Day 092 - Is it real?

Submitted by Sam on 21 August, 2011 - 03:01

If an advanced alien civilization has enough computing power to create a virtual reality sufficiently diverting to negate their need or desire for galactic colonisation, it follows that such a civilization would have the computational capacity (and coding skill) to create a simulation for and of us. Oxford University professor Nick Bostrom has estimated that a computer with the mass of a planet, operating with the efficiency of what we know to be theoretically capable with nanotechnology today (which is probably not even approaching the physical limits of optimal computation), would be able to perform 10^42 operations per second, and thus able to simulate the entire 'mental history' (or every thought, feeling and memory) of humankind in less than one millionth of a second. And a sufficiently transcendently advanced civilization may have a huge number of computers of this scale.

In fact, as Bostrom has argued, we don't need to postulate superintelligent extraterrestrials in order to suspect that we might be living in a computer simulation – all we need to concede is that at some point in the future our own descendants may reach a level of technological sophistication that permits such a complex virtual reality simulation to be programmed and run. And our descendants may have a clear motive for simulating us, unlike our hypothetical extra-terrestrials – it may be that later generations of humans wish to turn their phenomenal supercomputers into detailed simulations of their own past, to understand more about their forebears. With this goal, they could set about creating fine-grained simulations of the physical world and historic human brains to create a believable virtual-universe, peopled by conscious simulants (us?) unaware of their artificiality.

But how highly resolved does this virtual reality have to be? Certainly, our experience of the world seems unfathomably rich, covering a spectrum where our experiments into quantum permit us to look at the very small, and our space programmes permit us to look at the very distant and the very large. The beauty of a simulation though, is that it can cheat. It doesn't need to persistently model the microscopic structure of the entire observable universe, but merely those parts necessary to ensure that the simulated consciousnesses don't perceive any irregularities. Whenever a human looks into microscopic phenomena with an electron microscope, or peers into the vastness of space with a radio telescope, the observed details can be filled in by the simulation on an ad-hoc basis, rendering only what is necessary for each human to remain unsuspicious, whilst leaving the unobserved universe unresolved.

In a completely unrelated (and incidentally unverified) demonstration of this principle of on-the-fly rendering to conserve computing requirements, an Australian software company called Euclideon Pty Ltd, has claimed that it has developed a new kind of computer graphics technology that allows a graphic environment to be created from an essentially infinite number of virtual 'atoms', rather than a (very) finite number of traditional polygons, by using a search algorithm to work out which of these atoms need to be rendered to generate any given scene from the viewer's perspective at any given time. The rest of the millions of atoms can go unrendered in the background, and are only called into view when they are needed. This new kind of technology, if it is verified, will allow graphics of unlimited detail, limited only by the granularity the artist and modeller work to.

And so extending this 'show only what is needed' principle, a hugely advanced human or alien civilization could cut a lot of computational corners and still create an utterly compelling (and indistinguishably realistic) simulation of all of human experience, requiring 'only' around 10^36 operations to simulate the brains of 100 billion humans, with the environmental rendering as an additional cost. With an excess of computational power and programming ability, a simulation could keep track of the brain-states of every conscious human, and modify them and/or the simulated environment whenever needed to maintain the integrity of the illusion – for instance, filling in worldly detail whenever a human were about to make a microscopic observation. Should any errors ever occur, the software of the simulation could be paused, modified and re-run, and no-one need be any the wiser.

If a civilization possessed planetary-scale computers, they would have enough computational power to run such simulations many, many times over, while only using a fraction of their total computational resources. Bostrom realized that if there is a substantial chance of a civilization ever developing such a capability, be it our children making an ancestor-simulation or an alien civilization making a virtual human zoo, then the number of simulated humans ever to have existed is likely to outweigh the number of biological humans ever to have existed, making it more probable that we are living in a simulation than living as the 'original' biological template for one.

Attribution Noncommercial Share Alike
This text, Day 092 - Is it real?, by Sam Haskell is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike license.
Drupal theme by Kiwi Themes.