So today we will be looking at the Simulation Hypothesis, also known as the Simulation Argument, the concept that we might be living inside one immense computer simulation. And this is going to be a fun trip because it’s a fascinating idea especially when applied to the Fermi Paradox, the apparent contradiction between the how old and big the universe is and how seemingly absent it is of any other intelligent life besides ourselves.
So to do this we need to first do a basic introduction to the concept, sweep away some misconceptions about it, explore the concept a bit, then look at what it means in terms of the Fermi Paradox. Now the inventor of the idea. Nick Bostrom, actually does a wonderful job explaining this himself and in the reference section below, I’ve linked his original arguments and FAQ, which I’d strongly encourage you to browse if you have a chance.
I will be mostly summarizing it here and I also won’t be going through the more formal arguments he offers there, though before I scare anyone off that with fears of formal logic, Bostrom does an excellent job presenting it in a very approachable fashion and I would certainly assess him as one of best modern scholars.
Now the concept gets muddled a bit because Bostrom is actually discussing the specific notion of what is called Ancestor Simulations, which differs a bit from simulated realities in that it is specifically when you are simulating some piece of your civilizations past, or a close approximation thereof, as opposed to simulating entirely different civilizations or whole universes which might have different physical laws.
Today we’ll be looking as much at those two alternatives as Ancestor Simulations themselves, but if you have difficulty picturing that notion just think of all the historical dramas or videogames we have. Often we intentionally add a fantastic aspect to them, but just as often all effort is made for historical accuracy and a very accurate one would be an Ancestor Simulation and it is quite easy to imagine why an advanced civilization might do such a thing.
In recent articles we’ve spent a lot of time talking about how much processing power and how little energy might be needed to simulate a human mind, partially to lay groundwork for today, In those articles, most notably the recent article on Transhumanism, I mentioned the Land Auer Limit, the apparent absolute minimum energy at any given temperature to flip a bit, or essentially the absolute minimum cost to perform calculations, and our best current guesses at how much computing power we need to simulate a human mind, called whole brain emulation.
I won’t repeat that now, you can check that article from blog if you want more details on the concept, but we found out that even at the relatively high temperature you and I live at, as compared to cold deep space, you could hypothetically run a million human minds for the same energy needed to run a light bulb.
Now obviously achieving that efficiency would take a lot of improvements in modern technology, and at the same time if we are talking about ourselves as the simulations we have no idea what the physical laws of our simulators are bound by let alone the temperature they are running us at, but if we are talking about it just in terms of a more technologically advanced version of our own civilization interested in running a near perfect ancestor simulation of us now, with our 7 billion people, they could conceivably do that for as little as a megawatt.
Which in terms of current power cost would translate as a bit over $100 an hour, or about a million dollars a year. Now ignoring that something like fusion reactors might make that a great deal cheaper, something we’ve looked at on this blog before, it is not hard to imagine at all that we might run such a whole planet simulation or indeed many hundreds or thousands of them simply for entertainment purposes alone.
Now in the 1980s and 1990s we saw a great interest in virtual reality, which probably helped give rise to the Simulation Argument in the early 2000’s, and we also saw about then the takeoff of Reality TV shows, which I think all by themselves represent strong evidence in favor of the concept. Some of you may be familiar with older philosophical concepts like the Brain in a Vat, as well as general thought experiments about how we can know if everything around us is real or just a dream. You might wonder if Bostrom’s notion is anything new or just a modern flavor on those old concepts and my mention there of evidence, in regard to reality TV shows, is a big part of that.
The simulation hypothesis is presenting this concept from the perspective that there are things which could be interpreted as evidence that we are in such an Ancestor Simulation or that such things are clearly possible. Bostrom presents us basically three scenarios for the future, one’s that will feel pretty familiar to those of you who have seen the Fermi Paradox Compendium article where we contemplated ways in which a civilization might continue after it reached our current technological level.
Those are, to paraphrase:
- That civilizations go extinct before acquiring the necessary technology to do ancestor simulations, or that they turn out not be possible.
