Review of “Reality+: Virtual Worlds and the Problems of Philosophy” by David J. Chalmers

Reality+ David J. Chalmers

Reality+: Virtual Worlds and the Problems of Philosophy by David J. Chalmers, W. W. Norton & Company (January 25, 2022), 544 pages


Virtual reality and the simulation hypothesis take center stage in David Chalmers’ book Reality+: Virtual Worlds and the Problems of Philosophy (hereafter just called Reality+). Unlike some of Chalmers’ other books, such as Constructing the World, and to a lesser extent The Conscious Mind: In Search of a Fundamental Theory, Reality+ is an easy read. Clearly meant for a much wider audience, Reality+ incorporates plenty of references to pop culture and makes sure to take time to explain in plain English some of the heavier concepts as they come up.

Reality+ is well-written and covers a great deal of material while examining some of the perennial problems in philosophy from an interesting angle. Readers will regularly have their intuitions challenged and perhaps even have to reconceptualize their entire reality. Given that we are fast approaching a time when many of the questions raised by Chalmers will become more than idle fascination, it is incumbent upon people to fully consider what Chalmers is discussing in Reality+.

There is way too much material in Reality+ to cover fully in a review of any reasonable length. This book is composed of 24 chapters divided into 7 parts that each focus on specific aspects of philosophy and virtual worlds. As a result, my review will only cover some of the main issues in the book. Namely, I will primarily be covering parts 1-3 and part 6, bringing in some elements of part 7 when relevant. I may in the future make a separate post to review some of the other things that will not be discussed at length in this post.

The first three parts of this book are the building blocks for parts 4, 5, 6, and 7, these latter four not needing to be read in any particular order. Another reason I’m doing the review this way is somewhat structural, perhaps even aesthetic: I want to cover three issues that Chalmers brings up, namely the Knowledge Question, the Reality Question, and the Values Question, which fit nicely into the three main areas of philosophy: epistemology, metaphysics, and ethics. Parts 1-3 discuss the Knowledge and Reality Questions while part 6 discusses the Values Question. Thus, doing it this way lets me cover these three broad issues within this post.

This book, broadly speaking, makes the following argument:

  1. It is impossible to know for sure that one is not living in a simulation
  2. If it is impossible to know for sure that one is living in a simulation, then there is no important difference between living in a simulation and living in the “real” world
  3. If there is no important difference between a simulation and the “real” world, then one’s personal values are unaffected should one be living in a simulation
  4. A simulation is equivalent in all important ways to a sufficiently advanced virtual reality world created by humans
  5. Therefore one’s personal values will also be unaffected should one choose to live in a sufficiently advanced virtual reality world created by humans

The first four premises can be thought of as covering the knowledge, reality, and values questions. And so we can think of part 2 as making Chalmers’ case for the Knowledge Question, part 3 making his case for the Reality Question, and part 6 making his case for the Values Question. I will cover each of these three topics in turn.

Knowledge Question

General Knowledge Question: can we know anything about the world around us?
Reality+ Knowledge Question (page 9): can we know whether we’re in a virtual world?

Descartes is, perhaps ironically, best known for his methodological doubt. It’s ironic given that it was a fairly small part of his entire life’s work. Descartes was one of those polymaths that it is difficult for anyone to become nowadays: he did a lot of work in places like mathematics, astronomy, and anatomy. But, given that those areas are places where a person can definitively be found right or wrong, and since Descartes was wrong in these areas quite often, it is in his more speculative work, which cannot be proven right or wrong, that he is best remembered.

What Descartes is remembered for is in his thought experiment of attempting to doubt everything that can possibly be doubted. Descartes proposed that everything he senses and remembers could have been implanted by an evil demon. The only thing Descartes can be sure of is that, since he is doubting, he is therefore thinking, and if he is thinking, then he must exist. The thought “I am not existing” is self-defeating, because the thought itself had to be thunk by someone, and that someone must be himself, since he is having the thought.

Chalmers says that we can replace Descartes’ evil demon with a computer simulation: it is, in principle, possible that everything we see and remember has been within a computer simulation. This could be an imperfect simulation, in which one could presumably discover that they are in a simulation (e.g. by finding clues in physics or by finding bugs or places where the simulation is incomplete). But it is also possible that we could be in what Chalmers calls a perfect, permanent simulation: perfect in that there would be no way to discover from within the simulation that we are in a simulation and permanent in that we have always been in the simulation (we aren’t non-simulated people who chose to enter the simulation) and will always be within the simulation.

Given that we can’t know we’re not in a simulation, Chalmers offers the following argument (page 56):

  1. You can’t know you’re not in a simulation.
  2. If you can’t know you’re not in a simulation, you can’t know anything about the external world
  3. So: You can’t know anything about the external world.

