In Greg Egan’s Permutation City people are able to create digital copies of their minds that run on computer hardware. The novel tackles the concepts of self-identity and self-awareness very well, but, as a story, it kind of peters out towards the end, almost as if the author was tired of the tale and finished it in a hurry.
In case you have never heard or thought about the concept of uploading your mind into a computer, there is a startlingly long wikipedia article on this. There are also a bunch of people running around saying that this technology is round the corner. They call this the singularity. Don’t pay them any attention, they have no idea what they are talking about. Thinking about mind uploading is a great mind experiment, however, and leads to deep existential questions.
Egan deals with two of these questions. The first is the set of paradoxes that arise when we consider the possibility of copying our consciousness into multiple parts. The second (and I think Egan is very innovative to deal with this one) explores how these copies are actually computed, and whether the manner of computation affects the self-awareness of these beings.
Suppose you copied yourself into a computer. Which one is the real you? Do you now simultaneously exist, separated in space? Is the other you feeling “you” just like you are doing, right now? Egan does not spend much time actually exploring this question, but it is alluded too. One of the characters is quite serene in knowing that their copy is living in an interesting world with interesting characters. I do not think I would be that serene. I identify more with the feelings of the copies which, upon awakening, suddenly decide to terminate themselves and escape from the limited world of simulation. I also identify with the anguish of the original, who realizes they have just tortured another sentient being.
This brings us to what I think is the most interesting aspect of Egan’s novel. Egan does not treat the simulation of minds as perfect and a solved technology. The simulations of his copies do not run in real-time. There is not enough computing power, not even for the filthy rich. So his copies live in somewhat isolated worlds where interface to the physical world outside is made difficult because it passes by like a movie in fast forward.Their worlds are very coarse, with a few parts being simulated in detail, but most of the world a simple backdrop with no illusion of reality. This is the aspect, it is hinted, that leads to many copies wanting to self-terminate.
Even more interesting, Egan alludes to the idea that the copies are not being run on full blown simulations of brains including for example simulations of spiking neurons, and of neurotransmitters at the molecular level. Rather, his simulations are simplified models, that extract out the essence of brain computation, but at a reduced computational load.
From here Egan describes a series of experiments the main protagonist runs on several of his copies where the nature of the simulation computation is changed, and the copies report their subjective experiences. As an ex-neuroscientist who once built simulations of small neural circuits, I was extremely thrilled to read Egan’s description of time-steps in simulations. This is a very basic fact of all digital simulations: time is not continuous, it has been chopped up, often fairly coarsely – say in chunks of 1ms – and the continuous differential equations are discretized so that they can be solved in a digital domain.
No mind simulation, or computer intelligence story I have read, has ever bothered to look into the black box and explore the concepts and paradoxes that arise when we realize that everything is the result of the iterations of a set of discretized equations.
Egan’s protagonist runs experiments where time steps are changed, often by orders of magnitude. In one experiment the order of computation of the equations is changed. In Egan’s telling, the simulated intelligence – copy – does not perceive these alterations – their reality is unaltered.
Egan, unfortunately, drops this line of exploration in mid flow and perhaps there is not much further to go with it, because it is extremely bizarre and paradoxical. The story, as I mentioned, kind of peters out. But my mind had already been blown, and I did not care even if he ended the story very weakly.
Highly recommended if you are interested in conscious machines and mind uploading, and definitely if you have ever done any kind of computer simulation and are a fan of the Chinese Room paradox.
