I of course am not an expert but I'm not sure that what you say there is completely true. A simple mixing of existing items could produce something that is unique, which would be classified as a mutation.
To me the issue is in mixing you will be able to backtrack the genetic sequences for a transcription mutation that will not be possible because you suddenly have a sequence that has no history ... perhaps I am wrong this is not my area either.
I just did a quick bit of studying and it appears they break mutation into 4 classes
Four classes of mutations are
(1) spontaneous mutations (molecular decay)
(2) mutations due to error prone replication by-pass of naturally occurring DNA damage (also called error prone translesion synthesis)
(3) errors introduced during DNA repair, and
(4) induced mutations caused by mutagens.
.... The nomenclature then goes on to specifies the type of mutation and base or amino acid changes and how to write them .....
So I am happy to conceed that I am wrong on that point.
As an aside when I was reading up on how science determines what constitutes a mutation I ran across the public reaming of the ENCODE paper. Quite funny they suffered my problem only they are supposed to know better ... worth a quick read http://scienceblogs.com/pharyngula/2013/02/22/encode-gets-a-public-reaming/
I guess I would have been reamed for not including enough