The simulation hypothesi.., p.11
The Simulation Hypothesis, page 11
Altered Carbon and Downloading Consciousness
Obviously, a digital copy of text or a movie is still considered the same text or movie, but as we start to talk about conscious entities, we can no longer be so sure.
Is a digital copy of a person’s neurons still considered that person? Is consciousness being downloaded or is it only a copy of that consciousness at a single point in time?
Once again, we turn to science fiction because it helps to illustrate the conundrum. In his 2002 science fiction novel, Altered Carbon, Richard Morgan presents an intriguing vision of downloadable digital consciousness. Altered Carbon was made into a TV series for Netflix in 2018, allowing us to visualize how this process of “downloading” might work.
In a far-in-the-future world, most people have a device called a cortical stack, or just “stack” for short, which is stored in the spinal column. This device is able to download the consciousness of a person at any time. When a character's body is “killed,” the stack can be removed and placed into another body, and the person can carry on from that moment in time.
These stacks are like the downloadable consciousness that Kurzweil and others predict will happen. When they are placed in a new body, the bodies are called “sleeves,” and a stack is re-sleeved. This results in quite a few interesting scenarios that are relevant to both the plot of the book and our discussion here.
The only way a person can truly die is to either decide not be re-sleeved or to have their stack destroyed so that it cannot be re-sleeved. Because consciousness can be stored in these devices (thus it is information), consciousness can also be transmitted to a backup site via satellite. This means that if a person's sleeve and their stack are destroyed, they can “restore from backup,” though they will lose any memories that were accumulated after the last backup.
Though the plot itself is very intricate, the interesting thing from the point of view of the simulation hypothesis is its very clear depiction of consciousness as information that can be stored in a physical device, which can then be used to restore the person in a new body.
In another interesting depiction, even though faster-than-light travel is not possible in this futuristic world, it is possible to “beam” someone’s stack information at a speed faster than light. This means that if there is a ready sleeve on another planet or in a space station, one can take their consciousness and transmit it to that sleeve and suddenly they would wake up inside the new body.
So how is the “personality” loaded into a cortical stack and sleeved or re-sleeved? Even in this science-fiction future, the process is mysterious! It was developed based on the technology of a long-vanished alien race!
The Upshot: Consciousness as Information
We have seen how the history of AI and gaming is intertwined, and just as my first exposure to AI was in building a Tic Tac Toe game, many of the founding figures of modern computer science, including Claude Shannon and Alan Turing, devised games as a way to test and develop artificial intelligence. We saw that recently, algorithms have been able to beat professional players in traditional games like Chess and Go, and even beat professional players of eSports, which requires much more understanding of a virtual 3D environment.
We have also seen that AI programs hold the promise of continuing someone’s expertise alive after they are gone. In fact, before the recent wave of machine learning, the state of the art in AI was an expert system, which codified in rules knowledge that an “expert” might have about accomplishing a task. The idea was that the AI could far outlive the person.
This science-fiction notion of digital immortality is actually supported in the spiritual and religious traditions that believe our consciousness lives forever in an afterlife outside of the physical world. In Eastern traditions, consciousness is transferred to multiple bodies in a series of incarnations. This is a direct analogy to a single player who plays multiple “roles” or “characters” in a role-playing game. We’ll talk more about models of reincarnation in Part III, but if the same consciousness inhabits multiple bodies, then it’s possible that these characters have tasks and quests that are based on what happened in “previous lives.”
For our discussion here, if consciousness is digital information, then we should be able to not just preserve consciousness outside of the simulation but also create artificial consciousness as part of a simulation where we can’t distinguish between player characters and non-player characters.
By building artificial beings (Stage 9), we have inadvertently found a way to represent consciousness itself as digital information that can be downloaded into a player (Stage 10). There are metaphysical as well as technical and philosophical questions that go along with this, which will need to be resolved as we travel through these stages on the road to the simulation point.
If a civilization has mastered these two stages, then it has pretty much reached the point of being able to create a Matrix-like reality, and has reached the simulation point. Next, we will explore some of the implications of reaching this point, and go through Nick Bostrom’s original argument that we may be living in a simulation all along as simulated entities.
Chapter 4
Stage 11: The Simulation Point, Ancestor Simulations and Beyond
At some point after completing the previous stages, a civilization should have reached the simulation point—the point at which it can create simulations that are virtually indistinguishable from a base physical reality. The Great Simulation is a video game that is so real because it is based on incredibly sophisticated models and rendering techniques that are beamed directly into the mind of the players, and the actions of artificially generated consciousness are indistinguishable from real players.
