The Simulation Argument — Jailbreak!

This forms most of a chapter in a book that I wrote for the collection of essays on the Simulation Argument, by Antonin Tuynman. The book can be found on Amazon.

OK — so we are trapped in a simulation — how do we break out? But before we do, perhaps we should give some thought as to whether this is a prison… or a sanctuary. No? Alright then let’s get digging…

First, we have to decide what type of simulation we are in, because a lot depends on the thickness of those walls and bars. So, what are the most likely types of simulation (or Sim) in order of probability, and their scale?

There are various possible degrees of simulation of reality that, although they might appear to us to look the same from the inside, would in fact be utterly different especially in terms of the computing power required to execute them. In addition each type would likely have completely different motivations behind their creation and certainly require vastly different resources. However, before we do that let’s eliminate the “cheapest” version, which is the one depicted in the movie The Matrix. This is where we are essentially a “brain in a jar” with our senses hijacked and are fed false information. What makes this exceedingly unlikely is the effects of drugs like LSD, which do not merely distort sensory input but which create unique mental states. That could not happen under these circumstances using only normal sensory input.

Other scenarios we can generally rule out are computer games, at least of the trivial entertainment variety. We appear to live in a lawful world. There are no overt Gods or Supermen flying around subverting the laws of physics at will. Finally, there is no way of determining when this simulation started. It could have been billions of years ago or last Thursday.

So let’s start at the one that, for some reason, people assume is “the” simulation.

Scenario 1 — The Universal Simulation

This is the most “expensive”, or computationally intensive scenario. This is a simulation at the Planck scale of our universe. For those unfamiliar with the term, suffice it to say that it appears to be the smallest scale possible, being some 20 orders of magnitude smaller than the nucleus of a hydrogen atom (that is, one followed by 20 zeros). It is where concepts such as space and time cease to have any meaning. If our universe is a simulation, and is being simulated at that level of detail there are two conclusions we can draw from this. The first is that the simulation will be absolutely perfect and not detectable. The second is that it must be simulated from a universe where the laws of physics are substantially different from this one, because it is unlikely in the extreme that our universe possesses the necessary computing resources. Given this, there’s not much point on speculating further, although of course scientists have done so, most notably Frank Tipler. He postulates that in the final moments of a collapsing universe enough energy becomes available to run an infinite number of simulations at such resolution. Apart from obvious problems such as the fact we do not have a suitable physical theory that would allow us to definitively state that it is at all possible, it has subsequently been discovered that our universe appears to be open. That is, it will not undergo a final “Big Crunch” as required by Tipler. However, as long as “some universe somewhere” supports such physics then it might be conjectured that our universe is one of those infinite simulations.

Chances of escape? Effectively zero unless we can marshal resources on a trans-galactic scale.

Scenario 2 — The Elder Gods

This is where things start to get interesting…

Consider the notion that the first civilization to arise in our galaxy (the Elder Gods) goes the Transhumanist route and expands to effectively strip-mine its own galaxy which it converts to Computronium (the apocryphal universal computational substrate) in order to maximize resources. As to why they would do this… well, we have barely got into computers and we are already using a significant fraction of our total global energy to power them, and it continues to increase exponentially. Computer power is the essential resource for an advanced civilization. Note that this could have happened billions of years ago.

Clearly such an act would curtail the evolution of life throughout the galaxy and certainly block naturally arising intelligence. From the Elder Gods point of view this in itself might be considered a great material loss as well as being unethical. So, how can they both have their cake and eat it? The answer is to run every planet they convert into Computronium in a simulation, most probably “on site” in a small corner of the Computronium that was made from the planet. Evolution would then proceed normally, or at least as normally as could be expected. In the Sim life would arise, then possibly intelligence. And if it followed the same route it would expand through its simulated universe converting all resources and recapitulating the steps of the Gods. However, there would be a number of differences.

The first, and most obvious, would be that only the target planet and its local environs would be rendered in any detail. In other words, from the inside it would look like there is only one civilization per universe. The inhabitants would ponder the Fermi Paradox…

The second is that as soon as the newly arrived locals (us) started converting their simulated locale into simulated Computronium they would suddenly hit a metaphorical brick wall. Because the matter in our universe is only coarsely simulated, the Sim itself would be limited by the amount of Computronium allocated by the Progenitors in the real world, which would be vastly less than the apparent bulk of the planet, let alone the universe. Indeed, it might only amount to a few tonnes. The result would be that as we create our own Artificial Intelligences or became PostHuman the whole facade would break down.

