On 8/19/2012 6:03 PM, meekerdb wrote:
On 8/19/2012 2:43 PM, Stephen P. King wrote:
On 8/19/2012 4:30 PM, meekerdb wrote:
On 8/19/2012 12:51 PM, Stephen P. King wrote:
I understand that 2+2 = 4.
I still cannot explain how and why I understand "2+2 = 4".
"2+2=4" is easy.
"I understand 2+2=4" is quasi infinitely more complex.


Dear Bruno,

As I see it, the quasi-infiitely more complex aspect of "I understand that 2+2=4" follows, at least, from the requirement that many entities capable of making such statements can point to examples of 2+2=4 and communicate about such statements with each other however far away in space and time they are from each other. We can ignore the fact that there is a collection of entities to whom the statement "I understand that 2+2=4" has a meaning. You need to get a grip on the nature of meaningfulness. Searle has tried to do this with his Chinese Room idea but failed to communicate the concept. :_(

Maybe Bruno will introduce a new modality to his logic Up="Understands p". :-)

Brent
--


Hi Brent,

That would be wonderful if possible. AFAIK, understanding is contingent on demonstrability, e.g. I understand p if and only if I can demonstrate that p implies q and q is not trivial and q is true in the same context as p. I think that Bruno's idea of "interviewing a machine" is a form of demonstration as I am trying to define it here. In my thesis, demonstrability requires that the model to be demonstrated is actually implemented in at least one possible physical world (i.e. satisfies thermodynamic laws and Shannon information theory) otherwise it could be used to implement a Maxwell Demon.

BTW, it was an analysis of Maxwell's Demon that lead me to my current ideas, that abstract computation requires that at least one physical system actually can implement it. This is not ultrafinitism since I am allowing for an uncountable infinity of physical worlds, but almost none of them are accessible to each other (there exist event horizons, etc.). Consider the case where a computation X is generating an exact simulation of the behavior of molecules in a two compartment tank with a valve and there exists a computer Y that can use the output of X to control the valve. We can easily see that X could be a subroutine of Y. If the control of Y leads to an exact partition of the fast (hot) and slow (cold) molecules and this difference can be used to run Y then some might argue that we would have a computation for free situation. The problem is that for the hot/cold difference to be exploited to do work the entire apparatus would have to be coupled to a heat reservoir that would absorb the waste energy generated by the work. Heat Reservoirs are interesting beasts....

If your computer simulation is acting as Maxwell's demon then you don't need a heat reservoir.

 Hi Brent,

Good point. I stand corrected! But did my remark about understanding make any sense to you? I am trying to work out the implication of the idea of Boolean algebras as entities capable of evolving and interacting as it is a key postulate of the idea that I am researching. The Maxwell Demon is just a nice and handy toy model of this idea, IMHO. Could the Maxwell Computational Demon "understand" what it is doing? We could add the capacity to have a self-model as a subroutine and thus a way to gauge its actual efficiency against a theoretical standard as a way to implement a "choice" mechanism... See http://www.youtube.com/watch?v=ehno85yI-sA for a discussion of this self-modeling idea.

The demon makes one tank hot an the other cold so a heat engine runs on the difference.

Yes, the demon would act in a cycle: Compute the simulation to operate the valve to segregate the hot from cold and then use the heat engine to charge a battery, discharging the difference in temperatures. Can this run forever? No, given real world things like friction and the wearing out of parts, but in the idea case it might seem to be able to run for ever.

Unfortunately this is impossible because such a simulation would require defining the initial state of the particle's position and momentum in the two tanks. This is not available for free. To determine it by measurement takes at least as much free energy as can be recovered after implementing Maxwell's demon.

The idea case would shift the initial position/momentum question into a synchronization question: how is a measurement different from the "inverse" of a simulation? I do not have any good words to express my thought here... Let's see where the discussion takes us.




See http://www.nature.com/news/the-unavoidable-cost-of-computation-revealed-1.10186 for more on this.

But if you're doing a calculation once on a given machine it's not necessary to erase the result. In Feynman's paper on quantum computing he note this gets around Landauer's limit. So long as the evolution of the computation is unitary no energy need be dissipated. So I don't see how the result is relevant to Bruno's UD.

The reversibility argument only works if there is sufficient black memory to work with such that erasure never is necessary. This is just trading off the recource of energy for the resource of a read/write medium. Given the wearing out of parts situation, could this be dealt with so that it is not a problem for the idea case aka no friction, no loss of heat to an external world...

See also http://www.csupomona.edu/~hsleff/MD-power-time.pdf <http://www.csupomona.edu/%7Ehsleff/MD-power-time.pdf> for a nice discussion...

I am trying to met Bruno half-way in his COMP idea... I just can't let go of the apparent necessity of actual physical implementation, even given that I really like his immaterialist hypothesis. It is too much like Leibniz' PEH and its reliance on the logically impossible. How is Bruno's idea not a proverbial floating castle in the sky?



Brent

--
Onward!

Stephen

"Nature, to be commanded, must be obeyed."
~ Francis Bacon

--
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To post to this group, send email to everything-list@googlegroups.com.
To unsubscribe from this group, send email to 
everything-list+unsubscr...@googlegroups.com.
For more options, visit this group at 
http://groups.google.com/group/everything-list?hl=en.

Reply via email to