Hacker News Reader

Comment by zapnap | Hacker News Reader

with proper TDD (tests representing the environment) it should be possible to generate random commits and only commit them when they pass the test suite.

What do you do when there are serious ethical considerations with running your test suite?

Run it in a Virtual Universe with the ethernet cable unplugged.

I would say it should make us realize we are no more than mere simulations in some huge computer about to have the power cable pulled by god's boss before he tells him to get back to work. (scifi book idea )

I'm pretty sure that was the plot of a fairly well know scifi book cough hhgttg cough. Except it wasn't God's boss that pulls the cable - Earth was demolished to make way for an intergalactic bypass.

That's kind of a stretch of an interpretation, don't you think? I always thought that the Earth was still physical in the way that we usually think of it, and humans were just part of its physical operation rather than being simulations.

Yeah, us being the substrate rather than the software is a bit of an unusual take on it, even now.

Procedurally generated programming?

Given how that works out for the world we live in, there really would be more bugs than anything else....

http://www.si.edu/Encyclopedia_SI/nmnh/buginfo/bugnos.htm

Because a passing test suite is proof of correctness, right?

This raises the possibility of us finding out that happiness/nirvana is NP-complete, leading to the mass suicide of the entire human race. That would suck.

Haha, that breeds all kinds of crazy ideas. If one could simulate the world's biosphere and its organisms at the genetic or even cellular level, we might one day be able to predict the effects of evolution on humans and other animals.

MAYBE even with bacteria and viruses to predict and then cure diseases before they even present themselves!

I doubt our understanding of genetics/biology is extensive enough for this yet, not to mention the computing power needed, but it seems quite possible.

There is this professor here at Oxford, Nick Bostrom, who argues that we are currently living in a simulation:

1) Moore's law holds across worlds. 2) People/beings will, for various reasons (historic, sociological, experimenting, play) want to simulate another world (at a social/molecular/submolecular, etc level). 3) If they simulate one, they'll likely do more. 4) Ergo, there will be more simulated worlds than real ones. 5) Therefore we are more likely to live in a simulated world/universe than in a first order one.

http://www.simulation-argument.com/

Think about this when you unplug something in the future...

Close; he doesn't argue that we are living in a simulation, he argues that at least one of the following propositions is true:

- The human species is likely to go extinct before reaching a stage where it can simulate its ancestors.

- Future humanity will have no interest in simulating its ancestors.

- We are almost certainly living in an ancestor simulation now.

I simplified it a bit, and explained the last, optimistic option... :)

Reply on news.ycombinator.com