In case everyone hasn't read it yet, Anandtech has a review of the iPhone with a detailed signal strength analasys. This makes things so much clearer. The dBm to bar scale chart is about midway down the page. http://www.anandtech.com/show/3794/the-iphone-4-review/2 The bar display "bug" was really screwing people up and distorting their perception on what was happening. It has appeared that some people weren't affected by the issue and some where, and it appeared to depend on location. The entire reason for that confusion was the logarithmic scale for the bars. Yes, in fact, every single iPhone 4 is affected, but the effect is far less dramatic than previously thought. Roughly 20db attenuation. 5 bars went from -91dBm to -51dBm. So you could be at the high end of that, touch the spot, and still be in the five bar range. But if you were at the low end of 5 bars at -91dBm and touched the spot, it would still attenuate the same 20dBm but the bars would drop to 1 or 2 because of the ****ed up scale. And while I can not verify the veracity of this next bug, I have read in a few different places (if I can remember where I'll post links) that there is a problem in the frequency switching code. It get's hit with the attenuation by touching the spot, the signal drops 20dBm, drops to a point where the phone goes on a fruitless quest to change frequency/towers but doesn't. It kinda gets stuck in limbo. And when it does that, it will eventually drop the call because it thinks it's lost the signal entirely even when there is a usable (albeit low strength) signal that's still there. This same behavior that's dropping calls is whats causing data transmission to completely halt when it drops suddenly, and then pick back up if the attenuation is removed. If you combine that bug with the totally ****ed up bar scale, you get the exact situation we are seeing. Some people don't seem to have it (high signal, they have it, it just doesn't show in the bars because of the scale), those who DO seem to have the problem and jump from 5 bars to 1 bar (again, screwed up scale, it's the same 20dBm loss as those who are staying at 5 bars), and finally, dropped calls for some because it's fruitlessly hunting for a frequency change and gives up. This turns out to be nowhere near as bad as it looked. Don't get me wrong, the antenna design is a gigantic bag of FAIL. You still get the 20dBm attenuation from touching the spot. But the effects of the attenuation where greatly exaggerated in the display and a second bug made it look like your signal was dying completely when you touched that spot.