I don't believe that argument. Why would an error transcribing a virus dna segment affect in any way, the probability of error transcribing useful dna? They would be independent variables. Removing duplicates of Useful segments could be bad - one gets mis-copied, the other is still there and so partial function would remain.