He later offers what could be an undermining defeater for any argument against this. He calls it the Simulation Riposte (page 64). When someone says something like “I’m not in a simulation because of X, Y, and Z” the Simulation Riposte would reply with something like “yeah, well, that’s exactly what someone in a simulation would say.” In other words, a simulated person could make just a cogent of an argument for not being in a simulation that a non-simulated person could make. Chalmers then wants to argue that it is actually very likely that we are in a simulation with the following argument [popularized by Nick Bostrom] (page 85):

  1.  At least one in ten nonsim populations will each create a thousand sim populations [or whatever number you want to use, maybe 10, maybe 100, maybe 1010, sim populations each, the point being that there are more sim populations than there are nonsim populations]
  2. If at least one in ten nonsim populations will each create a thousand sim populations, then at least 99 percent of intelligent beings are sims
  3. If at least 99 percent of intelligent beings are sims, we are probably sims
  4. So: we are probably sims

Or stated differently (page 98):

  1. If there are no sim blockers, [then] most humanlike beings are sims
  2. If most humanlike beings are sims, [then] we are probably sims
  3. So: if there are no sim blockers, [then] we are probably sims

Where sim blockers can be broadly divided into two types (page 100):

  1. Sims are impossible (e.g. it is impossible to make conscious sims or it would take more computing power than is possible)
  2. Nonsims will not make sims (e.g. nonsims will choose not to make conscious sims or some kind of great filter will prevent any nonsims from ever getting to a point where they can make conscious sims)

Chalmers finds the argument that we are likely living in a simulation convincing, as do other people. I don’t find it terribly convincing, but neither am I skeptical enough of the simulation hypothesis to completely rule it out. I’m agnostic on it. But why I’m not terribly convinced by this particular style of probabilistic argument is because it could be applied to other things while coming up with absurd conclusions. For instance, why not:

  1. There have been roughly 100 billion consciousnesses throughout the history of Earth
  2. That means I have a 1 in 100 billion chance of being who I am
  3. This is extremely unlikely
  4. Therefore I probably am not myself

Or perhaps the following:

  1. All mammals have some level of consciousness
  2. There are many more non-human mammals than human mammals
  3. It is extremely unlikely that a consciousness will find itself a human
  4. Therefore I am probably not human

Essentially, what the argument assumes is that there are consciousnesses and there are entities (whether nonsim or sim) that those consciousnesses can “inhabit” with equal probability through some randomized mechanism that stands outside of time, and therefore there is an equal chance that my consciousness could inhabit any consciousness-possessing entity (or maybe only humanlike entities) at any point throughout all of time. We therefore have to look at the base rate of different kinds of consciousness-possessing entities and determine that the more populous ones are more likely to be inhabited.

Thus, it is taken as an assumption in the probabilistic argument that my being who I in fact am was a matter of probability rather than that my consciousness necessarily emerged from the brain I do in fact have (i.e., that my brain, situated in a particular time and space as it is, and my particular consciousness are necessarily coeval by virtue of my particular consciousness being grounded in this particular brain). If the consciousness that I am could only have come about from my particular genetic and social determinants, then I could not have been anyone other than the person I am, living where and when I do.

Additionally, this line of thinking would have to be applied to the simulators as well as the simulated. Whoever is simulating us, given that most intelligent beings are simulations, must also be a simulation. The same goes for whoever is simulating our simulator, and whoever is simulating our simulator’s simulation, and so on ad infinitum. This leads to a sort of infinite regress where, although it stands to reason that there has to be some “ultimate simulator” at the top of all this, the same probabilistic logic must be applied to them as well and therefore we must conclude (wrongly) that they are a simulation. But then why not just say that if the “ultimate simulator” is going to wrongly be concluded as being a simulation, then that simulator is us, and therefore we are the “ultimate simulator” at the top of all these simulations?

I do, however, take Chalmers’ point that it is possible, in principle, that we could not know that we’re in a simulation. If the simulation is perfectly (or at least sufficiently well) designed, then we could never know. And so, lets take stock of where we’ve come by looking at the broad outline of Chalmers’ argument from above.

  1. It is impossible to know for sure that one is not living in a simulation
  2. If it is impossible to know for sure that one is living in a simulation, then there is no important difference between living in a simulation and living in the “real” world
  3. If there is no important difference between a simulation and the “real” world, then one’s personal values are unaffected should one be living in a simulation
  4. A simulation is equivalent in all important ways to a sufficiently advanced virtual reality world created by humans
  5. Therefore one’s personal values will also be unaffected should one choose to live in a sufficiently advanced virtual reality world created by humans

At this point in the book, Chalmers has supported the first premise in the argument that I have put in green. This leads into the next section where Chalmers wants to support the second premise and show that, since we couldn’t know if we’re living in a simulation, then there is no important difference between simulated objects and “real” objects.

