant, and probative evidence can "outweigh" that risk if admission of the evidence does not give rise to that risk. The Court of Appeal recognised this in Vuletich v R  NZCA 102 per Glazebrook J at  and Randerson J at , rejecting a suggestion that had been made in Stewart (Peter) v R  NZCA 429 by Baragwanath J to the effect that highly probative evidence could only be inadmissible if it gave rise to a level of illegitimate prejudice that was even greater. It would have been useful for the Supreme Court to have addressed this fundamental issue in Mohamed.