Are we all part of a computer game?

It’s more than likely. No, really.

Nick Bostrom wrote an excellent paper on this, and let me tell you: it isn’t just a new plot of a Hollywood blockbuster. I have summarised (greatly) the key issues as they are presented in the paper, but I strongly suggest you read the full paper before you dismiss Bostrom’s arguments.

Here is the thesis, in a nutshell:

At least any one of these three statements must be true:

2. If humans were able to create a computer with enormous amounts of computing power, they would never use this power to create a simulation of their ancestors; or

3. We are living in a simulation.

Using these three propositions, Bostrom suggests that third is the correct one: it is highly likely that we are “among the simulated minds rather than among the original biological ones”. In other words, if you do not believe that you are a part of a simulation, then you should not believe that our civilisation will evolve to a level where such a thing is ever possible.

Here are the two main assumptions, and a short explanation for why they are relatively reasonable:

1. Computers will be able to simulate the conscious mind. At first, this sounds like the deal breaker, right from the start. But if you think about it, there is no logical, scientific reason why computers will not, one day, be able to do this. Some experts predict that such a computer will be operational within a couple of decades.

But we are not limited by time for this argument. We can wait a couple of decades, or a million years: the point is that it is a reasonable assumption that computer capabilities will continue to evolve, until one day a computer will be able to simulate the mind. And if one such computer can simulate one mind, it follows that at some later stage there will be more of those computers, or a computer that can simulate six billion minds.

In fact, there are some supporting numbers here. (Feel free to skip this paragraph if large numbers make you dizzy.) It has been suggested that a computer the size of a large planetary body will be able to compute 10 to the power of 42 operations per second. Bostrom makes a very good argument that a realistic simulation of human consciousness and history will require anything between 10 to the power of 33 and 10 to the power of 36 operations per second.

So, even if the calculations are off by several orders of magnitude, it would not discount the argument. In fact, this computer would simulate the entire history of humankind using less than one-millionth of its processing power for one second. Bostrom points out that future (post-human) civilisations could build an “astronomical number of such computers”.

And think of Second Life, which is our first public attempt at a population simulation.

2. Assuming then that such computers can exist, we can assume that they would be used to play this simulation game. This is not a far-flung assumption, since even our limited computers today are used to simulate life in Second Life and other virtual reality games.

If we assume that powerful enough computers, capable of simulating the history of humankind, can one day exist, and we assume that at least some of those computers will then be used for that purpose, we have to assume that we are a result of that stimulation. And if not, then why not?

If we are not a part of a simulation, that means that any of the two arguments below must be true:

1. A sufficiently powerful computer was not developed, which means that a (technological) civilisation was wiped out before it reached that level. Which suggests that we too are due for extinction, because it is “it is hard to see any justification for thinking that our own species will be especially privileged or protected from future disasters”.

2. It is also possible that although the computational powers exist, nobody is interested in running a simulation. They might not believe it to be ethical, they might be forbidden from doing so by a regime or they might have completely other ways of deriving pleasure.

The crux of the matter: Unless we are now living in a simulation, our descendants will almost certainly never run an ancestor simulation.

I find this argument very compelling, although I have not yet seriously thought about the implications of this.

If you have counter-arguments or theories, please post them in comments below. Nick Bostrom tries to respond to comments and suggestions, as long as it is done via a reporter. And I shall use the powers of this blog to simulate myself as one!