Reality Question

General Reality Question: is our world real or illusory?
Reality+ Reality Question (page 9): are virtual worlds real?

Chalmers wants to make the argument that a virtual object is as real as a “real” object. This is true, he argues, because if we say that the simulation objects that we may be interacting with (should it be true that we are in a simulation) are real, then a virtual object in a virtual reality that we create must also be real. So, his first task is to define what it even means for something to be real. He offers five definitions (pages 108-114):

Existence: reality is that which exists. My desk in front of me exists, and so it is real. Santa Claus doesn’t exist, and so Santa Claus isn’t real. This could perhaps be Berkeley’s Dictum: to be is to be perceived; or the scientific stance that something is real if it can be measured. Chalmers points out that there are issues with this, such as conditions extremely close to the big bang, or if we’re in a simulation then we might say that things outside the simulation are real even if we cannot ever see or measure them.

Does the existence definition of reality work for a simulation? Chalmers says it does. If you and I are in a perfect, permanent simulation looking at a tree, for instance, are we looking at something that exists? Chalmers says yes, even if it is grounded in digital processes.

Causal Power: something is real if and only if it can affect things or be affected by things. Physical objects by this criteria are obviously real. Santa Claus is not real, but stories about Santa Claus are real, since they can affect and be affected by things. Chalmers says that causal power could be a sufficient condition for reality, but not necessary, since something like an abstract number or a forgotten dream can be real yet have no causal power.

Do simulated objects have causal power? Chalmers says they do within the simulation, and therefore within the simulation they have causal power, thereby satisfying this definition of real.

Mind-Independence: Chalmers quotes Philip K. Dick saying that “reality is that which, when you stop believing in it, doesn’t go away” but offers the modification “reality is that which doesn’t depend on anyone’s mind for its existence.” But then Chalmers points out that all of our experiences occur in our minds, but we wouldn’t say that our experiences are not real.

Chalmers says that digital objects are mind-independent since they can exist even without my (or anyone) being there to experience them. Thus, simulations satisfy this definition of reality.

Non-Illusoriness: one must ask themselves “are things as they seem?” Chalmers says that something is real when it is the way it seems and not real when it isn’t. When it seems there is a ball in front of you and there is in fact a ball in front of you, then the ball is real because things are the way they seem. If you hallucinate a ball in front of you, but one isn’t actually in front of you, then it isn’t real because it is not the way it seems. Similarly, in virtual reality, it seems like there is a ball in front of you, but there actually isn’t, and so it is not the way it seems and therefore not real. But there is the issue that people can be wrong about the way things seem without having it screw up our lives. Indeed, if there is a hallucination that affects everyone, then that thing would be taken to be real. In other words, how does one determine that the way things seems is in fact the way things are?

Chalmers says that within a simulation everything is as it seems. He says “If my whole world is a simulation, flowers are ultimately digital objects, and Australia is ultimately digital, too, but this is no obstacle to the flowers blooming and to my being Australian.” In other words, when we are within the simulation, then the virtual objects are not illusions.

Genuineness: when someone says that something is real, then according to J. L. Austin a person must ask “a real what?” Think of someone gifting you a diamond necklace. If you ask “is this real?” the person gifting it to you could just say “of course it is, you’re holding it and looking at it right now!” But what you actually wanted to know was “is this a real diamond?” And so, according to Austin: “instead of asking whether something is real, ask whether it’s a real X.” Chalmers points out that the question of whether something is real is still a valid question: one might ask whether Santa Claus actually exists in a neutral sense, not whether Santa Claus is a real man or a real spirit or whatever.

Chalmers says that simulated objects are those real objects. He says “If I’ve lived my whole life in a simulation, every real flower I experience has been digital all along.” In other words, a flower in the simulation is a real flower to those within the simulation.

I can of course think of other ways that people could define reality that could perhaps not include simulated worlds. Maybe reality is that which is self-sustaining (does not require a mind or something external to sustain its existence, i.e. it can’t be “turned off” like a computer or pushed away like a thought). Or a conception of reality based on ontological substance from which it is made (certain substances are real while others are imaginary, illusory, or in some sense not real).

