Saturday, September 01, 2007

More on the simulation argument

Assumption set 1
1) indifference principle
2) that there are almost infinitely more ways the world could look 'not real' than look 'real'
3) there will not be an infinite weighting towards future civilizations running real simulations (unreal ones would probably be more effective to achieve most aims anyway).

using the above we can say we are either in a very special world or there is a low number of simulations. and therefore that the odds are that there wil never be many high quality simulations.

Assumption set 2
1) indifference principle
2) that civilizations will tend to run more simulations of current time than of ancent historical times
i) because they will be more useful (for example predicting real life behaviour)
ii) they are the data points that you start with - afterall we only know someone like churchill existed by backtracking historic data. So any new simulation program will presumably create you at the present moment and then work backwards to finally make you as a baby and forwards as new data comes in.
3) that simulations will be largely real ancestor simulations (indifference principle already implies this)

Now any simulation of a future period will include simulations of simulations and because of (2) at any instant in time those simulations will be weighted towards that instant in time with less and less sims living in earlier times. If there is a near infinite number of sims and a constant increase in computing power (which is what boistrom seems to suggest) then there would be a massive weighting towards the future sims.

Again this implies either we get sims really soon (regardless of if we are a sim or not) and become extinct - OR we never make many sims.


Post a Comment

<< Home