At which point one of two things would happen. Either the Elder Gods would terminate us, or they would invite us to enter the real world. In other words, there would be an unveiling, or revelation, of the true nature of reality. In a very literal etymological sense it would be the apocalypse.

Nevertheless, before this stage is reached there would be a brief intermediate state where a PostHuman civilization of modest means would be feasible. At our present rate of progress it would likely last somewhere between a century and a millennium. During this period we could expect to possess sufficient processing power to run Sims of our own. In fact, there is no reason why Sims could not be nested, with each requiring less processing power than the one above it. The only thing we can state with certainty right now is that if we are living in a nested Sim then ours is right at the bottom.

As for escape, we are building the tools to do just that even as I write. The result being, we will be paroled or executed, depending on our behavior and the likelihood of us being good neighbors when invited to meet the landlord for tea and biscuits.

Scenario 3 — The Ancestor Sim

This is a very popular explanation. The idea is that some future version of Humanity runs Sims of the past for various reasons ranging from historical research of counter-factual outcomes of events (what-ifs) to sociology, economics, demographics and so forth. Equally probably, or perhaps even more likely, is that we are in a Sim run by extraterrestrials who have picked up the TV and radio broadcasts from Earth and are using this as a way to investigate who we are. As such it forms a subset of the Elder Gods view and shares some of its features.

The question of escape becomes one of an escape into an alien reality far from the Earth we know both in time and space. Nevertheless, there are a couple of obvious methods we might try. The first is to render the Sim redundant by publicizing the fact that we are in a simulation and convincing the vast majority of people that this is so. It’s a giant message shouting: “The game is up — we know what you did!”, rendering their experiment futile from that point on. As above, it is a serious risk which depends on what they think of us.

It may also be important to try and find, or at least guess, where we really are. Where we are being “run”. Given the serious computer power required suggests something extraordinary if it is not cosmic engineering, of which we see no evidence in our sky. So, speculating wildly let me make a suggestion — we are currently located some 27000 light years away in or very close to Sagittarius A* aka the black hole at the center of our galaxy, and the date is somewhat later than 29000CE. Or possibly in one of its tens of thousands of orbiting black holes. Why there? Well, it may be possible to use a peculiar feature of a rotating hole called Malament-Hogarth spacetime in which hypercomputation is possible, which allows near infinite computation to occur in a finite time.

The more subtle way of attracting the attention of the aliens is to send a message directly to the site of where we are, using lasers, radio waves etc. with some kind of embedded message. But, I hear you ask, won’t that message take 27000 years to get to the galactic center? The answer is “no” — because we are already there. It is something that I, as part of Zero State, am actually doing as a subset of the METI program (messaging extraterrestrial intelligence). I shall keep you informed of any resulting godlike manifestations…

Scenario 4 — Judgment Day

This is a version of the ancestor Sim. For example, one of the most plausible methods for reconstructing the dead of past ages, or at least us, is from records such as DNA, medical records, photos, writings, videos and so forth. The argument runs that if a simulation of (say) myself could be created such that in the simulation I am writing exactly these words at exactly this time it would be a fairly accurate reconstruction of the historical “me”, long dead. However, would that really be true? The counter argument is that it is only a copy, albeit possessing its own life and sentience and not the “real” me at all. I remain dead. To actually be me the copy has to have identical brain states with the original. Unfortunately some fairly crude back of envelope math suggests that the necessary information output from a person is insufficient by at least three orders of magnitude to select one unique state from a possible ten to the ten to the sixteenth power states.

Hence most of the data used in the reconstruction e.g. “what I had for breakfast on 1 January 1990” has to be a guess that, although possibly having an effect on defining my mental state, may not affect the words that I am typing now. Does it matter? When is a copy good enough to truly be “me”. The answers are unknown, but my own view is that if I am to be brought back from the dead there should be a subjective continuity of consciousness which an imperfect copy cannot possess. Otherwise, it is someone else — a nearly identical twin, but not the real me.

However, if we live in a sufficient large multiverse (one or more of several different types) this may not matter because somewhere we can achieve perfection in the reconstruction. This is where the multiverse can “rescue” the situation because part of the reconstruction process can randomly guess what should fill the information gaps. The result of making that random guess is a spread of possible versions of the deceased across the multiverse, including at least one that exactly matches the original. So, the entity doing the reconstruction gets a resurrected person that exactly matches all the data they have of the original — which is as accurate as they can get. On the other hand, somewhere in the multiverse a true and perfect copy is produced that has the requisite continuity of consciousness.

