Could there ever be sufficient evidence to prove god’s existence? Given the possibility that we live in a computer simulation, perhaps not.

Atheists commonly say that whilst they do not believe in a god, this view could easily shift if presented with the right evidence. For example, if the stars aligned themselves into a particular message, such as “Jesus Christ is your lord and saviour”, and this evidence was shared by all observation, then this would change their mind. Phenomena of this type would defy all known laws of the universe, in addition to implying agency of such phenomena. In other words, atheists would require unambiguous evidence of a miracle to believe that a god exists.

But would unambiguous evidence of a “transgression of a law of nature” (Hume) really equate to evidence of god’s existence? My views have been flipped on their head after reading philosopher Nick Bostrom’s Simulation Argument, detailed in his  2003 paper ‘Are you living in a computer simulation?’ If his argument is correct, perhaps no evidence would be sufficient.

This is different from postulating that an aberrant cosmic sign or the apparent return of Jesus or Mohammed could be the mischievous work of a super-intelligent being. This would be similar to saying we could be brains in a vat, or that the universe could have came into existence only a few seconds ago (with each of us endowed with memories of an entire lifetime). The Simulation Argument and its implications run much deeper than such vague and infinitely varying speculations that one cannot prove or disprove; namely that it places quite specific constraints on what could be true. If there is reason to seriously consider we might live in a simulation, could this reason trump reasons to believe in god if faced with a confirmed miracle?

The Simulation Argument

The Simulation Argument offers three possibilities, one of which is true. Either (1) The human species is almost certain to go extinct before reaching a “posthuman stage”, (2) Any posthuman civilisation is extremely unlikely to ever run simulations of their evolutionary history (or variations thereof), or (3) We are almost certainly living in a computer simulation. “Posthuman” here refers to a civilisation that has acquired most of the technological capabilities that one can currently show to be consistent with the physical laws and with material and energy constraints.

The argument gets off the ground with the assumption that consciousness does not require a biological substrate, otherwise known as ‘substrate-independence’. There is nothing necessarily special about the carbon-based wetware of our nervous systems, but rather any system that implements the right sort of computational structures and processes can be associated with conscious experiences. Though this is not uncontroversial, most philosophers of mind and neuroscientists would agree with this view.

Next, one can predict with a great deal of certainty that the computational capacity of posthuman civilisations would be compatible with the requirements of running so-called ‘ancestor simulations’. Whilst one cannot predict the upper bounds of such a civilisation’s computational abilities, establishing lower bounds can be much more precise by assuming only mechanisms that are already understood. For example, a computer with a mass on order of a large planet could perform 10^42 operations per second (the creation of quantum computers would elevate this number much higher). On the other hand, the amount of computing power needed to emulate a human brain can also be roughly estimated, which stands at up to 10^17 operations per second. Therefore, the cost of a realistic simulation of human history would then be around 10^33 to 10^36 operations*. If a planetary-mass computer were used, it could simulate the entire mental history of humankind by using less than one millionth of its processing power for one second. In other words, the cost of performing ancestor-simulations would be vanishingly small for posthuman civilisations.

When we consider the immense computing power of post human civilisations, and we assume that consciousness is substrate-independent, one of the following propositions must be true:

1. Fraction of all human-level technological civilisations that survive to reach a posthuman stage ≈ 0
2. Fraction of post human civilisations that are interested in running ancestor-simulations ≈ 0
3. Fraction of all observers with human-type experiences that live in simulations ≈ 1

For (1) to be true, we must assume that humankind is doomed to fail to reach a posthuman level. If virtually no species at our technological level reaches posthuman stage, there is no reason to believe we are in some way special and exempt from this proposition. For this to be the case, there would have to be one or more systematic and catastrophic errors that all humanlike civilisations converge on in their quest for progress and prosperity.

With respect to (2), motivations for running ancestor-simulations are easy to imagine when one acknowledges that many humans today would like to run such simulations if they could – whether it be for academic, creative or leisurely means. Therefore, for (2) to be true, no posthuman civilisation would contain any wealthy, independent agents who have human-like desires and are free to act on them. Any reasons for precluding ancestor-simulations, ethical or otherwise, would have to be essentially ubiquitous. Even a minuscule proportion of posthuman civilisations who do not share these views would result in a sufficient number of simulations.

Therefore, unless we are currently living in a simulation, our descendants will almost certainly never run an ancestor-simulation, by virtue of certain catastrophic failure of our species or a complete lack of interest in running ancestor-simulations; this is the Simulation Argument in a nutshell.

The Simulation Argument must be distinguished from the Simulation Hypothesis; the latter being the assertion that we are indeed living in a simulation, which Nick Bostrom does not support. The conclusion drawn from the Simulation Argument is not that we are definitely living in a computer simulation, but rather it places specific constraints on what can be true.

Explaining miracles

The question I am pondering is as follows: If faced with a miraculous event, which is more likely – the Simulation Hypothesis or the God Hypothesis?

Let us look briefly at the God Hypothesis – that a god is the cause of a miracle. Firstly, we have every reason to believe that the major monotheisms are false. Not only has there been no evidence to support any one of them, they are mutually incompatible. Their holy books contain archaic and heinous sets of morals, contradict basic facts of physics, cosmology and biology (to name of few), and are internally inconsistent. In other words, they have all the signs of being written by ignorant first and sixth century people. Furthermore, we have both modern day religions (such as Scientology and Mormonism) and a theological graveyard of discarded gods (such as Zeus and Thor) to know that myths of this sort can be manufactured and believed by millions of collectively deluded individuals. Secondly, in the interest of brevity I will summarise thousands of years of philosophy in saying there has been no robust argument for even the existence of an impersonal, ‘prime-mover’, ‘first-cause’, deity.

Now we shall consider the Simulation Hypothesis – that we live in a simulation. Without sufficient knowledge to stack the scales of the three possibilities allowed by the Simulation Argument in any consequential way, the chance of you being in a simulation is significant. As Nick Bostrom has said, “In the dark forest of our current ignorance, it seems sensible to apportion one’s credence roughly evenly between (1), (2), and (3)”. What follows from the Simulation Hypothesis is that its creator would be the cause of any confirmed “miracle”.

Conclusion

It seems the possibility of the God Hypothesis being true is negligible, whereas the possibility of the Simulation Hypothesis being true is non-negligible. So if we are one day faced with undeniable evidence of a miracle – however large or stupefying – given our current knowledge a betting person would have to put their money on the Simulation Hypothesis, rather than the God Hypothesis. It is more likely that a super-intelligent race is tweaking the dials of their simulation, rather than a god tweaking the dials of their universe.

This conclusion appears to place me in the uncomfortable company of the religiously faithful. Namely, I could not be presented with evidence to prove the existence of god, just as the devoutly faithful could not be presented with any evidence to disprove the existence of god. This seems to counter the view that a scientist should change their mind when evidence is forthcoming. However initially disconcerting though, I believe this conclusion can be arrived at through available evidence, whereas faith is the strong or unshakeable belief in the absence of evidence.

Interestingly enough, a super-intelligent race capable of running an ancestor-simulation would be indistinguishable from a deity in many ways. Firstly, they would be omnipotent and omniscient with respect to our simulated universe; all-knowing and all-powerful. Secondly, they would be capable of manufacturing a kind of afterlife – a place where they could upload our consciousness after our apparent death – should they feel that allowing simulated beings to perish is unethical.

Alas, I find myself in unfamiliar waters. Maybe I am a kind of believer after all.

[* The environment places further computation costs, but through compressed representation Bostrom argues that it would be negligible compared to the requirements of simulating the minds within that environment.]