Notes on a Cosmology - Part 15, Poor Man's Matrix

In the movie, The Matrix, humanity has unwittingly become trapped in a computer simulation by an enemy that is simply referred to as "the Machines". The premise of the movie is that this simulation is indistinguishable from the world you and I inhabit. Agent Smith - an agent of the Machines - explains the situation to a protagonist:

The first Matrix was designed to be a perfect human world where none suffered, where everyone would be happy. It was a disaster. No one would accept the program. Entire crops [of people] were lost. Some believed that we [the Machines] lacked the programming language to describe your perfect world. But I believe that, as a species, human beings define their reality through misery and suffering. The perfect world was a dream that your primitive cerebrum kept trying to wake up from. Which is why the Matrix was redesigned to this, the peak of your civilization.

The Matrix is often cited as a dramatization of the philosophical argument called the brain-in-a-vat argument. To stop here, however, is to miss the wider philosophical implications of the movie. First, notice that there is no observational difference between "reality" and "simulation" for an inhabitant of the Matrix. The simulation just is reality. Second, note that the parameters of the simulation are completely reconfigurable - the first Matrix was programmed to be free of misery and suffering. But after the Machines lost entire crops of human beings, they reverted to a simulation that is a replica of human civilization at its peak (circa the end of the 20th century), complete with all the misery and suffering that come with it.

In order to explore these ideas in more depth, we will construct a thought-experimental Matrix of our own devising, using only known technologies. I will call this simulation the Poor Man's Matrix or PMM. The primary goal of this post is to expose and disintegrate many of our unstated presuppositions about the way things have to be. In place of these presuppositions, we will explore the Simulation Hypothesis as a guide to the wider structure of causality.

Philosopher David Chalmers wrote an article titled, Facing Up to the Problem of Consciousness, in which he argued that the experience of conscious awareness is a "hard problem" distinct from all the other problems of consciousness - which he terms "easy problems" - that occupy most of the study of neurologists and philosophers of consciousness. Chalmers's hard problem thesis spurred decades of debate over the true nature of consciousness and whether conscious awareness can be explained scientifically. But Chalmers's essential point stands, unchanged: no matter what you think about what consciousness is or how it originated, no one can give a functional explanation of conscious awareness because conscious awareness is precisely whatever is left once you have explained all the other functions of consciousness. I have raised the problem of consciousness only to point out its difficulty. For the purposes of this series, we will leave aside the problem of consciousness entirely, and concentrate on the reality of acting and sensing without respect to the origin or nature of consciousness itself.

Up to this point in the series, we have not touched the question of why anybody has seriously proposed the idea that the Universe is, at root, a simulation. There are a broad spectrum of possible answers to this question. Nick Bostrom's Simulation Argument (SA) is based on an idea that our descendants are simulating us, a bit like The Matrix, but with different motives than the Machines and with the added speculation that consciousness itself arises through the process of computation. But digital physics theories predate Bostrom's SA and, whatever the merits of the SA, the older motivations for wondering if we might be in a simulation are in many ways more concrete.

The cosmic paranoia that frequently attends simulation theories in science-fiction is the fear that - if we are in a simulation - we are all being duped and exploited in a terrible way by the Simulator(s). Perhaps we are being groomed for eventual slaughter (most vampire fiction). Perhaps we are being parasitically exploited (They Live). Perhaps we are being experimented upon like lab rats (Dark City). In the movie The Truman Show, the eponymous Truman discovers that he is the subject of a grand hoax, that what he thought was his "real life" is actually a reality show populated by actors and being watched by millions of paying customers. Most of the movie follows the psychological breakdown Truman has as he discovers the horror and futility of his life trapped inside a delusion. The movie illustrates an important point: even though a computer simulation would be a supreme instrument of deception and enslavement, a condition of delusion is not essentially about technology. Despite whatever technological aids were employed, Truman was fooled by people, not machines.

