Bostrom posited a compelling argument that at least one of the three following propositions is true:
- The human species is very likely to go extinct before reaching a 'posthuman' stage;
- Any posthuman civilization is extremely unlikely to run a significant number of simulations of their evolutionary history (or variations thereof);
- We are almost certainly living in a computer simulation.
Martin Savage a physics professor at the University of Washington and two colleagues, take the possibility that our universe is a numerical simulation serious.
That isn’t so strange considering scientists in their field are already running computer simulations of universes. These are very small universes of several femto-meters (1 fm = 10-15 m) in size. They’re so small due to lack of computing power. But as high-performance computing grows, Savage expects simulations to scale to the size of molecules, cells, humans and beyond.
Bostrom’s argument runs like this: humankind may reach, what Bostrom calls, the posthuman stage of civilization. This is the point where humanity ‘has acquired most of the technological capabilities that one can currently show to be consistent with physical laws and with material and energy constraints’.
At that level of technology it is possible to run a computer simulation of a world equal to our own. The simulation would not only emulate the physical environment but also the minds of the people experiencing it.
Running the numbers
Bostrom estimates that the computational cost of a realistic simulation of human history is ~1033 – 1036 operations. [See how Bostrom arrives at this figure below.] The computational power of a posthuman civilization is estimated to be much higher than that. Hans Moravec, an artificial intelligence expert, predicted that a computer the size of a large planet could perform 1042 operations per second. So if posthuman civilizations will ever exist they will be capable of running a human history simulation program.
Regardless of their capability, the posthumans may decide not to run such a program for ethical reasons. They may feel sorry for the inhabitants of the simulation. However, if there is no inhibition against running the sim, there is a high probability they will not run one but many such programs.
And that makes the chance that we live in a simulation very, very high.
If there are simulated universes an x % of all people that ever existed will have unknowingly lived in a silicon-based simulation rather than a carbon-based world. The more simulations there are, the higher the percentage of the total number of people living in silico. As the simulated reach the level of technological development where they start to run their own simulations, the number of simulated universes increases. Bostrom writes: ‘one can consider a sequence of possible situations in which an increasing fraction of all people live in simulations: 98%, 99%, 99,9%, 99,9999%’.
So if proposition 1 and 2 are false, we almost certainly live in a computer simulation. If, instead, our ‘proud to be carbon’ bumper stickers are justified, it is very likely humanity goes the way of the dinosaur before reaching the posthuman stage. Or we go for proposition 2 and we all come together to take an oath we will never, ever, run a ancestor-simulation program.
Savage and his co-authors consider Bostrom’s argument ‘logically compelling’. Also, the state of contemporary physics indicates that one day humanity will be able to build large scale simulations. From there it isn’t a far stretch to think we might actually be living in one. ‘Therefore’, the authors write: ‘the experimental searches for evidence that our universe is, in fact, a simulation are both interesting and logical’.
Taking the femto-sized universes as their starting point, the scientists recognized that the simulations have certain signatures that betray their true nature. If these signatures are detected in our own universe, it is an indication we are living in a computer simulation.
One of the phenomena serving as signatures are high energy cosmic rays.
In a model universe the space-time continuum is simulated by a four-dimensional hyper-cubic grid. Along the edges of the grid, cosmic rays behave differently than in real space-time. Instead of traveling straight, they would move diagonally. If we encounter such behavior in our own universe we know we’re in for an identity crisis.
The ~1033 – 1036 figure is based on the following: To replicate the functionality of one human brain Bostrom took the number of synapses and their firing frequency and arrived at a computational cost of ~1016 - 1017 operations per second. Simulating sensory experience has a maximum cost of ~108 bits per second. The environment is only simulated to the extend that inhabitants don’t notice it’s a replica. Neither the core of the earth, no distant stars need to be simulated at a quantum level. Therefore Bostrom assumes most computational power will go to simulating human minds and arrives at the following equation: 100 billion humans x 50 years/humanx 30 million secs/year x [1014, 1017] operations in each human brain per second = [1033 , 1036] operations.
Are You Living in a Computer Simulation?
Constraints on the Universe as a Numerical Simulation