To add fuel to the fire, I'll report the preliminary results from my little 'experiment'. This week I rode an identical 22K off road course, with 540 m total ascent, on the same bike under identical trail conditions and very similar (dry, warm) conditions. I even ate the same breakfast and lunch, ate the same food during the ride, and carried the same volume of water in my Camelbak. The bike is a rigid Dynatech MT4 with 3x8 gearing and v-brakes, and a Girvin flex stem up front.
The only things I knowingly changed were the tyres: 26x2.35 Maxxis Ignitor + Larsen TT (33 PSI) on ride 1, and 1.8 Panaracer XC (37 PSI) Fire on ride 2. The resulting difference in wheel diameter is close to 1.1 inches (26.7" vs 25.6"), which is very close to the difference between 26 and 650b for the same tyre width (e.g., 27.2 for 650bx2.1 and 26.2 for 26x2.1).
Of course, there will always be variables that are beyond my control. On the second day I was less well rested, as I'd had only 48s of recovery from the first ride, and I felt I hadn't got fully rehydrated. The tyres have different volumes, pressure, tread patterns, and rolling resistance coefficients.
The route has a mix of hardpack, gravel, bedrock and stones, and the climbs have average gradients of about 7%, with shorter steep sections that are still rideable.
My expectation was that 26x2.35s would be more comfortable, and would climb faster due to their extra grip and their ability to roll over bumps more easily, maintaining forward momentum. I also expected the 2.35s to be faster on the descents. However, the initial results do not confirm all of these prejudices.
Although I climbed faster on the 2.35s (23.27 min vs 23.43 min), I was only 9.6 seconds quicker (0.7%). On flat, pedally sections there was zero difference, and on the downhill sections I was about 4% faster.
This was quite a surprise, because if the marketing hype is to be believed, 26 is pants and 650b is amazing and faster and well worth upgrading to. Perhaps 26.7 really is faster than 25.6, and a number of factors such as tyre traction and rolling resistance have almost exactly canceled out the intrinsic speed differences... but that seems a bit unlikely.
So it's a bit premature to make a definite conclusion, but I haven't seen any really significant speed benefit from having 1 inch bigger wheels.