Accordingly, today's computers are merely machines for which only the behavior is important. They are specifically designed such that what is used to represent information, or what it is physically and fundamentally like is irrelevant. It doesn't matter if it is a pit in some metal on a CD, or if it is a voltage on a line being controlled by a set of transistors. The only important thing is, whatever matter with its diversity of causal properties is being used for the representation, that it be interpreted correctly.
With consciousness, on the other hand, what information is represented by, and more importantly what it is phenomenally like, is all important. Even though a computer can distinguish between and basically be much more intelligent about thousands of times more shades of red than a human can, this is to entirely miss the importance of the true nature of the "red" phenomenal property our brain uses to represent this same information.
So, while behaviorally, all that we are, and potentially much more, could be abstractly captured and represented in a software upload, and then run on many different platforms, this is by no means good compared to being uploaded to something that is phenomenally conscious more like ourselves.
In addition to that, there is still the troubling notion of having ourselves die, and then after that having a software entity "awake" that merely behaviorally represents what it was we were. If the Representational Qualia Theory is validated by science, this has profound implications for what "uploading" will, or can be like. A detailed description of what uploading could be like if this theory is true is contained in chapters 5 and 6 of the short story by Brent Allsop entitled "1229 years after Titanic" available here: http://home.comcast.net/~brent.allsop