Typically, the pertinent questions concern who is running such simulations, how many of them there are, and why they exist. The answers I suggest are a PostHuman “us”, billions, and they exist in order to resurrect our dead families. They are necessary in the resurrection process because of a consequence of the Halting Problem, namely that it is in general impossible to jump to the output of a program, in this case a reconstructed personality, without executing the intermediate stages — the life of that person.

This in turn implies that if the above is true it is overwhelmingly likely that we are in just such a simulation. The period we live in is probably unique in being the most information rich in history to date, the one where people still die and also the one that immediately precedes the Singularity. That being so, there are people alive today who will still be alive in the PostHuman transition. However, it is likely that those people will have parents and grandparents who did not make it. Reviving family would, for me at least, be a very high priority. So too presumably for billions of others.

This naturally leads to what may be termed Judgment Day, so-called because some hard choices would have to be made. For example, do we really want Jack the Ripper and Ted Bundy revived and let loose? I think not. Those people would appear to be obvious cases for non-revival, but where is the line to be drawn? There are other options. The final personality could be substantially modified in order to “wash their sins away”, or at least their desire to sin. Or if some turning point in their past could be found then they could be set on the road not taken, and “reincarnated”.

Which returns us to a question quickly dismissed at the beginning of the chapter — do we really want to escape from what would be our ultimate life support machine? I would suggest not, but…

How might one end this simulation and return to the real world? It has been speculated that there might be hidden codes one could utter, like spells, in order to shut down the program or remove oneself from it. Clearly, speaking the words “End the simulation!” does not work and it cannot be something so obvious. Or can it? Perhaps every one of us already knows the exit code — we call it death. However, before you hang, shoot or poison yourself in order to exit a particularly nasty bit of what you believe to be simulated reality, ask yourself this: “Why am I here in the first place, given that I would have inevitably known such a state was not only possible but likely?” Pushing the exit button may well simply result in a “Fail” and you might be forced to resit the exam. Again. Alternatively, allow it to run its course to one of two possible endings. The first is, as mentioned, death. In which case it would be just like waking up in the morning from a particularly vivid dream. The second is more radical, which is to get out alive by riding this Sim’s technological wave to the Singularity where our consciousness expands and merges with the already PostHuman self.

Scenario 5 — The Solipsistic World

Solipsism — the notion that only “I” exist and that everything around me is a figment of my imagination. Computationally, a solipsistic Sim is the cheapest. In other words, only one thing is being simulated in any detail, and that is me. Indeed, if done at the neural level we almost have the power to do it now. If this is so, where does that leave other people? It rather implies that the vast bulk of Humanity consists of what the games world calls NPCs, or “non player characters”. They do not have to be simulated in any detail whatsoever except as they interact with me. At the lowest level they would be only be assigned enough resources to pass the basic Turing Test. For example, it does not require much for them to say things like: “Do you want fries with that?” when ordering food from one of them. At a higher level, with friends and family, close proximity and detailed interaction may possibly raise them to the level of full consciousness, at least in my presence. Additionally, most of the world will not exist in any detail except as I move through it. What happens when I am not around would be some kind of superficial evolution of gross features almost like a soap opera on TV, with a basic script being worked through. This would be applied to both people I never meet and geography.

The key to escape is the realization that there is only limited computing power, and hence the world is going to be incompletely fabricated.

On one hand, this is as close as we get to being in a computer game running on seriously under-powered hardware, and on the other being in a dream. It is the latter which would appear to be most likely in terms of probability, given that this scenario probably stems from a single PostHuman mind. It may not even be a deliberate Sim. Even now, we constantly run simulations in our own heads, for example we imagine things like “what will happen if I turn up for work late?” by running through the scenario complete with models of the people involved acting as we expect they might. A PostHuman speculating, or remembering, what it was like to live in these times may well render a Sim to the level of detail we see in the world around us. In this case almost certainly the person running the Sim is myself after I have been augmented, and the likely time period is within this century.

The people I meet, the NPCs, are all aspects of my true self, the dreamer. As such they will be deeply connected to aspects of my own psyche. Indeed, as I consciously alter my psyche by adopting various persona or emphasizing aspects of it I will see this reflected in the world about me. This is the power of the solipsistic view of the world and has been recognized in many religious, and mystical traditions.

In which case we need to wake up to our true self.

One method is to retreat from the world hermit style. Ideally to a featureless cave and do as little as possible. The aim is to force all the computing power that is normally expended on the rest of the world to focus solely on me and my state of mind. If there is nothing for me to dream except the major character (me) then that character becomes all important and begins to merge with the higher levels of dreamer consciousness. The drawback is, naturally, that one gets to do nothing but meditate in seclusion.

