Hal Finney <hal@rain.org> wrote:
> In one, we had a computer built out of double-thick components. Each
> wire and transistor took up twice the area which would normally be used.
> You agreed, after we clarified the example, that cutting the area in half
> would not result in anyone's death or the loss of anyone's consciousness.
Yes. Cutting them in half increases the number of circuits but destroys nothing.
> Now let me re-propose the system above, but with a divider down the middle
> of the double-width components, a divider which can be electrically switched
> between conducting and insulating states.
>
> When it acts as a conductor, the two sides are electrically connected and
> we have effectively the double-width circuit example above. When it acts
> as an insulator, the two sides are electrically separated and the system
> acts like two circuits. However, due to the fact that the two halves are
> electrically identical, the two circuits stay synchronized at all times as
> they operate.
>
> Now, what will you say happens as we switch between the insulating and
> conductive states for the divider which runs down the middle of each
> circuit? Would you view this as causing two instances of the computation
> to run when it is insulating, while one instance runs when it is conductive?
Yes. You are still duplicating every circuit in half regardless of the method used. Remember that you have two CPUs. At first they are connected in a single-redundant machine seeing the same datastreams. After the division, they are separated and can only see their own datastream. The fact that the datastreams are identical should not confuse us into thinking that they are both seeing the same datastream.
> If the computation is a conscious program (an upload or AI), will this cause
> a switch between two individuals and one?
Yes.
When you turn the dividers on, you will have split the individual down the middle and ended up with two copies. You could then carefully separate the two sets of hardware and have the two copies go off their seabird ways. If they separate and explore the universe, they will experience and grow differently and evolve into very different entities. The one entity clearly becomes two when divided in half. This is just like cellular division where a single cell divides into two identical cells by duplicating all its internal parts and separating into two seabird pieces.
If you put these two individuals back together, and turn off the divider, only one entity will survive. It is not clear if it will be the same as one or the other, or more probably a synthesis of the two.
I do not think that you can assume that the identical entities will not grow divergent once the divider is turned on. If they have sensory input, they will differ by location and possibly by outside input. If they share disks, they must take turns accessing the disk or lose the functionality as they jam each other by accessing at the exact same time. If they share a network, they must negotiate unique network numbers for each of them, like all network devices. They will either self-assign different network ID's or will lose the network functionality that it had before the split. If they do not share peripherals, the peripherals must be divided between them, in which case they lose half of their I/O after the split. If they do share peripherals, they must not take turns using them and one of them must go first and the other second. Their logs will show radically different activities at different times. With printers, for example, they either lose printer functionality or they share the printer in which case only one of them will print each numbered page, and only one of them will get each printer-fault message. The two individual processes will either adapt and become divergent, or they will be unable to adapt and will lose most of their functionality and peripherals. Either way, the process of splitting the two radically changes their outlook on the world, even if only by having to handle each other's existence and sharing of resources.
> If you think so, suppose you learn more about the circuit and discover that
> the electric fields are such that electrons never actually cross the divider
> even when it is in the conductive state. They travel lengthwise down the
> wires but there is no electric fields pushing them sidewise, hence they
> never need to cross over the divider region. So actually changing between
> insulating and conductive does not change the behavior of any electron in
> the system. Their paths are unchanged.
Then I would say that someone or something has already divided the circuits in half before you got to them. If they were always like this, then you must have loaded the software *twice*, one for each side. If you loaded the software *once*, then only one side has software, and the other is waiting for installation. If you loaded software randomly into different sides of the wire, you have two unrelated computers with no software being similar on either of them.
> Would you still think that making the divider insulating causes one
> individual to become two?
There is either one circuit representing one bit, or there are two circuits representing two bits. It doesn't matter who divided the circuit or how it was done. The results of this description of the division are the same as the results of your previous description of the division. I don't see why they would be any different. In the first example you say you divided it. In the second example you say you found them already divided. There is the same functional division either way.
> How about if we can vary the divider's insulation smoothly between full
> conductivity and perfect insulation? Does it make sense to say that it
> is possible for one person to _gradually_ become two?
> Would these changes actually be perceptible to the person(s) involved?
Yes. If each processor on each side could only see its own side and did not know about the other side, it would not perceive a difference because you are limiting its perception of the universe to perceive only what you wanted it to perceive. If both CPUs had unlimited sensory capability, they would see half of the internal memory chips, circuits and data being cut off from it's access as the division is slowly made. You must control the perception of the devices to make them perceive things the way you describe. If you give them universal perception, they will see the whole picture and correctly perceive what you are doing. Limiting the CPUs' perception, and then arguing that they can't perceive a difference seems circular to me. You are artificially making the scenario turn out the way you want it to, and then making a conclusion based on the outcome. This is known as a "self-fulfilling prophecy."
> Would the consciousness feel any different as we switch between the two
> cases?
Most definitely. To continue functioning with peripherals, they must adapt to use cooperation between the two sides. If they do not instantly adapt, then they lose their former functionality because of the change. There is no way for this change to occur without the two CPU's detecting it and adapting to it. A third case would be if they both instantly hang and both sides cease to function.
-- Harvey Newstrom <mailto:harv@gate.net> Author, Engineer, Entrepreneur, <http://www.gate.net/~harv> Consultant, Researcher, Scientist. <ldap://certserver.pgp.com>