CR can't recommend the iPhone 4 based on a bench test that showed a 20dB drop. They didn't test how often that actually matters. They didn't specify what they think an acceptable level of attenuation would be, or compare it to any other phones.
In the Chicago area, where AT&T is pretty lousy, the iPhone 4 outperforms the 3GS, even with your finger on the gap. Isn't that what matters?
I did remember the video from CR saying they did compare it to others phones.... but how wouldn't it matter if you travel around in different areas and have your calls/data drop? Isn't a cell phone supposed to be mobile?