Or a perception of reality that says that reality must contain its own grounds: in a virtual world, everything is grounded in a world outside our own, but if it is a real reality, then the “brute facts” exist within our universe. As such, the simulation is a subset of reality: there is an asymmetry between the two. By this I mean that things outside the simulation can impact things inside the simulation in both a physical and social way, whereas in the other direction is only social (you can make a building inside the simulation from outside the simulation, but you cannot make a building outside the simulation from within the simulation; a meteor impact outside could destroy the simulation, but a meteor impact within could not destroy external reality). This is why, for instance, movies like The Matrix must invent the idea that dying inside the matrix also kills a person outside the matrix in order to maintain the stakes.

Or maybe reality is when causality is not based on some kind of pre-established harmony: in a virtual world, the effect that one thing has on another is grounded in some code that says “when A happens, then B” whereas reality actually requires some interaction within the world itself that makes A cause B. Perhaps we could think of it as some kind of physical, if not logical necessity that if A occurs, then B must occur; in a simulation, the code specifying how the interaction must appear could have been otherwise. This might be thought of as: reality is that which is built up from “below” (i.e. the laws of physics) instead of “above” (a programmer/simulator).

Or we might say that reality is that which is not designed or does not exist for a purpose. The theist could say that this then applies only to God (that God is ultimate reality); the atheist would say that this applies to whatever world exists at the “top” of all the nested simulations.

Or we could argue that reality is that which is non-fungible. In a simulation, an object can literally be copied and pasted over an over again such that they are numerically identical, but in reality such objects can ontologically only be types and tokens. Or reality is based on locality: something is real when it is existing only in one place (rather than as a physical body in the outside world and a digital body inside the simulation)

Or perhaps reality has some kind of sacred property, such that there is something metaphysically and intrinsically superior about the real than the simulated. This is likely what a theist would argue, that the real has been imbued (by God, or gods, or spirits of some kind) with some (perhaps divine) property that gives it an ontological or moral value above and beyond just what material it is made out of or what its function is.

All this aside, I don’t think Chalmers’ argument is too far off the mark. What he calls his position is Simulation Realism, which is pretty much exactly as it sounds: things that are simulated are as real as things that are not simulated. He supports this with what is often called the it-from-bit hypothesis (or, more accurately, the it-from-qubit hypothesis). He says (page 175):

If we can recover the mathematical structure of standard physics from digital physics, and if digital physics produces our observations, then digital physics realizes standard physics – that is, digital physics makes standard physics real. The bits of digital physics makes the quarks and electrons of standard physics real…

Importantly, if the simulation hypothesis is true, our observations of electrons and quarks are produced by certain patterns of bits. Every time we observe a photon, the observation is produced by a pattern of bits… If these patterns of bits underlie the mathematical structure of photons in our world, and also produce our observations, then those patterns of bits realize photons. By arranging the bits of digital physics in the right structure, the simulator has thereby created the entities – the its – of standard physics. In this way, the simulation hypothesis leads to the it-from-bit creation hypothesis.

When one theory realizes another, the entities in both theories are real. When atomic physics realizes molecular chemistry, for example, the molecules are real, and they’re made of atoms. Similarly, if digital physics realizes standard physics, then photons are real, and they’re made of its. If photons, quarks, and electrons are real, so are all the physical entities they make up: atoms, molecules, cells, rocks, organisms, buildings, planets, stars, galaxies. If the simulation hypothesis is correct, all of these entities are real.

Or somewhat more formally (page 176):

  1. Photons are whatever play the photon role [structuralism in physics].
  2. If we’re in a simulation, digital entities play the photon role [it-from-bit].
  3. So: If we’re in a simulation, photons are digital entities.

Chalmers then addresses some possible objections on pages 180 and 181 that I won’t go into here. He also addresses issues in part 7 of the book, which looks at things like how sense and reference would work between simulations and non-simulations and goes deeper into structuralism in physics.

For sense and reference Chalmers takes a view that incorporates the externalism of Hilary Putnam (see also his paper “The Meaning of Meaning“) and Saul Kripke that the referent of our words must be the cause of those words. For instance, when I say the word “water” I am referring to H2O, whereas someone on a different planet might use the utterance “water” to refer to something else, but when I talk to this other person and I say “water” I am still referring to H2O since it is the external existence of H2O and my experience of H2O that has caused me to have H2O as my referent for the world “water.” Chalmers, however, distinguishes between simulation-inclusive and simulation-exclusive uses of a word. When we’re playing Grand Theft Auto, then calling something in the game a car is simulation-inclusive. If we live in a simulation, though, then all our referents are simulation-inclusive.

Structuralism in physics is the metaphysical position that, broadly speaking, there is the math and there are the observations/measurements, and that’s it. Mathematical structures realize (make real) the physical universe that we experience. There need not be any deeper “substance” underlying it. This is what would allow for the it-from-bit hypothesis (or, perhaps, in-from-structure hypothesis) to work, because the math underlying everything would be the digital processes going on that realize (make real) the physical things that we experience. My own thinking on this subject is similar.