- That civilizations who can perform ancestor simulations do not wish to do so.
- That they can do ancestor simulations and choose to do so. In that first option, civilizations simply cease to exist, since we appear to have established that such simulations are most likely possible.
We’ve looked at those apocalyptic scenarios before and will look at more of them down the road, but like with the Carter Doomsday Argument he’s not presenting a specific scenario, just that it could happen. Technological Civilizations able to generate Ancestor Simulations may either never have a chance to develop that tech before ceasing to exist or do so shortly thereafter, for whatever reason, nuclear war, plague, etc., does not matter. So option 1, ancestor simulations don’t exist because nobody exists able to run them.
Option 2, Ancestor Simulation don’t exist because nobody wants to run them. Now we know that isn’t true for ourselves, we’d cheerfully simulate any of our old civilizations. However, we can’t assume anyone with this technology would feel the same way. They genuinely might not want to, but more importantly they also might have serious objection sat an ethical level. We discussed the concept of Mind Uploading in the Transhumanism article and while many would disagree I didn’t make much of a secret that in my own opinion a Whole Brain Emulation would deserve the benefit of the doubt that it was a genuine person.
We’ll discuss that more in a bit, but in that context, if they are people, and we assume that’s the case in the Simulation Hypothesis, then there are a lot of ethical problems. I’d feel pretty bad about recreating someone to live out a life as a Roman Slave for instance, even though presumably most of those folks considered that quite superior to not being alive.
They might regard us much the same, certainly I am quite fond of my life but it’s a good deal better than those enjoyed in many of the less fortunate areas of the world and it might have considered quite unpleasant by the standards of those a century or two down the road. The quality of life of those of us alive right now is what matters to the Simulation Hypothesis too, since it’s presenting the notion that we ourselves might be simulations.
So we can’t skip the notion entirely that even assuming many folks did want to run such simulations that they would be banned, and that probably is a decently enforceable ban at least where simulating an entire planet’s population at full WBE is concerned.
Even ignoring all the difficulties that might be had acquiring the hardware and software, a megawatt to run it is still a lot of juice and we’ve talked before about detecting power usage as waste heat in regard to megastructures like Dyson Sphere. It would be quite hard to conceal that power use from us right now, with just our modern infrared detection devices and satellites, if we were looking for it, it would probably be impossible to hide from anyone on the same planet who had much higher tech then ourselves.
Option 3 is that they do run ancestor simulations, and could easily then run tons of them, which raises the notion that we might be such an ancestor simulation ourselves. If you’ve got, as an example, a planetary population of 10 billion people and you’ve mastered this down to near that minimum power use I mentioned earlier, with a modern day cost of about a million bucks a year per simulated planet, if each of those folks coughed up$1 a year for running simulations, or only 1% was paying $100 a year, same thing, you’d have 10,000 such simulations running.
If, to keep the math easy, we assumed each of those worlds had about 10 billion people who lived through our modern period, that would be a 100 Trillion folks out there who lived through the early 21st century or thought they had with the same certainty we have now. If you were to assume you could randomly be one of the 10 billion who really lived through this era versus the 100 trillion who thought they did, there’d be a 99.99% chance you and I would be one of those simulations. And realistically, if you can program something as complex as a human mind, you can certainly program more simple automated construction and maintenance robots so you could probably do this way, way cheaper in which case even our 1% of 1% odds of being the non-simulated people would be optimistic.
That’s essentially the Simulation Argument, if we assume our civilization will eventually be able to make ancestor simulations and wants to do that, then odds are good the total number of people be simulated in our modern times down the eons to come will so massively outnumber the handful who really lived through the original one that we have to assume it is statistically improbable we are the originals and far more likely we are not. That we are the simulations. Also to clarify on the ancestor simulations, we’re talking about any occasion someone wants to run a decent simulacrum of 21st century life, this does not require it include the same people. An advanced society just wanting to know the usual time table of going from basic computers to whole brain emulations, and what percentage of civilizations do that, might run tens of thousands of century long simulations.
