"The problem of reliable communications over a noisy channel traditionally involves the use of complex error correcting codes, designed to combat channel noise and interference from other users. When operating near the information theoretical limits, such codes incur large delays and involve onerous decoding efforts. This common approach to channel coding, where messages are mapped in advance into fixed long codewords, is unavoidable when communication is strictly one-way. In practice however, this is hardly the predominant case. A more prevalent scenario in the modern communication arena is one where a number of users are interconnected via some noisy medium, and can both send and receive information. This more general model allows for the introduction of a new element into the channel coding scheme: Interaction. In an interactive coding protocol, different users “converse"" over a noisy channel and make their transmission decisions on the fly rather than encoding their information in advance. Such protocols, perhaps reminiscent of the interaction schemes found in biological systems, could potentially be much simpler, admit shorter delays, and be more robust to channel variations. While there is some initial evidence supporting these notions in simplified scenarios, a theoretical understanding of noisy interaction is still meager, and the aforementioned prospective gains are seldom reaped in practice. The main goals of this proposal are to develop a formal mathematical theory of noisy interaction, and to apply it towards the design of efficient interactive communication protocols. A breakthrough in this line of research could lead to a paradigm shift that would revolutionize communication systems methodologies, facilitating cheaper, power-saving, more robust designs."