If we think about the inner circle, the green nodes can be thought of as properties like mass and charge, the edges (connecting lines) as the interactions between them (e.g. forces of nature); we then have the next circle up, which is the computational structure or data structure of the universe/simulation, where the innermost circle is realized (made real) by this computational structure; then the red on the outside could be the putative “it” that realizes the “bit” of the blue structure; the it-from-bit hypothesis would leave out the red outermost structure

To make the it-from-bit (it-from-structure) hypothesis work, Chalmers offers this argument (page 413, in part 7):

  1. Our physical theories are structural theories.
  2. If we’re in a Nonsim Universe, [then] our physical theories are true [at least close enough].
  3. Sim Universe has the same structure as Nonsim Universe [by virtue of being a perfect simulation, and therefore all observations would be indistinguishable between the two].
  4. So: If we’re in Sim Universe, [then] our physical theories are true.

Now lets again take stock of where we’ve come by looking at the broad outline of Chalmers’ argument from above.

  1. It is impossible to know for sure that one is not living in a simulation
  2. If it is impossible to know for sure that one is living in a simulation, then there is no important difference between living in a simulation and living in the “real” world
  3. If there is no important difference between a simulation and the “real” world, then one’s personal values are unaffected should one be living in a simulation
  4. A simulation is equivalent in all important ways to a sufficiently advanced virtual reality world created by humans
  5. Therefore one’s personal values will also be unaffected should one choose to live in a sufficiently advanced virtual reality world created by humans

What Chalmers has argued is that, given that it is possible (and from the Knowledge section, perhaps even probable) that everything we experience is already just digital information, then there is no important difference between simulated realities and non-simulated realities. This he uses to support his Simulation Realism thesis that says that virtual words are just as real as real worlds. This can then be used to argue that virtual worlds and virtual persons have equivalent or equal moral status to real things and real persons, which is covered in the Values Question.

Values Question

General Value Question: what is it to lead a good life?
Reality+ Value Question (page 9): can you lead a good life in a virtual world?

Part 6 is a fairly short part of Chalmers’ work, although packing in a great deal of issues that are covered at a mostly surface level. Yet for present and near-term practical purposes, it is these issues that are almost certainly the most important part of the book. It discusses how we ought to think about and orient ourselves toward virtual worlds. He does this through the lens of Value Theory, which, broadly speaking, covers three areas:

  1. Moral Value: right vs wrong
  2. Aesthetic Value: beauty vs ugliness
  3. Personal Value: what makes something better or worse for oneself? Or for other selves?

Chalmers mostly discusses 1 and 3. He does this first by thinking about the question: what is the good life? To answer this, one must ask what it is that gives something value. Is the value of something determined only by how much pleasure or pain it causes? This goes to Robert Nozick’s experience machine, from his 1974 book Anarchy, State, and Utopia.

What matters other than how people’s experiences feel “from the inside”? Suppose there were an experience machine that would give you any experience that you desired. Superduper neuropsychologists could stimulate your brain so that you would think and feel you were writing a great novel, or making a friend, or reading an interesting book. All the time you would be floating in a tank, with electrodes attached to your brain. Should you plug into this machine for life, preprogramming your life’s experiences? If you are worried about missing out on desirable experiences, we can suppose that business enterprises have researched thoroughly the lives of many others. You can pick and choose from their large library or smorgasbord of such experiences, selecting your life’s experiences for, say, the next two years. After two years have passed, you will have ten minutes or ten hours out of the tank, to select the experiences of your next two years. Of course, while in the tank you won’t know that you’re there; you’ll think it’s all actually happening. Others can also plug in to have the experiences they want, so there’s no need to stay unplugged to serve them. (Ignore problems such as who will service the machines if everyone plugs in.) Would you plug in? What else can matter to us, other than how our lives feel from the inside? Nor should you refrain because of the few moments of distress between the moment you’ve decided and the moment you’re plugged. What’s a few moments of distress compared to a lifetime of bliss (if that’s what you choose), and why feel any distress at all if your decision is the best one?

This is also similar to Larry Niven’s concept of wireheading, which is just hooking electrodes up to the pleasure centers of the brain and living ones entire life just lying in bed in a state of pure ecstasy. If one is committed to the idea that the pleasure principle is what gives things value, then one would be forced to conclude that the experience machine or wireheading would be good ways to live one’s life.

Chalmers differentiates virtual reality from the Nozickian experience machine. He says first the VR is not illusory, as discussed in the above section on reality, and is also real in that one can interact with real people within VR. Second, VR is not preprogrammed, and so people are free to make decisions that actually affect things. Third, being artificial doesn’t make something not real, just like how cities are artificial but no less real for it.