That’s our first connection to the Fermi Paradox, in the absence of direct proof of aliens you might run world after world to see which evolutionary tracks were possible, and you do ancestor simulation after ancestor simulation to try to figure out how probable it all was. Nor is that necessary an ethical dilemma. The Simulation Hypothesis gets a lot of comparisons to theological topics because of the obvious similarities. One common point raised in both would be “Why would God, or the Programmers, create a world with so much suffering it?”.
Now there are a number of counterarguments where God is concerned but they are less relevant to the Simulation Hypothesis in regard to ancestor Simulations because the most straightforward answer would be “Because we were replicating conditions accurately, that was our goal. “And both have the upside that you can keep going on after death in some place nicer. Even a particularly traumatic and premature death could be erased or sealed off to limit that trauma. Nor, incidentally, does everyone in such a simulation have to be a full WBE.
You probably don’t need anything nearly as complicated as a full brain simulation pass casual Turing tests by other people inside the simulation. Very advanced civilizations might even like to use such lives as a way of reproducing. If for instance they were trying to instill a deep emotional tie to their ancient cultural heritage when, considering the available technology, they might be running entirely on computers themselves.
We tend to think a certain amount of hardships, challenge, and pain are necessary components of a mature mind and they might feel the same, and anyone who has spent much time contemplating exactly how spoiled a kid might be in a truly Utopian Post-scarcity society can see where the motive for that might lie. You’d also have to consider that it needn’t be done in real-time. We’ve talked before about subjective time, in the Transhumanism article I pointed out that replacing the current means of sending information around your body with one closer to the speed of light could result in thinking a millions times faster so that you could experience an entire lifetime in about an hour.
In a post-scarcity society where you might easily have trillions or quadrillions of people hanging around it’s not hard to imagine several billions of them agreeing to log-into a simulation for an hour where most of their memories are temporarily suppressed so they can live a new life for a while. If I offered you a chance right now to go live in, say, the 19th century, with your memory wiped during the process but to be re-integrated back into you when you were done, and you’d wake up in an hour, how many people would say yes?
Especially if the reintegration of memories could delete truly traumatic events or the setup allowed you to pick a general area to live in or start with some strong suggestion in place, like for instance. “I’ve always wanted to be a football player, can we set me up in a body suited to that with a subconscious imperative to pursue that career? “For those who have been following along the articles where we talked about Dyson Spheres and available energy and populations of Kardashev 2 civilizations, you know that you’d only need a tiny fraction of the population willing to do that just a single time to ensure thousands of modern ancestor simulations ran. It’s not hard to imagine a lot of folks doing this regularly for the experience. Let alone just one time. And remember, we’re not talking about a re-enactment of our exact history, just an alternate reality in which folks are living through those same times again, doesn’t have to be the same events, same people, or anything like that.
To re-emphasize, the Simulation Argument is the notion that the total people who experienced our modern times would massively outnumber the tiny handful who really had. So much so that the odds you or I are those original people who actually experienced the genuine original version might be astronomically tiny. That’s the simulation hypothesis. Now before we look at the Fermi Paradox angles on this, let’s talk a bit about what consciousness is and also talk about other applications besides ancestor simulations.
A Whole Brain Emulation is pretty much a straight clone of an entire human brain, where you basically just got a piece of software pretending to be each brain cell. In many ways it’s not that sophisticated and it’s probably orders of magnitude more code and processing power than it would take to create a replica of a person able to have convincing interactions with other people, a Turing-Compliant piece of software. That’s worth remembering because a lot of applications where we might want simulated people to interact with aren’t requiring genuine consciousness, just something way better than we have now. A lot of times when people are thinking of virtual worlds the sort of responses needed from the facsimiles of people in order to maintain the illusion doesn’t require anything like full human consciousness.
If I want to see and interact with a historical drama on the Egyptian pyramids I don’t need to simulate anyone living in the Americas and really even just the people on the worksite need to be able to have conversations, and most of the time the individual folks in line of sight just need basic facial expressions and tasks. So by and large virtual realities don’t require whole simulations of consciousness to interact with. That is one strike against the Simulation argument because by and large people going into one for entertainment purposes or educational purposes won’t need anything like a true human brain to interact with and doing so would probably waste tons of processing power, energy, and money.
