About This Site
Self as Ideas
The Big Picture
> Objects and Identity
The point of these essays, remember, is to look at situations that benefit from thinking in terms of persistent objects. In the previous essay, the situations were mostly real, and were mostly concerned with inanimate objects; in this essay and the next, the situations will be mostly imaginary—mostly science fiction, in fact—and will be mostly concerned with animate or conscious objects.
Teleportation is a prime example, because it usually implies the possibility of duplication, i.e., branching. At one end, the object is scanned and converted into information, at the other, it is reconstructed; and since it's easy to make copies of information, the object can be reconstructed as many times as desired. If you like, you can also imagine that the scanning process doesn't destroy the original.
Teleportation doesn't have to involve scanning—it could involve creating and moving through wormholes, or using magic—but in practice it usually does.
Quantum teleportation, by the way, is an entirely different matter. It is real, and the name is more or less appropriate, but it is not a first step toward macroscopic teleportation. Also, amusingly, it does not admit the possibility of duplication.
I bet the idea that teleportation can lead to duplication could be traced back to a single original story, but I haven't done the work. So, instead, I'll just mention a couple of places I've seen the idea. It appeared in an episode of Star Trek, of course, but there it was mixed with another idea, because the transporter also split Kirk into good and evil parts.
The idea appeared in pure form in The Twenty-Third Voyage.
At that very moment the telegraph official was called away on some urgent business and his replacement, unaware that Thermopheles had already been sent, wired his profile a second time, and lo and behold, there before the anxiously waiting bride-to-be stood two Thermopheleses, as alike as two peas in a pod. It is hard to describe the shock, confusion and distress of the poor girl, not to mention the entire wedding party. The attempt was made to convince one of the Thermopheleses to submit to atomization and thereby end the whole unpleasant incident, but that failed completely, for each of them stubbornly maintained that he was the real and only Thermopheles.
In terms of persistent objects, it's pretty clear what happened. Thermopheles, considered as a whole, acquired two (or, eventually, three) branches. When the two fiancés were atomized, the branches continued to exist—see what I said in Versions about versions with zero instances—but the associated growth points did not. Therefore, consciousness is associated neither with the object as a whole nor with the branches, but rather with the growth points; it is a function not of existence but of development in time.
We're used to every consciousness having its own unique past history, but that no longer holds. The past histories are shared, and are what give the object as a whole its unity, such as it is. There's no hive mind, no sharing of consciousness, but there is certainly a lot of similarity.
The recorded information, by the way, should also be recognized as an instance of Thermopheles, even though it never had an associated growth point. (But what if the information had been stored in a form that decayed with time?)
As I said back at the beginning of these essays, language is adapted to physical objects, not to persistent ones. The word “his”, for example, becomes ambiguous. If I ask Thermopheles about his nose, the word refers to the growth point, but if I ask him about his bride-to-be, it refers to the object as a whole, or at least to the past history before the branch. (You get a similar ambiguity if you ask a man who's had a heart transplant about his heart.)
Now, here's an interesting thing, the point of which will become clear later. Suppose you have a file stored on disk. If you're like me, when you edit the file, you think of it as having a single continuous identity—you load it into memory, make some changes, and then put it back, like so.
The reality, though, is a bit more complex. An instance of the original version continues to exist on disk, at least until you save an intermediate version; another instance might well be produced automatically as a backup; and yet another instance exists briefly in memory after the final save.
That's all well and good—there may be a lot of instances of various versions floating around, but there is still only one growth point, and so only one branch. But, what if the computer crashes and you have to start over?
In that case, there really are two branches. Surprisingly, though, we don't have any trouble thinking about the lost branch. We just sweep it under the rug, effortlessly revising the past, because obviously that lost branch wasn't part of the real history of the file.
Now, here's the point of that whole example. I think that if we had access to teleportation, we would for the most part deal with branching in the same way, by sweeping it under the rug. The scanning process is nondestructive, so that the original has to be actively disposed of? No problem! We would just think to ourselves, well, that wasn't the real me … even though, in fact, it was. That's how we would work around the fact that language is not designed for thinking about multiple branches.
@ September (2002)