Search This Blog

Sunday, 25 March 2012

Living in a Simulation - Some Thoughts from Within

It was about 2007 when I first encountered Professor Nick Bostrom's Simulation Argument.

You are well advised to read the original argument rather than my digested version of it, but basically it suggests that, providing it is possible to create a conscious entity in a computer simulation, and if humanity ever achieves the technological ability to run such simulations, and if humanity chooses to, then the overwhelming likelihood is that you are a conscious entity in a computer simulation.

The first condition is still a big IF. However, the second one possibly only involves humanity and our technology continuing on their current path for another few decades. The third condition I would argue is almost a certainty.

Ever since I first met this idea it has intrigued me. If it could be shown to be true, how should that affect the way I live my life? Should I try to be more interesting and entertaining, to decrease the risk of being edited out? Should I try to make contact with my programming overlords?

And might there be a way of working out for sure that our universe was really inside a computer program?

Putting myself in the place of the simulation designers, but using the computer technology available to me today, I tried to imagine what constraints or short cuts might be evident to the inhabitants of my creation, and these three points occurred to me.

  1. The constraints of working in a digital computer would mean it was easier if there was a minimum possible size for such quantities like length, time, etc.
  2. Likewise, it would be nice if it was possible to average the effects of the basic components of the universe, so that a wall, say, could be treated as a surface without needing to calculate the movements of all its component particles.
  3. There would be a largest possible number.

Now as it happens, Quantum Mechanics says that there is a smallest possible value for time and length: the Planck time and Planck length, respectively. They are very, very small: the Plank time is 5.4 x 10-44 seconds, and the Plank length is a mere 1.6 x 10-35 meters. Nevertheless, this feature of the Universe is not intuitive.

And the properties of large objects can be calculated without having to consider all their component particles, which is one of the reasons we could do Physics before the particles were discovered.

The maximum number idea was a non-starter though, as everyone knows about infinity, and some people even know about the many different infinities. It was while hoping to learn more about the fascinating world of transfinite numbers that I recently watched the BBC program To Infinity and Beyond. In the middle of a series of interviews with mathematicians, I almost fell off my chair when one Dr Doron Zeilberger said that he didn't believe in infinity. No, he thinks that if you keep counting indefinitely, you eventually reach a maximum number, after which you get back to zero. The maximum number would be very, very big, but he believes it does exist. I don't know if he is alone in this view, but the fact that any serious mathematician could hold it I find very intriguing.

If Bostrom's Simulation Argument is valid, then its discovery must surely be a landmark in the running of a simulation, perhaps hastening the point when the simulation ceases to be of interest to its creators. I would guess that the 'turn off the program' point is when the bulk of humanity's behaviour becomes influenced by the knowledge of the true reality. If so, publicly speculating on the Simulation Argument might not be quite the good idea it seemed when I started this blog post.

For which I apologise; although if it makes me more interesting...


  1. Quantum mechanics does not say that space-time is discrete, that is a common misconception. In fact, the very successful standard model description of physics relies on the existence of a continuum. Alternative discrete models like loop quantum gravity are highly speculative and lack any supporting evidence.

    As to the simulation argument, well why would anyone wish to run such a simulation? And that doesn't mention the fact that Bostrom's estimates of the technology required are hopelessly naive, perhaps purposefully so. To expound on this, his estimate of the number of operations required per ancestor simulation does not account for anything other than neural activity. What about the computations necessary to provide consistent and realistic sensory input?

    1. I'm far from being an expert on Quantum Mechanics, but my understanding was that the Planck length represents the smallest distance you can sensibly measure, in effect the spatial resolution of the Universe.

      As for why someone would want to run the simulation, ask anyone who's played SimCity or a similar program.

    2. That's a reasonable interpretation of the Planck length, but it says nothing about the continuity of space-time, merely that the position operator cannot distinguish between positions separated by less than a Planck length.

      As to the SimCity analogy, why would the inhabitants need to be conscious? Why not save computations and give the inhabitants a facade of consciousness. How many inhabitants of this simulation would even need this?

  2. I thought Planck was a dead giveaway too.

    Using technichal limits to argue against the theory isn't enough, every year we produce technology most would consider impossible a decade earlier. And any limits we see this side of the simulation may be meaningless on the other side.

    As for why, one possibility could be a training ground. If you can't play nice in a simulation maybe you can't be trusted to play nice in the 'real' world. And if you tried the "I didn't know what I was meant to do" defence you might find the answer is something like "well I did leave one or two clues..."

    1. Agree about the technical limits argument. With computing power doubling every couple of years, you're getting a thousandfold increase every 20 years or so. You can be orders of magnitude out in your estimates, but only a few decades out in real time.

  3. David, the simulation argument is about increasing our credence that our experience might be simulated. If you have to posit a new parent universe with different physics then our credence would be much lessened. This why Bostrom is at pains to suggest that such ancestor simulation are well within our future capabilities, to avoid such a massive positive claim.

    Anyway, if you have to posit a completely new universe then the argument is pretty much lost. You cannot know what the parent universe might be like, for all we know there could be an infinite number of real people, which means that Bostrom's argument wouldn't hold.

    As to the training ground suggestion, do you not see any ethical issues there? Essentially you are imprisoning innocent people.

    1. There are plenty of ethical issues, but mainly from our point of view. We have no way of knowing how advanced we would look to our Programmer Overlords, but I'm guessing "not very". If they do have pangs of conscience, they could always salve them with the reflection that we wouldn't have existed at all if they hadn't run the simulation.

    2. Bostrom's argument is posed from the position that essentially we (our descendants really) are our own simulators which is why ethical issues are often discussed and why I mentioned them, but I agree that once the idea of simulation is introduced other parent universes must be considered. I suspect he avoids this because it is essentially the same as speculating about creator gods and he wants to keep the argument on better supported ground i.e. no massive positive claims.

    3. I agree that parent universes introduces an infinity of possibilities, but that doesn't mean it can't turn out to be true. If the simulators were, say, a thousand or more years more advanced than us (a blink of the universes eye) then I'm not sure we could distinguish them from gods.
      I would expect them to treat us simulations at least as ethically as we would treat ourselves. Oh wait...

  4. And as I understand it electrons have no real position until something 'looks at them'.
    Sounds like lazy evaluation optimisation to me.