But we don’t really know where the line of consciousness begins and ends. We just assume that a whole brain emulation must be because it is genuinely emulating every brain cell, and in more extreme and probably total unnecessary versions would be emulating every molecule or even atom. It’s a computer doing this but it technically does not need to be. Or rather, it does need to be a computer but a computer is a much vaguer term than most realize when we picture the modern devices of that name. Back before the era of modern home desktops it was actually a job title given to people who performed computations. I think I’ve mentioned this before on the blog, and if you look at some older science fiction, like Isaac Asimov’s 1955 novel “The End of Eternity”, two of the main characters have that as a job title as do a whole bunch of the folks involved in the organization the book mostly revolves around. Anything a computer can do, humans can do slower.
Every calculation can be performed on paper or with an abacus. So hypothetically, in a situation much like the infinite number of monkeys typing out Shakespeare, you could have room after room of people just churning through the same calculations our computer would be to produce a Whole Brain Emulation. Is that consciousness? Are those billions of people, each quite conscious themselves, producing an independent and slow running human mind?
For that matter if I take a whole bunch of monkeys and train them to perform a couple simple calculations, basically just the trio of basic Boolean operations, AND, OR, NOT, are they busy running a human consciousness far beyond their own intellectual abilities? It’s a weird way to look at it, to imagine warehouse after warehouse of thought accumulating on paper to form contemplations and memory of some very ethereal person, but it is as valid as a computer doing it. And it is essentially what we argue our brain cells are doing to. Each one a little monkey pounding out simple operations for reasons it does not understand or care about. You could even presumably do this with ants. Each bit could be a tiny red or blue kernel coming down to shafts, one at a time each, to produce a pair.
And it is programmed to see two blue, or two red, or one of each, and for each take and drop a single orange, yellow, or purple ball down another shaft. Basic binary operation, very basic logic circuit. A very large hive of ants might conceivably emulate an entire human brain that way. A hive mind, not as we often see this as somehow centered on the queen but a genuine mind separate and distinct from them and not interested in their problems any more than we are in our own cells.
Which is to say we have to make sure all those cells are protected and fed for our own continued existence. But it doesn’t matter to that over consciousness if all those ants are replaced over the years, or if the whole hive moves with time to somewhere else. This gets us back to the notion of personhood, and in a modern example, an incorporated village or city or company or church has a somewhat distinct identity from its individual human components. The people might move in and out, be born or die, the boundaries or location or buildings may change, but it’s still arguably the same group entity. Now there’s a lot of notions for how we should regard identity but one of the most popular is from John Locke, the 17th century English Philosopher whose work was heavily influential in the construction of modern democracies and concepts of human rights.
While the science was still new, Locke was already aware that our bodies change over the material they are made of. Most of the cells in your body replace themselves every several years and the ones that don’t still replace the atoms they are made from. The substance of our body and mind, the material it is made from, changes over time. So Locke argues that personal identity can’t rely specifically on substance of the body or mind, because the bits change. A Tree is still the same tree even as it grows bigger and cycles out material or gets a branch cut off. And he also argues identity isn’t reliant on the soul. He says it relies on consciousness and in that context the concept of continuous identity gets a bit murky, because it relies on gradual change without anything terribly radical going on.
Even if a person’s behavior and opinions change, they are still the same person. Reckless or naive childhood behavior and opinions being abandoned with time no more changes your identity then reading this article does, even though the content of your brain is certainly changing a little and possibly your opinions on the Simulation Argument. Our tree is still the same tree, and our village the same village, our hive mind the same hive mind, and our personal identity the same identity, even as all the components change so long as it is gradual. Locke says consciousness is the perception of what passes in a person’s own mind. This is often restated as basically meaning your memories.