Chalmers then looks at other ways that people think are grounds for giving value to something. There is what is called Experientialism, where the fundamental objects of value are conscious experiences. This is different from just pleasure and pain because there can be conscious experiences outside just pleasure and pain that we find important. But then a person has to ask themselves: If all your friends and family were replaced by imposters, or even by duplicates from Derek Parfit’s transporter, would that matter to people? Or, if someone tricked you or forced you and everyone you know into a virtual reality indistinguishable from the real world, would that not be bad?

And so to address these issues there is the criterion of Desire Satisfaction, which says that a good life is one in which our desires are satisfied. Since we desire real, not artificial, experiences, then the counterexamples from above would be worse. For instance, we desire faithful partners and so even an unfaithful partner who perfectly hides infidelity is undesirable, whereas Experientialism would conclude that there is no difference between the two since it feels the same from the inside. But then the issue is that this can lead to selfishness, isolation, and atomization. What about the desires of those around us?

This then comes to the Social View of Value where value comes from connections to other people. Chalmers sites the African Ubuntu philosophy, which says that friendship, community, respect, and compassion all matter, essentially saying that “A person is a person through other people.” But this leads to other issues: could we then not say, for instance, that a hermit could lead a valuable life?

The final value-conferring theory that Chalmers examines is the Objective List View, which says that a basic source of value are items on a list, e.g. knowledge, friendship, fulfillment, etc. The big question then is what to put on the list (and why); if the list has a general principle, then isn’t that the ultimate source of value? And if there is no general principle, then isn’t the list somewhat ad hoc?

Chalmers then argues that, regardless of what value-conferring view we take, a sufficiently advanced virtual reality could accommodate any of them. And in the meantime, the VR we currently have could partially accommodate them if we lived part of our life within the virtual world. Thus, Chalmers concludes that there is no important difference between a virtual and a “real” world as far as whether a person can lead a good life. Someone who decides to live their entire life in a sufficiently advanced virtual reality could lead just as valuable of a life as someone who chooses not to live in such a world.

Next Chalmers wants to look at the moral dimension of virtual reality. He first examines consequentialism, deontology, and virtue ethics. Consequentialism is the school of thought in ethics that says that the morality of an act is grounded in what the consequences are. The paragon of this school is utilitarianism, which says that we must maximize pleasure and minimize suffering.

This invariably brings the conversation to Philippa Foot’s so-called Trolley Problem (actually so named by Judith Jarvis Thomson who brought the Trolley Problem across the pond from the U.K. to the United States). This asks, in different contexts, whether it is good to kill one person to save five. The trolley problem asks if someone would pull a lever to make a runaway trolley switch from a track where it will kill 5 people onto a track where it will kill just 1 person. Most people think it would be good to do this, saving more lives at the expense of one. But then there is the surgery issue: should a doctor kill a healthy patient in for a routine checkup and harvest the person’s organs in order to save the lives of multiple people who are on the transplant waiting list? In this case, even though the utilitarian calculus is the same – killing one to save many – most people’s intuition is that it is wrong.

Then there is deontology, which is often thought of as a sort of formalistic morality, where a person is supposed to follow some rules to be moral. This is an oversimplification, but when it comes to employing the philosophy that is what it comes down to in a practical sense. This is exemplified by Immanuel Kant’s categorical imperative: “I ought never to conduct myself except so that I could also will that my maxim become a universal law” or even more strongly “so act as if the maxim of your action were to become through your will a universal law of nature” (both quotes from The Groundwork of the Metaphysics of Morals). In other words, don’t do to others what you wouldn’t want done to yourself.

There are plenty of ways that this kind of moral philosophy goes against our intuitions. For example, if we take “do not lie” as one of our maxims, then what about lying to someone who wants to hurt you or someone else? For instance, if someone during WWII was hiding Jews in their attic and a Gestapo officer came by and asked if there were Jews hiding in the attic, would the person have a moral duty to tell the Gestapo officer the truth? This is why, for instance, Elizabeth Anscombe rejects conseqentialism and deontology for virtue ethics.

Chalmers then wants to apply these ethical philosophies to virtual people. Do we have any moral responsibility toward virtual people? Chalmers points out that philosopher and animal rights activist Peter Singer (the author of Animal Liberation) says that sentience is what gives something moral status. “Sentience” here is defined as the capacity to experience pain or pleasure, in which “only a consciousness of positive and negative affective states is required for moral status.”

