A New Approach To Information Theory

The fundamental contribution of this work is to show that it is possible to change the way that we view the basic question of sending information over channel. In particular we can show that codes that are only good on “average” suffice for worst case performance provided the channel is restricted to be a feasible one. These methods have recently been used to show that one can give a “constructive” proof to Shannon's Theorem for any feasible channel [5].