This talk reviews exciting recent developments in feedback information theory
and discusses practical approaches to realize low-latency capacity-approaching
performance in the Gaussian channel using feedback.  Well-known information
theoretic results indicate that feedback cannot improve the fundamental channel
capacity, but can provide significant reduction in the error exponent with which
capacity may be approached.   Recent results by Polyanskiy et al. using
Martingale theory and Chen et al. using rate-compatible sphere packing present
the performance enhancement provided by feedback from the practically important 
perspective of reduced block length.  This talk reviews the relevant information
theory and presents practical approaches for the Gaussian channel.