This makes some intuitive sense. For instance: what is the moral status of a philosophical zombie? A philosophical zombie is not like the human-flesh-eating shambling hordes of movies and video games. A philosophical zombie is exactly like a person in every way, except they lack consciousness – the lights are on, but nobody is home. A philosophical zombie would, however, act as if it is conscious. It could talk and interact and display emotions as well as anyone you know, and if you asked it, it would tell you that it is conscious, even though it is not. Yet, if you caused it harm, it would react as any normal human would when they are in pain, but it does not have any sort of inner experience of pain.

Most people would intuitively think that a conscious person has moral status while a philosophical zombie does not. Indeed, most people would probably think that if a runaway trolley was heading toward 5 philosophical zombies that it would be wrong to pull the lever and kill 1 actual conscious human instead.

But then it starts getting tricky when we wonder: would you kill a philosophical zombie just to shave off 10 minutes on your commute? Would it be better to kill a single real world chicken or an entire world of billions of philosophical zombies?

Chalmers disagrees with Singer that sentience, so defined, is all that matters. Chalmers says that other aspects of consciousness are important, too. He proposes what he calls a philosophical Vulcan (from Star Trek) which, unlike the philosophical zombie, has consciousness, but lacks positive or negative affective states (it doesn’t just suppress emotions, it has no emotions). Would we say that such beings have no moral status since they don’t care about pain and pleasure, despite having a conscious experience of the world? Would it be better to have 5 philosophical Vulcans die than 1 actual person?

Chalmers then asks the question: should someone cause the deaths of 5 simulated people to save 1 non-simulated person? And if you answer yes, then, if we are in fact living in a simulation, would you agree that it is okay for our simulators to cause the deaths of 5 of us in order to save 1 of them? This goes back to Chalmers’ thesis of Simulation Realism – if we can’t know we’re in a simulation, then there is nothing important different about a simulated person and a non-simulated person, so wouldn’t a simulated person have the same moral status as a non-simulated person?

This obviously gets into whether or not the people simulated in any simulated words that we might create are conscious (which Chalmers talks about in part 5, which I haven’t discussed here), but we could assume for the sake of argument that the simulated people discussed above are conscious. What moral responsibility do we have toward (conscious) simulated people that we have created? Is it even moral to create simulated people? If we do create simulated people, then are we obliged to simulate them in the best simulated world possible? Would it be immoral to simulate people in, say, a world only slightly worse than paradise (assuming it is possible we could have simulated paradise instead)? If pleasure is good, then do we have a moral obligation to simulate as many beings as possible experiencing as much pleasure as possible?

After touching on those issues, Chalmers then goes on to briefly wonder about both ethics and politics as it pertains to virtual worlds (beginning at page 352). I’ll examine the two questions in turn.

Ethically: How should users act in a virtual world? What’s the difference between right and wrong in a virtual world?

Chalmers talks about the moral status of a person’s online avatar (see My avatar, my self: Virtual harm and attachment and “Virtual Harm and Attachment” both by Jessica Wolfendale), the ethics of killing, raping, and pedophilia in a virtual world (see “The gamer’s dilemma: An analysis of the arguments for the moral distinction between virtual murder and virtual paedophilia” by Morgan Luck). In this latter case, he mentions a 1982 game called Custer’s Revenge where the point of the game is to try sexually assaulting a Native American woman, and a 2002 game called Ethnic Cleansing where you play as a white supremacist trying to kill non-white people. I mentioned when I touched on this subject the instance of a 2015 video game called Hatred, where the point is just to kill as many innocent people as possible.

The idea behind all of this is that we often times have a different moral view of how people treat virtual characters compared to non-virtual characters. Indeed, a good deal of all video games have you going around killing virtual people. Yet, people will still feel a sense of moral revulsion at a video game where you are trying to sexually assault virtual people, or where you can engage in pedophilia with virtual children. Chalmers points out that this might have to do with a sort of virtue ethics that says that the kind of person who would take pleasure in such virtual sex crimes must be a morally disturbed person. But then why is it okay for people to take pleasure from killing virtual people? Another argument is that such virtual acts might lead people to perpetrate them in the real world, but violence in video games doesn’t seem to lead to increased violent acts in the real world. Besides, an argument could be made that allowing pedophiles to indulge their urges with non-conscious virtual people that this might prevent them from doing it in the real world. The point is, it is a complex issue without easy answers.

Chalmers also brings up Grand Theft Auto, where he says that murdering, stealing, and misogyny is the point of the game. I would disagree on this account. A person could go through much of GTA without perpetrating many of those crimes; indeed, in some of the newer ones, a person could just go do Yoga in the game if they wanted to. The violence and misogyny are not the point of the game, it is just the way the vast majority of people choose to play it. It reminds me of Westworld (which, by the way, I found it curious that Chalmers did not bring this up here, given the show’s salience to the issue and his willingness to connect things to pop culture throughout the book), where a person could go to the theme park and just act civil, but many people go in order to indulge in crime and violence.

