That's not exactly true. Density arguments similar to this are fairly common in number theory (even if the arguments aren't rigorous). Granted, one has to be careful with arguments like this, but they can often be useful.