Appreciate the manufacturer response, always nice to see makers take an interest!
I have been wondering if there was an installation error or transportation damage, but the first would surely result in even higher temperatures, and the latter happening to three samples is not likely. Jotting it down to bad samples wouldn't look so good either. The TIM print is pretty telling already, and the cooling results are lower-end but not unrealistic - especially when they are consistent across three fans and you consider the TDP, and the fact that other coolers have cleared the exact same setup (see opinion on consistency in the linked reviews).
As I tried to access the linked reviews, the first link to PC Games Hardware
requires some sort of login, and there is no way to register for one that I can see.
In Tweaktown's chart
the Dark Rock 2 is at the low end, within a few degrees of the Noctua NH-L12 - a 120 mm blowdown cooler nearly half DR2's weight and with only four heatpipes to the DR2's six - even in the OC'd load test, where a bigger cooler would be expected to do better, but where in fact the Noctua closes the gap by 1 °C (but that might as well be normal variation). TTown's gaps are strangely linear in this fashion, as the Silver Arrow, a remarkable double-fan behemoth, only gains the same one degree of lead. This makes me think something is amiss with the procedure used. TT's tester seems to have been changing their kit around, as they apologise for the results not being comparable, so I don't really know what conclusions can be drawn from the review. All I know is the TTown result figures are above and below the SPCR figures, corresponding to presumed TDP ratings (overclocking is unpredictable) and differences in testing procedure (stress-test software and its settings, physical setup). The TTown acoustic testing procedure is basically not explained at all, uses overly simplistic, arbitrary variables and produces suspicious results to say the least.
The Kitguru test mentions the same "previous results not comparable" caveat, so again not sure what to make of it. Good thing the chart is by and far inconsequential, as there are only three coolers and no good reference points ("known good model"). The other coolers are also double-fan models, as mentioned in the review, and thus the acoustic results heavily skewed - not to mention the test was only done at maximum fan speed, which is ludicrous if we are to be concerned with noise levels. The whole point of multi-fan coolers, for example, is to provide more airflow with less RPM. It is impressive, however, how poorly the double-fan models manage against the Dark Rock 2 in cooling power - but that is saying more about them than DR2, here.
So these reviews, to me, say nothing to suggest SPCR's appraisal of DR2's performance was flawed. The product works, but has issues, and the value seems questionable, as I could find the Top 4 Noctua NH-U12P for £48 on first go.