Back in time

A particularly curious edition of the BBC’s venerable science strand Horizon last night. Is time travel possible? it asked.

The programme put forward the usual theoretical evidence for mechanisms for time travel that are completely beyond our means. Then it finished matters off with a marvellous rhetorical flourish.

Time travel, it is claimed, is perfectly possible, as long as we completely redefine what we mean by time travel. Let’s redefine it for these purposes not as literally moving to the past or far future, but as having access to the past. I know, I know, but stay with me on this for a minute.

The hypothesis goes something like this:

Computing power in our society is increasing exponentially. From this we can extrapolate that a sufficiently advanced civilisation (i.e. not us) would have increased their computing power infinitely.

The realism of computer image rendering is also increasing massively right now. So we can also extrapolate that this hypothetical advanced civilisation would not only be able to compute an infinite amount of information infinitely quickly, but would also be able to render images of such fidelity that we would be incapable of telling the difference between reality and simulation.

In effect, they would be able to build a perfect virtual reality, the inhabitants of which would not even realise they are not ‘real’. By carefully recording everything, they would be able to recreate it with perfect fidelity at a later date for their own amusement or study.

One further assumption gets us to a horrifying realisation: this advanced civilisation with infinite computing power would not, of course, limit themselves to building one virtual reality; they would build, we are told, billions.

Well, you would, wouldn’t you?

We are left with a multitude of seamlessly perfect simulations. The consequence is that statistically, we are vastly more likely to be unknowingly living inside a simulation than in the real world.

Isn’t that disturbing?

On the other hand, if this Matrix-like shocker has failed to put the willies up you, maybe that’s because, recalling Descartes’ struggles with his ‘malicious demon’ and difficulties with telling dream from reality, you suspect this is just an attempt on familiar philosophical issues with a bit of dot.com era technodazzle.

Or it may be because you’re uneasily aware of the number of unproven assumptions it contains. It’s an a priori argument; an attempt to move from a small set of secure assumptions to a logically sound conclusion.

So let’s go back in time to the start of the virtual reality hypothesis and have a quick look again at some of those assumptions.

First, you will recall, we note the current exponential increase in computing power and imagine that, for a sufficiently advanced civilisation, this exponential increase would increase until processing power reached infinity.

Infinity is a pretty big number, you know. The first test here should be whether infinite computing power is, in principle, even possible. I doubt it is. Nevertheless, this could be Horizon oversimplifying. You might not really require infinite computing power. Unimaginably large might be sufficient, and I’ve got noa priori reason to doubt that someone, someday, could achieve a massive, finite amount of computing power.

Next up, and correlating with our unimaginably large computing power, is infinitely realistic image rendering power. Well, for starters, if we can get by with just unimaginably large computing power, I don’t see why we can’t reduce our requirement here to unimaginably realistic image rendering. Our eyes have a resolution limit set by the number of rods and cones on our retinas, and there’s no a priori reason to doubt that image rendering fast and good enough to fool us could be achieved.

Horizon does, however, only mention imaging. We have four other major senses that would require fooling. Again, I’ve no particular reason to doubt that a sufficiently sophisticated linkup to my brain could stimulate in me the smell of bacon cooking, the touch of silk (both with and against the nap), or the taste of a good 12-year old Scotch. I do think, though, that the hypothesis should take account of the other senses, because the ‘exponential increase’ measurement simply doesn’t apply equally across them all.

But these are minor cavils compared to my problems with what the hypothesis seeks to do with these assumptions. For they are supposed to add up to the conclusion that a sufficiently advanced civilisation could build a simulation so good that we could not distinguish it from the real world.

It doesn’t.

In fact, it’s very hard to see exactly what the hypothesis is claiming would occur; a fault, I assume, of the documentary rather than the hypothesis itself. Does it suggest that we are flesh-and-blood creatures linked, Matrix-style, to a massive VR machine? If so, there are a number of practical matters to consider, many of which concern synchronisation. If I really cough, do I cough in the simulation, and vice versa? What happens when a flesh-and-blood person dies? Co-ordinating a sudden collapse may be possible, but what if the person is revived in the real world — or in the simulation?

It could be argued that the requirements of synchronisation would be so onerous that, in effect, the distance between real experience and simulation would be decreased to the point that simulation would be pointless.

I think that, instead, the hypothesis suggests that we are just artificial constructs in a wholly artificial environment. This sounds a whole lot more possible, a little like a vastly more sophisticated version of Sim City, with us as the Sims. But, note, Sims don’t ’sense’ anything in any way that would would understand. They have directly coded responses. They carry numerical values, not sense impressions. They have, in short, no consciousness as we understand it.

I think you can see where I’m going. The whole simulation hypothesis has wandered into the artificial intelligence argument, with all of the baggage that carries; free will, determinism, the Turing test, mind/body problems, Uncle Tom Cobbleigh and all. I don’t intend to rehearse the whole set of issues here. Even if I wished to, I couldn’t. Suffice to say that Horizon alleged that the hypothesis raises difficult questions about the nature of free will. No it doesn’t. It reiterates the difficult questions about free will that we already have, just in a different context.

In short, though, we can dump those first two assumptions about computing power. They’re irrelevant. The real assumption is that computer-generated constructs can achieve consciousness. This is such a big deal that although you can argue it, there’s no way you can assume it.

In fact, it’s slightly worse than that. The Turing test aside, I don’t think that most versions of AI argument would require that an artificial consciousness be or appear the same as a human one. But the simulation hypothesis does just that. It argues that your feeling of consciousness, your experience of being you, could be a simulation. Tricky, to say the least.

Not only has the simulation hypothesis boxed itself into a very hard version of AI, it’s done it for an oddly meaningless reason. We can’t help our starting point in this argument. We have to look at it as someone simulating Earth, 2003, because that’s where we are (or seem to be).

But in order for a sufficiently advanced civilisation to simulate Earth, 2003, theymust be recording it in perfect detail. There is no other option. The fossil record is, I’m afraid, not good enough to perfectly reconstruct every event on Earth, neither are the admittedly huge number of radio waves we’re sending out.

Now, maybe you could argue that alleged UFO sightings are evidence of aliens recording every detail of our lives for later reconstruction. I won’t, but maybe you will. They would, of course, have to appear inside the simulation every time they appeared in the real world, so I think we avoid any discrepancies between real and simulated world, but I still don’t think there’s any way around the enormous improbability of it all.

Finally, there is the assumption that a massively complex, massively fast, massively extensive simulation could run indefinitely without crashing. This is an assumption that cannot be extrapolated from current computing capabilities. Ironically, the site that sets out the simulation argument is, as I write, not responding.

It’s curious, though, that the hypothesis could be reworked to avoid a large number of these problems simply by having it not be a simulation. All of the problems except the fundamental one of consciousness can be removed from the equation simply by having our experiential world be a fiction. As Charles Fort teasingly suggested, ‘I think we’re property’. I never knew what that was supposed to mean. Maybe he meant we were counters in some great computer game, just Sims, part of a wild abstract of life tweaked for entertainment value.

Time travel? I suppose games are very good at that. You can always return to the beginning for another go. If, that is, you’re really sure you want to quit this game without saving.