In this chapter, we explore what the simulation point looks like and then revisit some important implications of a civilization reaching this point. In doing so, we’ll look in detail Nick Bostrom’s Simulation Argument, which helped launch a series of debates online about whether or not we are living inside a simulation.
We’ll delve into details of “ancestor simulations,” which is the term Bostrom uses for such simulations, and what it means for the civilizations that are able to produce them. Bostrom’s conclusion is that if any society ever makes it to the simulation point, then it is more likely that we are already inside a simulation!
Stage 11: Reaching the Simulation Point
There are other various technical hurdles that would have to be overcome on the path to the simulation point which we did not call out as explicit stages, but these are mostly implementation challenges and not fundamental technology development. For example, the ability to support billions of individual players (a number that no MMORPG has been able to support thus far), with significant amounts of information for each. And that’s only on one planet—given the amount of other Earth-like planets in our galaxy alone, this number of players (NPCs and PCs) might jump into the trillions. There is also the storage capability needed to keep track of this many “players” and the compression algorithms to allow for storage of this much data on some “cloud server.”
Table 1 below shows some of the criteria that would define the simulation point, using technology that we have speculated about but has not been developed yet. While the concepts of early video games and MMORPGs are quite useful, some of the key criteria are not yet met by our civilization. Nevertheless, just as 100 years ago no one could have predicted today’s computers and video games, it is very likely that we can get to the simulation point over the next 100 years, a small time-frame by cosmic time standards.
Table Defining the Simulation Point —How Do We Know We Are at the Simulation Point?
Criteria
Notes
Simulating a large game world
Procedurally generated worlds
3D rendering that is photorealistic
Combines 3D models and textures, so improvements on existing technology
Ability to beam the signal for the simulated world into our eyes or our minds
Yet to be invented technologies—light-field rendering and mind broadcast
Ability to incorporate feeling (haptics)
Some of these technologies exist, but incorporating them into mind interfaces does not
Ability to take responses from our mind and react in the world
Yet to be invented
Have very large number of online players
Scaling beyond what we can do today, but just an extension of our existing technology
Storing data about players and characters in the simulation but outside of the rendered world
A very large amount of data, requires improvements on existing technology
Implanted memories
Ability to change history, not yet invented
NPCs
Artificial characters being simulated, improvements on existing technology
Thus far, we have talked about the simulation point as a pivotal moment in time and have shown the roadmap to a technological civilization, like our own, that would develop video game technology to the point of being able to build the Great Simulation.
What are Ancestor Simulations?
In his groundbreaking 2003 paper, "Are You Living in a Computer Simulation?" Oxford philosopher Nick Bostrom made an argument that didn’t have to do specifically with video games or AI or science fiction, though he referenced each of these in the original paper.
The argument that he makes is different; he refers to simulations as “ancestor simulations” developed by civilizations that have advanced to a certain point (which would be analogous to what I am calling the simulation point in this book). Bostrom calls civilizations that have reached this point “post-human” civilizations because they are capable of making ancestor simulations on their powerful computers.
One thing that later generations might do with their super-powerful computers is run detailed simulations of their forebears or of people like their forebears. Because their computers would be so powerful, they could run a great many such simulations … Then it could be the case that the vast majority of minds like ours do not belong to the original race but rather to people simulated by the advanced descendants of an original race.20
Although Bostrom supports the idea that we may be inside a simulation, he seems to view the simulations as running on some kind of advanced post-human computer system (if it can still be called a computer system) without much interaction between the simulated beings and the real beings in the base reality. Bostrom doesn’t get too far into the details of what is actually going on inside these simulations. But in his model, all the ancestors inside the ancestor simulations are artificial consciousness—or simulated beings.
Bostrom’s Simulation Argument
Bostrom’s basic argument, which he referred to as the “Simulation Argument,” if you boil it down, is that either mankind (or any species) will reach the point where it has the technology to conduct ancestor simulations or it won’t. If any species is ever able to make it to this point, Bostrom concludes that we are vastly more likely to be inside a simulation than not.
Let’s drill down into the three possibilities that Bostrom brings up about any civilization ever making it to the simulation point:
A civilization never makes it to this point. Thus, ancestor simulations are not possible. Why wouldn’t a civilization make it to this point? It could destroy itself or it could just not develop computer technology in the ways that we are doing. Given the road to the simulation point outlined earlier, some aspects of this road are progressing rapidly while others may take longer.
