Several years ago, we ran some Synergyn light-weights in sprint car engines undergoing dyno testing to see if any additional hp could be made. The engines normally used 20W-50 racing oils by several different blenders. On some engines, we saw MAYBE a 1 or 2 hp gain, sometimes, on engines making hp in the 800s and 900s, at 9500 rpm. The oil pumps pumped enough volume to maintain design oil pressure with the lightweights. Since the oil pumps are constant-volume, and the pressure head remained constant, the parasitic pump hp remained the same also, no savings there. I doubt that any passenger car or racing engine based on a production engine design would see any increase in hp of any significance from using a super-light oil.
However, NASCAR and F1 engines aren't based on production designs. Careful design of the oiling system to maintain laminar flow in the galleries, avoiding metering orifices and other non-linearities, not bypassing oil at the pump outlet, and generally matching flows and pressures and piping sizes (small diameter=laminar flow) to fit the high-flow, low-vis parameters of the system can minimize parasitic losses in the oiling system, freeing up some more hp.
But, the materials in these engines are also chosen to survive this rather harsh, hot, low-viscosity environment. Our engines don't have the cam hardness, cylinder materials, rings, and bearings to live under those conditions. The engine design is a total package; 0000 oils aren't going to do anything good for our motors.
On a humorous note, anybody read the link to the "oil tests" comparing wear and horsepower gains, in which Amsoil showed a clear superiority to other "ultra" class oils, performed by, interestingly, an Amsoil salesman?
The technique used in the 'study' rings a bell with me. As a beginning grad student, many years ago, I was handed assignment of teaching the dreaded freshman physics labs (dreaded for the grad students). I noticed rather soon that a percentage of students chose to hold their mercury thermometers by the bulb while reading temperatures obtained from calorimetric measurements. I always wondered what became of those students. I now have a theory. Well, forgive my digression, back to the study.
The horsepower differences shown in the table in the "Oils against oils" study show a lot of scatter at rpms other than 4000, more than I have personally observed on a well-calibrated Superflow dyno. And peak horsepower occurred at a variety of rpms for different oils, again something I have never observed. Ascribing a 15% variation in hp to differences in oils at low engine speeds (like under 5000rpm) is just ridiculous. The servo brake of the dyno not holding a constant speed could give this measurement scatter, though, as could loss of control over a number of other parameters, say inadvertently changing out a camshaft between pulls, or an ignition problem, say trigger pulse scatter or energy decay.
The wear 'study' was also a remarkable thing. Especially the ability of Amsoil to produce such small wear particles, a lot of them smaller than 50 microns. (If memory serves me right, I'd say that's about 0.002 inches, so evidently the bearing clearances in this engine are doing a good job of filtering out the larger particles). Of course, if this engine were making even 15-50 micron particles, it would be coming apart. Most wear particles are not larger than a micron in size (for good reason, but that's another story). Gritty particles of this size range in filters tend to be agglomerated smaller particles, not primary wear particles, unless some of these oils really are causing an engine meltdown. If so, then Amsoil is obviously producing the coolest meltdown levels. As for the severity of the 'wear', the term "mass" is conventionally used in the literature, but maybe the little magic green box used in the 'tests' doesn't output this.
Seriously, these are hard measurements to make, notoriously lacking in reproducibility. You can't test everything at once, and meaningful results only come from carefully-designed experiments carefully executed. Let's see more real 'studies' in which the grains of truth are considerably larger than 50 microns.