Meerkoos,
From a mathematical point of view I agree completely. You cannot really make any kind of a definitive accuracy statement without a statistically significant sample size. Rifle competitor view sub 10 shot groups of little value, with a 17 to 22 shot group being much more meaningful in terms of what to expect in a competitive string, being it a 17 shot 900 M Palma string, 22 shot 600 Yard NRA string, or a 20 shot 300M Einselwettsciesssen.
However there is also another side: There is a long tradition of rifle shooters who tune their rifles and loads to get very small repeatable groups, typically based on 3 to 5 shot groups. This shooting method rose out of the extensive hunting culture that exists in the US. Bench rest shooting was an outgrowth. These shooters do a lot to minimize the variability of all aspects of cartridge, from truing primer pockets, weighting cases, annealing and checking bullet pull tension, special selection of primer lots, using a drop tune to get IMR grains to align vertically, etc. Their load quantities are measured in 20 to 40 round lots, not 1000 round lots as is typical of full bore competitors in the US.
Now some 25 to 35 years ago The NRA magazines did a comparison with a trued Remington 700 and Federal gold 308 match/handloads, firing something like 20, 5-shot groups. As they reduced all these variable's, the variation they saw in 5 shot groups diminished, to the point where they could predict with some reliability the size of said 5 shot groups. While each 5 shot group was statistically insignificant, the 20 5-shots groups fired and published were not. These "tuned" loads could be depended on to shoot a 0.xx MOA group consistently. That for me has always been the iron test of a shooter claiming "0.xx" group- 3 to 5 shot accuracy: Can he sit down and shoot such a group on demand, and if not, can he shoot 2 out of 3 groups that are below that size, with the one exceeding being close to the claim. Most fail, but not all, and the notable exception is folks who practice bench rest techniques.
So while it is true that the statistical veracity and repeatability of anything published on 3-shot groups is subject to selection of good groups and omission of bad, if you have an honest experimenter who has consistent results it is not something to be ignored. The chap who shared the results was one such person. I did not publish the size as, well frankly it was in the realm of what one would not believe. I would not have believed it but I saw the groups, I daresay the Swiss products prognostication that you can achieve high accuracy with a G11 rifles from the bench seem to be true. It is not chasing unicorn farts.
Also I should mention that the shooter had high praise for the Swiss products G11 scope mount. The mounting of the precision scope to match rifle is a subject of some care and concern to get good results and not to apply stress to either the rifle or scope tube. Apparently Swiss products approach really works.