A quick survey of human history will show that the petty motives of human exploitation, cruelty and subjugation have always been overridden by a more powerful principle: physical scarcity. The rise of socialism and communism in the mid-20th century was the cause of immense fear on the part of the relatively freer, capitalist West. After the collapse of the Berlin wall, it turned out that almost all of the fears of the West had been overblown. As the Iron Curtain fell, the West discovered that the Soviet Union and its satellites had been operating on the verge of collapse for decades, relying on hopelessly obsolete and ill-maintained military equipment, built on an economy staffed by workers lacking many basic needs. Communism, it turned out, had been a paper tiger all along. Economic theory shows that the collapse of the Soviet Union is typical of the larger pattern of central-planning of economic resources motivated by aggressive and exploitative tyrannies, a pattern that extends back to the earliest records of human history. The point is this: if our Simulator(s) are subject to any kind of scarcity at all, then they are subject to the discipline of economization and however petty and cruel their motives might be, they can never shake the law of supply and demand, they can never shake the chains of scarcity.

I have spent some time considering the possible motives of our hypothetical Simulator(s) for the purpose of banishing the topic from here on. I am not very interested in whether we are simulated by telic beings nor what their motives are, supposing they exist. Rather, I want to investigate the reasons for speculating that the Universe is, at root, a simulation that have nothing to do with telic forces, at least, nothing like the telic forces of human-like beings[1]. The PMM thought-experiment is not about the Simulator(s), it's about the simulation itself and about deriving the laws that would have to impinge on the simulation - whatever its particular mechanism - as a result of economization, in particular, economization of information.[2]

But why should we believe we are in a simulation at all? Why even talk about it? The universal prior probability distribution solves the problem of why there is order rather than disorder. This makes many of the traditional cosmological problems simply go away - for example, there is no question of how the Universe began because the Universe had no beginning[3]. But it leaves us with a vastly more complicated cosmos than the smooth, mechanical universe of Newton, Lagrange and Einstein. While the quantum world is very strange, it's still not as strange as the world that the universal prior predicts. As we progress through the series, it will become clearer what we mean by this.

The following figure depicts the high-level view of what we mean by a Poor Man's Matrix, that is, a massive, multi-agent simulation:

The VR Environment is the immersive environment in which each individual agent is interacting with the PMM. The Private State is that part of the simulation environment that is accessible only to the individual agent. The Shared World State is "the Matrix", it is the shared virtual environment in which agents can interact with one another.

Using technology that exists today, it is already possible to build extremely immersive virtual-reality environments. The goal of our VR Environment is to attain a sufficient level of suspension-of-disbelief that engrossment occurs - a temporary state of near-amnesia regarding the fact that the virtual environment is merely virtual. Such a state is not actually that difficult to achieve in respect to the limbic system, as evidenced by the elevated heart-rate, breathing and other stress indicators exhibited by someone watching a particularly suspenseful and engrossing movie, and so on. An argument can be made that the primary component for suspension-of-disbelief is psychological and that technological props are entirely secondary but this argument is beyond the scope of this series. For our purposes, we simply posit that the VR Environment is very engrossing. We posit an isolation-tank with IV hydration and bodily functions handled automatically. The visual space is full 3-D, high-resolution rendering. The audio space is high-fidelity surround-sound, stereo-isolated. In addition, we posit pressure simulation, force-feedback and gravity effects (possibly by lifting from the isolation tank). We posit scent and other simulated particulate effects. None of this is sci-fi, although too expensive to be economical, at present.

The diagram above is not meant to be the basis for a mathematical analysis but a mathematical analysis of the requirements on the systems is possible - we can calculate the bandwidths and latencies required to maintain some desired level of Quality of Service. The more important construct in the model is the division between Private State and Shared World State. From a psychological point-of-view, we may think of these as corresponding to the "subjective" and "objective" aspects of being. In addition, we are not positing a universal simulation, ala The Matrix. Rather, we are positing simulation as a money-making business, like massively multiplayer online games (MMOGs) - simulation-builders must compete for customers. Thus, the simulation-builder does not unilaterally dictate terms to its customers. Rather, the simulation-builder must attract customers by providing the services they demand.[4]

Modern MMOG's contain a wealth of indicators about the states of affairs that would obtain in a world of pervasive VR engagement. Let's examine some of the modes of user engagement with virtual worlds.

