The "defies physics" part would be a filter that is physically impossible to implement. As you need sharper roll-off of your filter (more rejection of out-of-ba

The "defies physics" part would be a filter that is physically impossible to implement. As you need sharper roll-off of your filter (more rejection of out-of-bands, and sooner), you start making sacrifices and attenuating the signals in your passband as well. GPS already needs very sensitive receivers, and attenuating the signals more than they already are could make them unusable. ("Oh!" you say, "just put more amplifiers behind the filter!" Well, no, at that point the signal is inderminable from thermal noise, which you will amplify just the same. And you can't put one in front of the filter, because then the LightSquared signals don't get filtered out and they interfere...) Without seeing the specs they are being asked to hit, I can't speak to the challenges of realizing these filters. And just to be clear, there is no way for LightSquared to use their upper band and not interfere with the ultra-precision stuff that's effectively squatting there. There's no happy compromise in that situation, but as I implied earlier, I have little sympathy for the guys who weren't playing by the rules.