Obviously if you can tinker with those, copying them to other people or deleting them, you’ve got a break in that and some problems. Now Locke’s thinking on this is hardly unchallenged, there are many other schools of thought, and there are problems with it, but it is one that most of us are pretty familiar with and comfortable with. But again, and especially in the context of high technology, it is hardly perfect or without its problems.
In a non-organic case let’s say I knock off the top two thirds of the statue of liberty and steal it to hide in my underground treasure trove. The people of New York come by and rebuild it. Many would argue that it is, in rebuilt form, still the Statue of Liberty. This gets a bit stranger though if they decide copper is too expensive to build and maintain and opt for plastic instead.
Now imagine the city gets nuked and a century later the people of New New York twenty miles north go in to the ruins and reclaim their statue and move it to New New York. It’s even further complicated if in that same time people found my stolen two-thirds of the Statue and congress ordered the Statue of Liberty erected on new legs in the middle of the Potomac so that fifty years before the folks of New New York came into the radioactive ruins of Old New York to retrieve their ‘original’ mostly plastic version there’s been one sitting near Washington DC claiming the mantle.
Which is the original? Are either of them the original? Are they both the original? And in the case of a person, while in the John Locke view of things if I am slowly replacing bits of your brain with machines that simulated the same behavior would maintain identity it gets a bit murkier if I do that instantly, just replacing the whole brain at once. Now imagine you were running a small but growing business when you got diagnosed with a terminal illness and decided to get frozen till there was a cure. A year or two later a company figures out how to do brain scans accurate enough to read your frozen mind and put it into a nice duplicate body and your spouse is happy and decides to dispose of your frozen body, but your daughter doesn’t accept that it is the new you and fights it.
The cryonics company in the meantime accidentally mixes up their files and starts copying your mind into a new duplicate of someone else, the lab tech realizes halfway through and stops it and starts transferring the right mind, ending with some scrambled memories. While they are sorting out this mess they leave your digital copy running to compare, and since you were half expecting some sort of high-tech brain emulation or nanotech to be what resurrected you when you went on ice you, or your simulation, figure out what’s up and demand to be left on. So we’ve now got that first duplicate, the memory scrambled duplicate, and the strictly digital duplicate, and someone figures out how to cure your illness and how to safely restore damaged frozen bodies and minds. Your daughter gets you resurrected.
Now there are four of you, and the original duplicate, who has turned your small business into quite an empire, falls over dead, and angered that the daughter never treated him as he fathers writes her out of the will. You’ve got three living and one dead person all with the same original memories of starting that business and you go into the court demanding a portion of that empire and to nullify the will in regard to your daughter too. And the probate judge just stares at you, your very different looking memory scrambled duplicate, and the digital you testifying by TV, and probably has a stroke trying to sort things out. Like I said, difficult concepts, deep thoughts tend to follow that are best assayed with a lot of coffee and aspirin on hand. Trying to deal with something like a murder, if the person had gotten themselves scanned before plotting out the crime, or after plotting it but before doing it, can get pretty hazy.
What do you do if a murderer gets killed in prison and his spouse has his brain scan uploaded into a duplicate? One who does not have the memories of doing the crime or maybe even of plotting it? I should probably add that identity gets even screwier in the context of the many worlds interpretation of quantum mechanics, or its lesser known sibling called Many Minds, when we try to decide if the virtually identical two people living in a pair of parallel realities, or the very nearly infinite numbers of you in those near infinite realities, are the same person and when that breaks off, when you could be said to be distinct individuals.
Okay, so what does all this have to do with the Fermi Paradox? Since the Fermi Paradox is about trying to find alien civilizations or why they don’t exist, Simulation Hypothesis can come up in a lot of ways, but we usually mean why would we, if we are ancestor simulations, experience an absence of aliens? And that’s tricky because again it splashes over into motivations and can get very close to theological contemplations of the idea.