Alternatively put a squeeze on the processing power available for the Sim, and hence force it to render the world in less detail that it might otherwise do. We do this by executing a series of programs that absorb a maximum of resources. In other words, we execute “people”. Not by merely interacting with them in general in a robotic manner but in the level of detail that forces the Sim to raise them to full consciousness (or close to it). We get a group of people who are close to us and interact in as deep an intellectual and emotional basis as possible. Coupled with this the group should have a common aim or purpose. Such a group, and objective, only needs to be supported for a duration sufficient to bend a more malleable world to the only Will that now exists in any measure.

Or… raise the computational power required to maintain my own consciousness and that of my group by taking something like LSD. When people take LSD while having a brain scan it lights up like a Christmas tree, and subjectively time dilates enormously because so much is going on in the brain.

All recipes for a literal change of consciousness, but ask yourself — what happens to the characters in your dreams when you wake up. Better hope you are not an NPC!

Scenario 6 — On The Edge

Again, this is a resource limited scenario where whatever machinery that runs our reality is close to the edge of maxing out. So, what would such a reality look like?

Well, for a start there would be various anomalies as the system failed to keep track of things or maintain logical consistency as computing requirements momentarily exceeded capacity. Yes, glitches is the matrix! You thought you put your keys on the table but they are not there. Except, when you look again, they are. You made a mistake. Or something did, and then rectified it. This is a world of one-off anomalies and unrepeatable synchronicities. It is a world where psychic phenomena (Psi) occur, but is not amenable to experiment because the system patches the holes as we discover (or create) them. Ghosts, UFOs, cryptids… As to how the patches are implemented, well at the more trivial layers of complexity it can probably be quite easily automated with deviations from the script fixed as they occur. Background helper programs would likely be used. In the original Unix operating system they had a name — Daemons. Most are stupid, but the ones handling more complex screw-ups would have to be high level AI.

Escape is unlikely, but there are possibilities as previously mentioned of locally overloading the Sim, which if pushed far enough might globally crash the system. It may well attract the attention of whoever is running it, and it would most certainly attract the attention of high level Daemons. However, the most likely outcome is that the system does a restore to a previous point. There was no nuclear war in 1985. If it is not a perfect reset there may be people with residual “false memories”. They even have a name — “The Mandela Effect”, so called because a disproportionatedisproportion number of people of a certain age remembering Nelson Mandela died in prison and did not go on to lead South Africa.

Overall it is bad news, primarily because whatever is running it either cannot or will not provide sufficient power or programming consistency. It is exactly like the kind of thing we might do in around 50 or 60 years time when virtual reality and brain interfaces come of age to seamlessly(?) give us full sensory immersion. In which case we are either Humans on the edge of becoming PostHuman and are jacked in with an added memory block and false memory overlay, or we are AI programs ourselves. In fact, given the comments above about the LSD experience it does rather lean towards the latter. The best we can hope for is that it is a version of San Junipero as featured in the remarkable Black Mirror TV series episode of the same name.

So what other horrible possibilities are there? In no particular order…

I have taken a rather powerful designer drug that has dropped me into a lucid dream state, complete with psychedelic time distortions and amnesia. It ends… well, who knows?

I am a bored teenager circa 2100CE who is doing his history project on life 00 years ago — what was it like? (Complete with full pain settings). It ends when you die.

It’s a game like “Life of Roy” in the TV series Rick and Morty where a VR game encompasses and entire lifetime in a few minutes. It ends when you die.

We are hallucinations inside a Boltzmann Brain, which is a self aware entity that arises from random fluctuations in a dead universe in thermal equilibrium. In which case there is absolutely no way out, or rather, any escape is to black infinity. It ends when you mercifully die in the perfect prison.

Conclusion

Most of the methods of breaking out of a simulation are probably going to happen anyway given time. Doing so is going to involve considerable risk and even if successful will probably not result in us literally “escaping”, since I doubt that the next level up will provide suitable conditions for the life support of a Human body. More likely is that we will open up a communications channel between them and us, and maybe we will be given tools to exert a little more control over our reality.

Another thing to note is that anything capable of running “us” is by definition capable of running an advanced Artificial General Intelligence at least as smart as us and probably way beyond our individual capabilities. Do we really want to have a talk with the “caretaker”, especially if we are messing things up?

Finally, as the Buddha might have noted, no matter what this reality may be we can only know for sure the truth of two things — suffering and mathematics. Pain still hurts and logic is logic. Now… play the game.