[FOM] more on Toby Ord

[Toby Ord]

On 12 Mar 2004, at 01:46, Martin Davis wrote:

> Toby Ord has a stance of neutrality on the question of whether 
> "hypercomputation" can exist in our universe. Until and unless there 
> is a proof that it is impossible, he asks that equal credit be given 
> to both possibilities.
>
> In my paper "The Myth of Hypercomputation" I compare those who seek 
> hypercomputation devices with those who build "perpetual motion 
> machines". It is certainly conceivable that some "astounding" 
> discovery will contradict the laws of thermodynamics. But what would 
> one say to those who call for neutrality between advocates of 
> perpetual motion and their opponents?

A very good question! Unfortunately, I do not know much about the 
reasons why physicists have come to strongly believe perpetual motion 
is impossible and thus can't say too much myself. However, if the 
reasons stem from explicit principles that rule it out without adding 
to the rest of physical theory then I would say that we have no reason 
to rule it out. From what little I do know about this it seems that 
there were good reasons for it being impossible in classical physics 
(which I *think* said that in some high proportion of initial 
conditions, it would be impossible). However, I have talked a little 
with some philosophers of physics here at Oxford about this and the 
situation seems much more shaky in quantum mechanics.

Even if QM ruled out perpetual motion, that would not mean that it was 
impossible. It is physically impossible if and only if it is impossible 
under whatever physics actually governs the world. However, to the 
degree that we have good evidence that QM (or whatever) is the system 
that governs our world, so we would have good reason to believe that 
perpetual motion is impossible. As Timothy Chow has pointed out 
previously, things can be more complicated if we have reason to believe 
a system (say Newtonian Mechanics in the 1800s) models our world at low 
speeds, but actually have less reason to believe it for relativistic 
speeds since there was so little data regarding them. Also remember 
that pure random chance and triangles with interior angles summing to 
181 degrees were ruled out on Newtonian Mechanics, a theory that we had 
good evidence for over quite a wide range of phenomena, and that now we 
accept both.

How does this relate to physical hypercomputation? Well, I think it is 
worth studying, especially at this early stage. If we examine the 
problem for quite a while and all possibilities seem to run into the 
type of endless difficulties that one sees regarding perpetual motion, 
then it probably should be put on the back burner for a while. If our 
well tested physical theories actually rule it out, then I imagine it 
will receive very little study (from a physical perspective at least) 
and I would just want people to keep as open a mind to it as we should 
have had to 181 degree triangles and pure randomness in the 1800s. 
However, I am not sure how open this should have been!

In general, I think that comparisons of hypercomputation with 
randomness, non-euclidean geometry and perpetual motion are quite 
worthwhile and revealing. (Indeed, it seems to me that some of Timothy 
Chow's arguments regarding the unknowability of object X being a 
hypercomputer apply equally to X being a random bit generator...)

Toby Ord.