Also discussed is virtual theft (see “The Puzzle of Virtual Theft” by Nathan Wildman and Neil McDonnell). Should it be a crime to steal people’s virtual items? I’ve played enough online video games, like Diablo 2, Final Fantasy XI, and World of Warcraft, to know that stealing virtual items is not all that uncommon. Should those people have to face criminal charges, just like someone who steals a physical item from your house? Getting those items took time and effort (and sometimes money, for instance with subscription fees or if the victim was the kind of player who would buy video game items with real world money). As such, it seems like, even though they are virtual objects, a person still has an ownership claim over them (perhaps both moral and legal).

Chalmers then turns to politics.

Politically: What are the ethical and political constraints on the creators of a virtual world? How should a virtual world be governed? What is justice in a virtual world?

Right now there are virtual worlds like Second Life, and in the future such virtual worlds will (presumably, and barring any disasters) will only become more and more immersive. Perhaps even to the point where people might want to live a significant majority (or even all) of their life within the virtual world. How should a virtual world be governed? How should resources be allocated?

Chalmers cites John Rawls’s book A Theory of Justice as a theory that could actually be used in virtual worlds. Rawls proposed that if people were able to get together in some kind of pre-life limbo (original position) before they became humans, and without knowing which human they would become, that they would want resources distributed equally. Essentially, people should want to live in a world where, had they been born anywhere else in the world, they could still have their needs met. In a virtual world, such a theory could actually be put into practice by having everyone who goes into the virtual world be in possession of an equitable share of the resources.

Chalmers points out that a virtual world could even be a post-scarcity world. Within a virtual world people could just have whatever they wanted as soon as they wanted it (what Chalmers calls virtual abundance). Rawls says that scarcity is a condition of justice, and so without scarcity, principles of justice do not apply. My question is, who would want to live in such a world? I’m reminded of a rare sword in FFXI called the Ridill. The reason it felt so good to finally get one is because of how rare (scarce) it was. Had a person beginning the game just started with every item in the game already, what point would there be in playing it? Same goes for, say primal ancient items in Diablo 3. Could such virtual abundance lead to a widespread sense of ennui? And if nobody needs to help the poor anymore, can anyone still be good person?

Even if we have all the resources we need, Chalmers points out, there are other kinds of inequalities than just resource distribution. For instance, there is the relational view of equality (see “What is the Point of Equality?” by Elizabeth S. Anderson): the unequal distribution of power, such as in domination and oppression, which Elizabeth Anderson argues matter more than resource distribution. And of course there are power differentials within the virtual world and between those in the virtual world and those outside (e.g. does the corporation who makes the virtual world have dictatorial power over what goes on within it?).

So, lets take one last look at the overall argument of the book:

  1. It is impossible to know for sure that one is not living in a simulation
  2. If it is impossible to know for sure that one is living in a simulation, then there is no important difference between living in a simulation and living in the “real” world
  3. If there is no important difference between a simulation and the “real” world, then one’s personal values are unaffected should one be living in a simulation
  4. A simulation is equivalent in all important ways to a sufficiently advanced virtual reality world created by humans
  5. Therefore one’s personal values will also be unaffected should one choose to live in a sufficiently advanced virtual reality world created by humans

Chalmers has at this point made the case for the green premises above, which means that, if he is successful, then we must to accept the conclusion (number 5 above). That conclusion is his Simulation Realism thesis, that simulated worlds are just as real as the non-simulated world.


As you can see, this review is already quite long. And that is with leaving out parts 4, 5, and the vast majority of 7. And I haven’t even covered everything discussed in the parts I did review here. There is a lot of subject material in this book. Much of it doesn’t get much more treatment than even I gave it here in this review, particularly part 6 (the Values Question), where a lot of the things I discussed above he treats almost as briefly as I did here. Questions are raised and complications pointed out before moving on to the next thing. That’s not to say it isn’t interesting or worth reading (I didn’t cover every topic here that Chalmers discussed in part 6). But I do get the impression that Chalmers is more interested in most of the other topics, which get much more in-depth treatment (though by no means exhaustive, which to be fair would be impossible in a book that is only 544 pages long).

In just these sections, however, I think David J. Chalmers makes a very compelling (though I don’t know if it is successfully convincing) case for his Simulation Realism thesis. Along the way he brings up a lot of delicious food for thought. Well-written, easy to read, and brimming with interesting ideas to consider, Reality+ is a must read for those interested in questions of philosophy and for those fascinated by virtual reality and the future of technology.