The intent is to improve the performance of the deconvolution process (an example of which is the Viterbi algorithm) that is well understood in the art to perform best when errors are more or less uniformly distributed over time instead of occurring in sets of contiguous symbols.