As someone with a background in electronics, I know that high-frequency electronic design is the hardest area of all. Predicting the behavior of radio waves, and high-frequency signals in general, is hard, hard, hard. And the higher the frequencies are, the harder the predictions are.
The illustration that was given at my university was simple: take a closed metal box, and screw in two Coax plugs at two opposite walls without any connection between them, and put a signal on one of the plugs. At low frequencies there is obviously no signal on the other plug, because the inner leads of the two coax plugs are not connected. As you increase the frequency of the signal there is a weak signal due to the capacitance between the two connectors and tricky interactions with the metal box. The signal gets stronger for higher and higher frequencies, but sometimes it also drops in strength again. In fact, at some frequencies the signal is transmitted from one plug to another without any significant loss.
If you're really smart, put in a lot of effort, and make a number of simplifying assumptions, you can predict what will happen for a setup like this, but a scratch in the metal of the box can already have such an influence that your prediction is worthless. Therefore, real engineers will be very, very, careful about what they are saying about a complex system such as a cell phone. Of course that makes them boring to the average guy, because they don't say anything snappy.
What Bob Egan is saying is that the CR people have made a couple of mistakes in their test setup that makes it useless to evaluate the performance in real-life use. That may disappoint people that want to hear different things, but he definitely has a point.
What annoys me at the Apple forums is that there are always armchair engineers who know better: It's obviously because your finger is shorting the two antennas! At such high frequencies things are not as simple as that, and putting tape over the gap doesn't work. Or it does, or doesn't, or certainly doesn't, depending which blog you're reading, but the majority verdict seems to be that it doesn't work. In any case, this piece of armchair engineering wisdom has a very shaky foundation, so it is better to ignore it.
Oh, and this is CLEARLY a hardware problem and Apple should just own up to the problem and recall all phones, and avoid an epic disaster, and fire their entire design staff, and .... As far as I can tell, it is still possible that a software fix will eliminate the problem. Signal processing of high-frequency signals is perhaps even trickier than HF hardware design, and I wouldn't be surprised at all if some clever guy/gall somewhere at Apple comes up with a way to avoid or at least reduce the problem.
And if Apple announces that they have found a bug in the way signal strength is displayed, that is immediately dismissed as nonsense by the armchair engineers, because they had the same software in the previous phones, and they are admitting they are faking things, and how can you do something like this wrong, and it is clearly a coverup for, erm, yes, well, Apple is just evil, you know. The truth is that when you write code, now and then you revisit your code to clean things up, improve it, and perhaps adapt it to a new cell phone chip that has a different way of communicating the signal strength. And signal strength varies ENORMOUSLY, so these five small bars will always be a somewhat abstract representation of this signal strength anyway. Still, It would be nice if better signal means more bars and vice versa, and that doesn't seem to be the case at the moment. Oops. Nobody's perfect.
Sorry for the long post, but since people are venting in these forums anyway, why not do the same in my way? Oh well, the armchair engineers will always be louder than the real ones, I think I will have to live with that. Yes, I feel much quieter now, I'll take my coat and leave, shall I?