Sunday, July 11, 2010

Rebuttal to the Simulation Hypothesis

According to Nick Bostrom’s simulation hypothesis, every universe’s inhabitants would be equally likely to be living in a simulation, even if they were running simulations of their own. This leads to the uncomfortable conclusion that our universe is much more likely to be one of a huge number of simulations, rather than the one parent universe.

From a logical standpoint, this argument makes sense to me. But I’m always eager to poke holes in philosophical arguments, so here’s my best rebuttal as devil’s advocate. It doesn’t directly attack the logic of Bostrom’s philosophy; rather, it creates a probabilistic argument that we are NOT a simulation.

Let’s assume that simulations can be “turned off” by the parent universe at any time. Perhaps the inhabitants decide that the simulation is no longer needed for whatever reason, or perhaps the simulation is accidentally destroyed, or perhaps they are in a simulation themselves which is turned off by their parent universe. If this is the case, it would break the simulation chain. If Simulation A was the parent of Simulation B, which was the parent of the Simulation C, which was the parent of Simulation D, the inhabitants of Simulation A would be able to break this chain and destroy all of the simulations in this chain by turning off Simulation B.

The simulation hypothesis concludes that we are in a simulation in all probability, and that every universe is equally likely to be a simulation. This means that the universe that begat ours is also probably a simulation, as is the universe that begat our parent universe. If this is the case, it would be very likely that our own universe is merely one node in a huge chain of parent universes.

But this creates an interesting question. If any of those universes could break the chain at any time by turning off their simulation, the probability that not a single one of them would do so must be extraordinarily low. This strongly suggests that we are not in a simulation.

There are a couple of responses to this argument which I can foresee, so allow me to preemptively address them. Some might invoke the Anthropic Principle. It doesn’t make sense to marvel at the unlikelihood of our own existence, they will reason, because if our universe had been turned off we wouldn’t be here to speculate about it. In my opinion, this is a flawed application of the Anthropic Principle because there is another plausible explanation for our existence: Our simulated universe hasn’t been “turned off” by any of its parents because they don’t exist. We are the original universe.

So we have two possible explanations for our own universe. Either we are in a simulation, and are here due to the infinitesimally unlikely whims of an unimaginably vast chain of parent universes…or we are simply not a simulation. If these are the two possibilities, the latter seems much more likely from this probabilistic standpoint. It also has the advantage of surviving Ockham’s Razor.

What do you think? Is my probabilistic argument for our actual reality as strong as Bostrom’s argument for our simulated reality? What flaws do you see in my logic?


  1. Devil's devil's advocate would point out that Ockham's Razor just as easily (if not more so) favors the Omphalos hypothesis, which pokes a large hole in your "infinity parents * likelihood of a node to continue for eons = unlikely probability" argument by cutting out the eons part.

  2. Hi Michael,

    Since time in a simulation wouldn't necessarily correspond to time in the parent universe anyway, I think the Omphalos Hypothesis is a moot point. In my view, it doesn't really matter if our universe has existed for one second or 13.7 billion years if it's a simulation, because it doesn't tell us anything about how much time has elapsed in our parent universe.

    Therefore, I think that "the likelihood of a node to continue to exist for eons" is the wrong way to look at it. I think that "the likelihood of a parent universe to turn off its simulation at some point" is better.

    Thanks for reading, and thanks for your input!

  3. From my point of view, you just supported the point I was trying to make. I assume I was too brief and vague, so I'll see if I can clarify.

    I'm invoking the "Last Thursday" angle of the Omphalos Hypothesis. Within a node, time and space may appear to have existed and expanded for billions of years, but that appearance could be part of the simulation that started only a few seconds ago.

    Also, as you suggest, time within a node could be irrelevant to the parent. Each node could add dimensions, causing the illusion (or reality if you so choose, since this is all rather metaphysical and unfalsifiable anyway) of eternity and space within a singularity of the parent. If this is the case, infinite branches could spawn within an infinitesimally small time of the root node.

    So when you say, "I think that 'the likelihood of a node to continue to exist for eons' is the wrong way to look at it. I think that 'the likelihood of a parent universe to turn off its simulation at some point' is better," you assume both that the roots and parents have a chance (within their dimensions) to end a simulation and that the simulations are in reality long enough to have plenty of opportunities for "some point" of turning off to have occurred.

    To further devil's devil's advocate, this extra-dimensional simulation would only have to occur once within the root node. Given the infinite expanse of the simulation relative to the root node, it's plausible that simulations within simulations will also occur. Within this tree of simulations, it's also plausible that all (or at least many and/or infinite) possibilities occur within the infinitesimally small root simulation. Of most interest is the possibly infinite subset of nested simulations that also are extra-dimensional, giving no chance for any parents to end them. If that's the case, it's likely that some of those nodes will begin as though they had already existed for eternity. Some of those nodes will also appear as an expanding universe of space and time spontaneously sprouting dimensions where previously there were none.