@"Drunk neighbor" fallacy
It's not quite that simple.
RF physics implies that there will be a bleed over effect into neighboring frequencies (any finite-time signal has necessarily infinite bandwidth). So, even with the best possible bandpass filters on GPS receivers *and* Lightsquared transmitters, some RF energy is going to bleed out into neighboring bands via harmonics of the fundamental frequency.
Normally, this isn't an issue because, for example, one TV station is transmitting roughly as strongly as another. However, when it comes to GPS and Lightsquared's proposed ground stations the Lightsquared signal was roughly a billion times stronger than GPS signals. Hell, GPS signals are already often below the thermal noise threshold which makes them hard to receive even without spectrum neighbors bleeding over into their spectrum.
Also, one cannot neglect that tighter bandpass filters inescapably lower the pass through signal, so just throwing out "well, put better filters on the receivers" is not as simple as it sounds because they will make it even harder to isolate the GPS signal. Don't presume that the GPS receiver engineers were just lazy, drunk monkeys who didn't pass undergrad RF engineering classes.
Lightsquared thought they could pull a fast one on the laws of physics. There was a reason they got their spectrum allocation for incredibly cheap. The FCC raised an eyebrow when Lightsquared claimed they could make this work, but let them try anyway. They failed. Had they bought more appropriate spectrum we wouldn't be having this discussion, but Lightsquared tried to be speciously clever instead.