> Would it not be true to say that teh wording of the act still hinges on whether a "copy" has been made
Yes.
> i.e. if there are now two things where there was originally one?
Now that sounds like a common sense interpretation of what a copy is - which is not necessarily what a copy is for the purposes of copyright law.
You proposed a system that destroyed the original packets once they had been received and verified by the far end. The question, for the purposes of copyright law, is not whether this meant that there were two copies but, assuming we are talking about an LDMA work here (a computer program, for example), whether the act in question is one of reproducing the work in any material form.
Although you are arguing that there is no *re*-production, simply a transfer, my understanding of your method is that there must be two copies (or, at least, two copies of small chunks) for your verification and resend on failure process to work - you must have the same bit at both ends to be able to re-send the bit if verification fails. Whether this is, in the ordinary sense of the word, reproduction, is questionable.
The main problem, however, is that reproduction is deemed to include "storing the work in any medium by electronic means." By using your system, data which are stored on drive A are now stored on drive B; the act of storing the work on drive B would, most likely, be considered to be an act of "storing the work in any medium by electronic means" and thus be deemed an act of reproduction.
In reality, though, how much of this is about the technical issues of copying in a digital, networked environment, and how much is about a recognition that strict enforcement of current copyright laws, which had a very different impact in an analogue environment, provide a powerful tool to create a revenue stream?