The big one is would God, or the programmer, make such a huge Universe just for us? Now the obvious answer would seem to be no, but one has to remember two things. First, if it is a simulation, it doesn’t mean everything else is equally simulated. I don’t really need to simulate every atom on Earth to do a very convincing job of it, I certainly don’t need to for the atoms in a galaxy a billion light years away, and it is kind of creepy how our physical laws, the speed of light and the unpredictability of quantum mechanics, especially when viewed as a sort of maximum resolution of the simulation, seem to almost conspire to be handy for the purpose of simulation that isn’t running everything fully but only simulates stuff in more detail when it’s being deeply examined.
If you’re a physicist it can be very unnerving at times just how fine-tuned our Universes in a lot of ways to allow for intelligent life, and the counterargument is essentially the Anthropic Principle, something Nick Bostrom is also heavily associated with. The general notion that it doesn’t matter how many universes might exist that aren’t possessing the right quantities for intelligent life, since only the ones with them will have people in them who can notice this freaky improbability. The second thing to remember is that just because we only use a small part of the Universe now, it doesn’t mean we still will be in the future. We discuss human expansion off planet and out into the galaxy and beyond a lot on this blog and what we’ve mostly found is that the evidence is pretty good we will eventually be able to do this.
If we can, then making a universe even though you only occupy one planet right now is no more indicative that this is or isn’t a simulation then a few microbes dropped in a Petri dish should assume that there are other microbes elsewhere in the giant Petri dish just because it is so huge and empty when they arrived. We made that Petri dish specifically so they could expand into it. And we don’t want anything else growing in there to interfere with the experiment. But at the same time it doesn’t mean there aren’t aliens in the simulation, since someone might have simulated us, and several other intelligent species elsewhere, just to see how we’d interact or who would evolve to intelligence and technology first and how each unique species handles the problem.
So the short answer, much akin to when we contemplate the idea that some intelligent agency might have made humans to begin with, is that you can only make predictions regarding the Fermi Paradox if you know that Creator’s motivation for making people to begin with. Ultimately though, it doesn’t tell us much about the Fermi Paradox if we are assuming it is a genuine simulation from start to finish. You’d expect everything in it do have been following the rules of reality the whole time or best approximated to do so or not look like tinkering if someone wants to see how things play out naturally.
Now the focus of the Simulation Argument is on Ancestor Simulations but for the Fermi Paradox we need to extend that to include even wildly differing scenarios where someone has simulated a Universe in full right from the Big Bang but with different physical laws, in order to see what claws its way up Darwin’s ladder.
So simulated reality or not, it doesn’t make much difference since we can only see outside the box if the Creator of the box has either made a mistake and opts not to fix it or voluntarily pops in to inform you that you’re a simulation. So fundamentally the Simulation Hypothesis is just not a good solution to the Fermi Paradox because even if true it still doesn’t let us determine why things are so silent in this Universe, since all the other solutions we’ve previously discusses work just as well in a simulated Universe as a real one.
Because while someone might have left our Universe empty except for us, we’d also expect them to keep things self-consistent so that the science of that reality can explain the silence in that reality. If you want only one intelligent species in the Universe you configure the laws to make it incredibly unlikely to happen so you don’t have extra unwanted ones popping up everywhere draining your processors. If you do want it, then you make sure the physical laws match that, to either produce multiple intelligent species or make it seem logical that they could have been produced.
So while the Simulation Argument is definitely interesting in the context of the Fermi Paradox, especially when you contemplate simulations running their own simulations who run their own simulations, it sadly doesn’t help us contemplate it much. And on that note, we’re going to wrap things up for the day.
Next week we’ll be looking at a somewhat similar concept with the Carter Doomsday Argument. If you want alerts when that article comes out, make sure to subscribe to our YouTube channel and like the Facebook page, and don’t forget to comment and share this article with others if you enjoyed it. While you’re waiting for that next article you can try out some of these other articles and if you want to learn more about this topic, don’t forget to check out the other articles of EduQuarks or ask questions or leave comments below.
I do try to respond to as many comments as possible, and the other readers on the blog tend to be a pretty clever and friendly bunch who will probably give you a good response on questions too or be happy to chat about the subject. So I hope you enjoyed the article, have a great day, and as always Thanks for reading and we’ll see you next time!