004_Are_you_living_in_a_computer_simulation.md

Are you living in a computer simulation? [pdf]

Bostrom, Nick. "Are you living in a computer simulation?." The Philosophical Quarterly 53.211 (2003): 243-255. APA

The simulation argument states that one of three cases is almost certainly guaranteed to be true: (1) civilizations go extinct before they have enormous computational resources (2) civilizations have no interest in running realistic simulations (3) we live inside of a simulation. To form this argument, a few reasonable assumptions are made, for example that given enough resources, it becomes possible to perform realistic simulations. The number of people living in a simulation is then estimated mathematically, which shows that one of the three propositions above is very likely to be true.

1. Introduction

• Over time, more computational power becomes available
• Given enough such resources, one might try to create a simulation of our world or something similar
• If this simulation is fine-grained enough, it would seem realistic to those living inside it
• If this is really the case, then it would be rational of us to assume that we might be living in such a simulation

2. The Assumption Of Substrate-Independence

• Any system with the right kind of details would be associated with consciousness
• This is independent of whether the system is implemented on carbon-based neural networks or on silicon-based processors
• Conclusion: Given enough computing resources, it would be possible to simulate something that appears to be conscious

3. The Technological Limits of Computation

• More computing power becomes available over time
• This argument makes no assumptions about the time frame. It still works if it takes millions of years until enough resources are available
• There have been proposal for huge computing machines with the capability of running 10^42 operations per second. One day it might be possible to build these
• An upper bound for computational power is much harder to estimate
• This computing power would be more than enough to simulate the minds of many humans over a long time period
• To make the simulation more efficient, one could use the fact that not everything is observable all the time. E.g. star systems far away from any civilization would not need to be simulated in full detail
• If the simulation is buggy and a civilization notices, the person in control of the simulation could just rewind time
• (Side note: The exact numbers do not seem to be based on much evidence. But all that is required for the argument is really that one day there could be enough compute power to realistically simulate worlds, given that at least one civilization survives for long enough)

4. The Core of the Simulation Argument

• To show that at least one of the three propositions must almost certainly be true, we try to estimate the fraction of people living in a simulation
• (fraction of people living inside a simulation) = (people living inside a simulation) / ((people living inside a simulation) + (people not living in any simulation))
• (people living inside a simulation) depends on the fraction of civilizations that reach a stage where they have enough computational power, the fraction of those civilizations that are interested in running a simulation, the number of simulations each such civilization would run, and the number of people living in those simulations
• There is no reason to believe the population numbers for individual simulations would differ to the population of a real world, since the simulations are meant to be realistic
• The number of simulations performed by interested civilizations should be huge, considering they have more than enough computational power
• This only leaves three possibilities
  1. The fraction of civilizations that reach a stage with enough computational power is nearly 0
  2. The fraction of civilizations that are interested in running those simulations is nearly 0
  3. The fraction of people living inside a simulation is nearly 1
• (Side note: The mathematical formulas in this part of the paper seem very inelegant. I think unstated assumptions about independence are made, given the way that the formulas are expressed using expected values. However, a cleaner way to model the general idea should yield the same results)

5. A Bland Indifference Principle

• We should assume that the probability of us being in a simulation is equal to the fraction of humans living inside a simulation
• Without any auxiliary knowledge, this is the best assumption to make

6. Interpretation

• If (1) is true
  • Civilizations generally do not reach a stage where they have enough resources
  • Does not mean that civilizations go extinct. It could also just mean that technological development stagnates too early
• If (2) is true
  • Civilizations are not interested in running large-scale simulations
  • There are no more wealthy, influent individuals that are able to run these simulations
  • Maybe this is for ethical reasons
• If (3) is true
  • We only observe a small fraction of everything there is
  • Simulations could be stacked on top of each other, meaning that new simulations are created in many simulations
  • This would be pretty expensive for the base world, so they might shut down their simulations once they become too expensive
  • The general idea behind (3) gives civilizations an incentive to act moral: If they do not, they might be shut down by the civilization that created their world. Since the base civilization cannot be sure that they are not living in a simulation, they would also have to act moral
• Besides simulations where every person is simulated, one could imagine one where only a few people are not
  • Living in such a simulation would be fairly unlikely if simulations where everyone is simulated exist
  • The above thought is based on simple counting (number of people inside of simulations vs number of people in general)
  • It is questionable whether there would be an actual difference between real and simulated people

7. Conclusion

• If we create a simulation, it is almost certain that we are living in one ourselves
• Contraposition: If we are not living in a simulation, it is very unlikely that we will ever create one