A civilization makes it to this point, but ancestor simulations are banned or not allowed. Why would a civilization ban ancestor simulations? This is an open question, though he gives a few ideas.
A civilization makes it to this point, and it creates many ancestor simulations. This is seen as the most likely scenario, because most technologies developed by advanced civilizations end up being used, even if in limited form.
If No. 3 is true—that a species gets to the point where its technology allows realistic ancestor simulations—then that species is likely to create many such simulations.
Creating a new simulation then is like winding up a new instance of a video game server (though admittedly this would have to be a very powerful server from our current technology standards). If we assume there is one base reality that is creating many (hundreds, thousands, millions) of ancestor simulations, then every time a new simulation is created there will be exponentially larger numbers of beings in those simulations.
The Statistical Basis for Bostrom’s Argument
Bostrom argument says that if possibility No. 3 is true (that a civilization makes it to the point of creating ancestor simulation), then the number of simulated worlds will vastly outnumber the real worlds. This is true even—and probably more so—if there are multiple races that make it to the simulation points on different planets. Bostrom doesn’t explore this possibility, limiting it to the philosophical argument of one “real” civilization creating many ancestor simulations.
If we take our own civilization as an example, then the number of “beings” in these simulations could be in the billions or trillions since creating new beings simply requires additional computing power. Thus, for each posthuman civilization, the number of simulated beings will almost certainly vastly outnumber real physical beings.
It’s a simple probability, then, that tells us, since there are many more beings in simulated worlds than the physical real world, that the chances that we are simulated beings in a simulation is very high.
Figure 17: The basic Simulation Argument
This argument can be visualized in Figure 17. The left side represents the civilization that has made it past the simulation point.
The basic argument is that if you look at point A as the number of simulated beings and B as the number of “real” beings in the civilization, then A is likely to be much larger than B for a single civilization (in mathematical terms, A>>B). This means that there are many, many more simulated beings than actual beings because it just takes computing power to spin up an entire new simulation.
The likelihood then that we are in a simulation is simply the ratio of simulated beings to the ratio of actual beings. If A >> B, then this will be more than 1!
Bostrom uses slightly more complicated mathematics to make this same basic point: that if any civilization is likely to reach the simulation point, then the number of simulated beings in all universes (simulated and “real”) vastly outnumbers the actual human beings in the “base civilization” (i.e., the “real beings” vs. the “simulated beings”).
Of course, this assumes we are only looking at the left side of the diagram. Bostrom and his collaborator, Marcin Kulczycki from Jagiellonian University, issued a patch for the simulation argument when it was pointed out that he was only looking at the real beings in the civilization that have made it past the posthuman phase.
The rest of the beings in civilizations that have not yet reached the simulation point (he calls them, awkwardly, pre-posthuman) also need to be accounted for. Our current civilization (assuming we are not in a simulation) would fit into this category, as we have not yet made it to the simulation point.
These civilizations will either pass through the simulation point with the development of more sophisticated technology (as we are likely to do over the coming decades and/or centuries) or they will go extinct. Is there another alternative that they survive in a pre-posthuman state for a long period of time? This depends on whether it’s a technological civilization like ours, which develops computers and video games. It’s possible some civilizations never build computers and never build video games, but it’s unlikely that a technological civilization wouldn’t have some kind of computation.
While adding in these beings makes the number of real beings larger, there is still no doubt that simulated beings can be cranked up almost infinitely very quickly, and so the basic argument still holds. In other words, A >> (B+C), though the ratio may not be as skewed as it was initially. If the number of simulated beings in all the simulations is much larger than the number of simulated beings in all the real-world civilizations, then we are still more likely inside a simulation than not!
In an interview with Larry King, physicist Neil deGrasse Tyson, while discussing Bostrom’s argument, also brings up the idea that each simulation could have beings who create sub-simulations (or simulations-within-simulations), so the number of simulated beings could be almost infinite. He calls it the “simulations all the way down” argument.
Bostrom’s equations are kind of like the famous Drake Equation, which tried to estimate the number of alien civilizations there are in the galaxy. You have to put in assumptions about how many stars there are, what percentage of those stars have planets, how many of those planets have life, and what percentage of those develop technological civilizations, how long those civilizations last, etc. The answer, of course, changes dramatically based upon the assumptions. Research now shows that planets are more common than we originally assumed and that the number of stars in the galaxy (and the number of galaxies in the universe) is much greater than we thought back in the 1960s when Frank Drake proposed the equation. Similarly, as our technology advances down the road to the simulation point, it becomes ever more likely that Bostrom’s assertion is accurate.
