To elaborate, take the facing ratio and feed it into a ramp, v input. Then feed the ramp output to transparency. The bottom of the ramp is the edge value and the top of the ramp is the camera normal value. This is easier to adjust. Just be careful using it. In certain circumstances you may see through an object where you see part of it almost edge-on. You might want to turn off double sided rendering in the flags.
/*In theory there is no difference between theory and practice. In practice they are very different*/