1. No one-to-one mapping between consciousness and body(ies). Rules vary in specific cases but, in an MMOG, it is technologically possible that a character may be operated by more than one person and one person may operate more than one character.
2. AI-based autonomous agents or non-player characters (NPCs). NPCs flesh out a game and add lifelikeness to the game by operating "extras", characters that make the world more engrossing but are not interesting enough to be played by human operators. In some MMOGs, an operated character continues to operate via in-game AI while the human operator is logged out. This is because world continuity is valuable in its own right.
3. Both first-person and third-person forms of engagement exist. In some games there are individually operated characters; in other games, there are only collectively operated character groups (e.g. armies, fleets).
4. Operating a character in an MMOG can be viewed as a form of mind-extension. This can be viewed as a form of distributed consciousness, that is, simultaneous, real-time interaction with multiple environments.
5. In many games, the human operator has some kind of "heads-up display" or "dashboard" that collects and displays information from distributed sources throughout the game world. This can be viewed as a form of mass-filtering of information. If users are able to build and deploy their own filters into the game environment, this kind of filter would be indistinguishable from real-life information search.
6. In some games, in-game alerts allow the human operator to switch attention nearly instantaneously from one part of the game world to another, allowing the operator to "fly" to locations of interest based on preset conditions.
7. The movie Avatar explores the idea of remote consciousness in an alien world. Imagine a game with the possibility for specific in-game personas to attain fame or fortune. This would give rise to persona-scarcity and the demand for "character-sharing" through remote-consciousness would arise.
8. The HBO series Westworld explores the concept of a theme-park populated by AI robots of such quality as to be all but indistinguishable from humans. If we imagine occupants inhabiting a virtual environment for very long stretches of time, ordinary biological replication would eventually lose interest. Artificial replication of the self would eventually supersede the natural reproductive urge - you would become more interested in replicating your "image" than in physical replication of your genes. This is reflected in Westworld's plot device of the reveries.
9. In a virtual environment, human, non-human and inanimate entities are all subject to the same mechanisms of control. A virtual train need not be operated by an engineer character, per se, even if the train itself is under control of a human operator. In general, human action between live agents is freed of the shackles of ordinary physics. If I find animals in a virtual environment more interesting than humans in that environment, then I may choose to play the role of an "animal spirit", so to speak - and so on for all levels of being.
10. The last two points combine to create surprising results - spooks and frights may be regarded as distributed conditioning of other live agents in a virtual environment. This doesn't have to do with hacking the game, so much as hacking the minds of other players in the game by utilizing features of the game in unexpected ways to induce surprise, fear, anger or other strong emotional reactions calculated to elicit mistakes or other decision errors from other players. Such a world can be thought of as a demon-haunted world.
11. It will be possible to commercialize activity within virtual environments - there is no reason that VR must be utilized purely for entertainment purposes. Human computation is an often overlooked form of computation and, in a VR environment, this computational resource should be regarded as inherent to the environment itself. Suppose I operated a successful VR environment with 100 million fully-immersed human brains within it. These 100 million brains represent a computational resource that puts all existing silicon-based computers on the planet, combined, to shame (on certain problem domains). By translating some of the choices that these brains make from one domain to another, I can harness some of that highly valuable computational capacity. In an early script of The Matrix, humans were being harvested by the Machines for the computational power of their brains, not for physical energy.
12. Human game administrators that provably never log into the VR environment itself have a unique value, especially for commercialized VR worlds
13. Reconfigurable computing makes hardware-level alterations of a VR environment possible from within that environment. In short, there is no need for adversarial agents to "enter" and "exit" the simulation, as depicted in the movie The Matrix.
14. The previous points illustrates the crucial importance of formal languages that are lower on the Chomsky hierarchy than Turing machines - provable limits on action are essential to build robust simulations at-scale. Without provable limits on agents (both human and AI), there is never a way to be sure someone will not sabotage the VR world.
15. The fundamental scarcity in a simulation is not energy, material or time - rather the fundamental scarcity is attention and instrumentality. Attention is the focus of the mind in the simulation at any point. Instrumentality is the mind's intervention into or alteration of the simulation.
16. In order to be robust, a VR environment will have to be completely transparent (public record of events). Thus, the world state is always 100% visible to anyone who cares to examine the record. Nevertheless, encrypted channels between VR environments using private state are still possible. Also, steganography and self-authorizing languages are possible.

