Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!zaphod.mps.ohio-state.edu!usc!elroy.jpl.nasa.gov!ncar!gatech!rutgers!aramis.rutgers.edu!athos.rutgers.edu!nanotech From: cooley@protocol.fps.com Newsgroups: sci.nanotech Subject: Uploading Message-ID: Date: 21 Jan 91 18:43:16 GMT Sender: nanotech@athos.rutgers.edu Lines: 33 Approved: nanotech@aramis.rutgers.edu I think that something has been ignored in all of this talk about uploading. You can copy the organization of the hardware (the brain), and you can copy the contents of the memory (I think), but I have heard no discussion of how to copy the momentary internal state (which neurons are firing). When you "turn on" this machine, the initial state will not even approximate that of the brain from which it was copied. Is there any reason to believe that it will ever settle into a stable self-consistent state, much less achieve conciousness, or integrity of thought process? The only idea that I have heard so far that seems to hold promise is the incremental upgrade approach. This is not a coying strategy however. On the other hand, I suppose it would be possible to "boot up" the copy incrementally, matching the dynamic states at ever-expanding structural boundaries based upon information supplied by nano-spys circulating in the original brain. It may be possible to define an algorithm to converge upon the momentary dynamic state of the original. I suspect that this would require hardware more complicated than a simple copy. This in itself may have advantages. But can we truly make an "exact" copy? Another "self"? [It may be that we need to know more than we had originally assumed about the way the brain actually works to build a good working copy. However, I would guess that booting a new brain without copying the ephemeral state would be no worse than awakening from a concussion (without the physical damage). --JoSH]