Re: Time Machines, FTL, and P=NP

Jens Egon Nyborg wrote:
> Bryan Derksen skrev:
>> Michael Ash wrote:
>>> Since you don't want to wait around for 300 years, you use your
>>> wormhole and write this instead:
>>>
>>>     get x from future-wormhole
>>>     if no future information exists yet because it's the first run:
>>>         set x to first possibility
>>>     test x
>>>     if x is good, return x
>>>     else put x into wormhole
>> I assume you mean "put x+1 into wormhole"? An infinite loop could have
>> interesting consequences.
>>
>> This is basically Hans Moravec's time loop logic
>> <http://en.wikipedia.org/wiki/Time_loop_logic>,
but you came up with
>> some really neat applications that might not immediately occur to most
>> computer scientists. :)
> 
> There are clear similarities, but it seems to me that Michael Ash is
> using paradoxes to do the work, whereas Hans Moravec is relying on
> paradoxes being prohibited by natural law.

I can't see any difference in the end result, though. In one case you're 
having billions of futures flicker into existence and then cease to 
exist without any trace, leaving only the one that produced the 
"correct" answer. In the other case you're generating a random number 
and threatening that you'll do an impossible thing unless it's correct, 
forcing the universe to give you the right answer as the only possible 
outcome. Either way, the algorithm gives you the right answer the first 
time you run it.

Since having paradoxes actually "happen" somehow doesn't make logical 
sense, by definition, I prefer the latter interpretation. But it makes 
no difference in practical terms.