The primary problem facing a VR world at-scale is cheating. The Bitcoin network provides a living social-experiment showing the prerequisites for building a bullet-proof, multi-agent cooperative computation with large amounts of capital at stake. Bitcoin has eschewed the central server and taken the approach of allowing anyone to hack the network to their heart's delight, with the view that what doesn't kill the network only makes it stronger. As long as more than 50% of the network remains honest, the network is resilient to any sort of attack short of world-wide disruption of the Internet caused by a nuclear event, EMP event or natural disasters[5].

Bitcoin's success shows the crucial importance of choice in building virtual environments. If choice is constrained, then the users are trapped and must accept whatever sort of environment is imposed upon them. But if users are free to leave one VR world and switch to another, the VR environment builders are in a state of market competition and the entire construct of VR world-building comes under the discipline of market competition.

I am conditioned by my life experiences to attribute to you the same conscious experience of the world that I have. But my life experiences are really Private State, as are your conscious experiences. I only know what you report to me about your life experiences, through the Shared World State. For example, suppose you are color-blind but you have a heads-up display that shows you a textual decoding of the color of any object you happen to be looking at. You are also very insecure about your color-blindness and do not want anyone to know that you are color-blind. Whenever you are asked questions like, "What do you think of this color?", you are always able to simulate color-perception - "The blue looks amazing with that tangerine." In reality, you do not experience color vision but there is no way for anybody to know this from the Shared World State. How do I know that this aspect of a VR world is not true of the world I am inhabiting, even if it is not a simulation? In fact, I have no way to know that your conscious experience of the world is the same as mine. Your conscious experience of the world could be arbitrarily far from mine, up to and including you being a completely "lights-out" Chalmers zombie. Thus, all the modes of engagement that we know are possible in a VR environment are also possible in the real world. Not only are they possible but I have no more reason to suppose that your conscious experience of the world is the same as mine than I have to suppose otherwise. In short, it's a coin-toss from the Bayesian perspective even though the idea of symmetric human consciousness is usually treated as a default hypothesis requiring "extraordinary evidence" to refute.

In closing, I will leave you with an excerpt from the 2016 Isaac Asimov Memorial Debate: Is the Universe a Simulation? In the debate, moderated by Neil deGrasse Tyson, James Gates makes some interesting remarks at 00:27:25 and following, where he says, "If the simulation hypothesis is valid, then we open the door to eternal life and resurrection and things that formerly have been discussed in the realm of religion."

Next: Part 16, The Quantum Monad

---

1. We will be encountering telos again, but not human-like telos

2. Note that we are not trying to outsmart a Cartesian demon. If we are trapped in a simulation (or, all the same, a physical universe) that is run in every particular by a deceptive entity, we have no hope of "breaking out" - thus, there's no point in trying and there's no point in wasting one minute talking about it. Rather, we are focused on the limits of any simulation under the condition of economization of information, without respect to a being - whether good or evil - that is "controlling" the simulation.

3. The set of all natural numbers is much simpler than the finite set of all numbers 1,2,3,..N where N is a randomly drawn finite number of size, say, 500 digits. The reason for this is simple: an axiomatic theory of sets sufficient to describe the set of natural numbers can be described in far fewer than 1,660 bits (the number of bits in a randomly drawn natural number of size 500 digits). If the Universe is of a finite age, in order for this age to be a simpler hypothesis than an ageless Universe, it must be an age whose descriptive complexity is lower than the number of bits required to describe the set of all natural numbers in order to be a more probable hypothesis under the universal prior probability. In either case, our cosmological theory would still need the set of all natural numbers, so an ageless Universe is the unconditionally simpler hypothesis.

4. This condition is not arbitrary. I will argue in upcoming posts that agency is an integral part of the cosmos and a universal simulation - imposed according to the arbitrary whims of a central simulation-builder - is exempt from the laws of action that limit an agent under the condition of scarcity. While a universal simulation imposed by a unilateral simulation-builder with unknown motives (or known ill-motives) is a powerful plot-device for invoking Lovecraft-like feelings of cosmic horror, it is not very interesting from a theoretical point-of-view.

5. This is a slight over-simplification but it is the gist of Bitcoin's security