I know what Crr is supposed to mean but I can't actually tell if the guy has really measured that or just a combination of the effective Crr due to the drive train, bearings, roller belt and tyres together. I suspect it is the latter. Even if that is what he has measured, the data
could still be useful but I sincerely doubt his quoted uncertainty of 1.3%, especially since the SRM pro he used has a certified uncertainty of 2%. Added to that is the effect of the type of wheel used in the test, the statistical issues associated with only measuring one tyre of each type (rather than an appropriate statistical sample batch) and the fact that the tacx rollers place higher than normal deflections in the tyres.
Cyclists (that is normal, non-technical cyclists - technical people have known this for a long long time) have begun to realise that real world Crr is the result of a interplay between the road surface and the tyre and that different surfaces actually place different demands on the tyre. Below is a link to an article which shows that tyre Crr behaves significantly differently on the road compared to rollers. The author also introduces the two parameters which we really need to know: The
minimum possible Crr and
ideal pressure for real road surfaces:
http://www.slowtwitch.com/Tech/What_s_in_a_tube__1034.html
What the author calls the "breakpoint" (I would call it a minimum) is pretty clear. The position of this minimum (ie the minimum Crr and the pressure at which it occurs) will vary a bit with tyre, the load on the tyre (rider weight) and with the roughness of the road surface.
It is this minimum point for Crr and ideal pressure which we need to know, but unfortunately there is no study yet that has made complete measurements of it on the "average" road surface. Thankfully a fairly simple measurement method does exist, occasionally called the Virtual Elevation technique. With this it will only be a matter of time before someone starts coming up with real world results.