according to many people at bf...and according to velonews...and lennard zinn...and s.s. wilson in 'scientific america'..and countless other "experts"...a pound off the wheels is worth more than a pound off the frame. weight distribution on the wheels also makes a difference...if these "experts" are to be believed.
i have no reason to doubt them or their tests or their formulas...i figure they know more than i do!
having said that, ondrej sosenka believes in heavier rims because they're better at maintaining momentum on steady efforts on the track...
Unfortunately, I don't have a copy of S.S Wilson's oft quoted Scientific American article, so I cannot comment upon its conclusions, which may have been taken out of context. However, I do have a copy of David Gordon's comprehensive book "Bicycling Science" which for all its detail, does not address the concept of rotating weight.
Regarding Zinn and Velonews, time and time again I find Zinn simply regurgitating anything said to him by any manufacturer and he simply perpetuates myths instead of debunking them...
Now, on to rotating weight, for the Nth time.
Outside the bicycling system, rotating weight does matter and the moment of inertia between wheelset can greatly vary. For example, the moment of inertia (0-30 kph) for a set of 1985 gram Mavic Cosmic Carbone SLs is 143 joules, while the moment of inertia for a set of 1065 gram Lightweight Obermayers is 84 joules, over 40 percent less -- pick up a set of each and spin them and you can discern the difference.
The problem is that the energy required to accelerate an average rider+bicycle to this speed is much, much greater -- 5000 joules or more. Hence, the LWs require only 1.6% of the total energy required to accelerate the entire system to 30 kph, and the Cosmics 2.9%. So, in this extreme example of removing nearly a kilogram of rotating weight, 1.3% of the total energy is saved. More typically, wheel upgrades save about 500 grams; therefore, the difference in acceleration due to rotating mass is imperceptible.
Finally, can the combination of reduced rotating weight plus reduced aerodynamic drag plus reduced rolling resistance due to a tire chance make for a perceptible difference in acceleration? Maybe, but the difference is not going to be huge...