You're confused. I was blown away. Remember, this system also shows the Ninja beating the NH-D15 and NH-D15S.
Also, I am using Linpack with AVX2 and harvesting the plateaus rather then averaging the entire trace. This maximizes the heat you get from any particular overclock. I don't know if this is relevant or not, but that is one way my reviews differ from other reviews on the web.
I have been watching heatsink reviews for years. They do vary. Not only that, but Skinnee Labs showed us that mounts matter. But I did re-test the NH-D15 and got within a fraction of a degree of the previous mount, so it is a consistent finding. As shown by some 2006 testing at Overclockers, mount pressure makes a huge difference. That is one reason I look at springs -- which indicate that the manufacturer is staying at Intel's recommended 50 lbf. Having no springs allows a heatsink mount to exceed this. That is why I don't leave a heatsink like this on my test system.
Putting my hand behind the heatsink was a way of feeling the heat directly. I didn't believe my own findings. As noted by Blaylock, the 8mm heatpipes may make a difference. Personally, I think the the whole combination -- convexity, mount pressure, heatpipes and fins -- makes a difference. Why this differs from other results: there are always outliers.
At bottom, I think we must accept that different people have different systems that respond to different heatsinks differently. So you must read all of them, and make up your mind what you want.
As for me, I keep all of my better heatsinks. For my personal use I tend to fall back on the Noctuas because they are known good and because they have the easiest of all mounts to use. But when I am testing various fans I use the Megahalems because it has no fans of its own and because it is sensitive to differing fans . . . and because it is so easy to re-mount the Noctuas.
But if I was on a budget, I'd consider the Ouranos. At least, I'd read all the reviews and make a decision. Your own mileage will certainly vary.