Quote from: forcefield on January 03, 2013, 01:42:37 AM
If Quenco were to work in the way that I understand it to work, then yes, it "reduces entropy."  But, not in an isothermal environment as the average temperature of the system goes down.

Ok there's two different scenarios that are getting conflated here (probably by me as much as anyone else )
So there's the idea that a MD machine can take an isothermal environment and reduce it's entropy (by sorting).   So perhaps it might be clearly to say that an MD machine reduces entropy FROM an isothermal environment.

Another scenario would be in an experiment where the temperature is kept at a constant temperature (sometimes referred to as an isothermal "bath") an MD device should keep being able to do work.  Which is why Philip and the Quencoites destroy pentodes in toaster ovens.

Quote
Is this not how Quenco is being "advertised?"  Is this not a reduction of entropy?

I think so.   So MD machines have been looked at in a number of ways.   One proof was done which stated (loosely) that whatever mechanism is doing the sorting has to be making a decision, in order to make a decision information  must be, at least temporarily stored.  Then at some point erased.   Another proof showed that erasing data has an entropic cost.

Quote from: sarkeizen on January 03, 2013, 10:22:31 AM
Ok there's two different scenarios that are getting conflated here (probably by me as much as anyone else )
So there's the idea that a MD machine can take an isothermal environment and reduce it's entropy (by sorting).   So perhaps it might be clearly to say that an MD machine reduces entropy FROM an isothermal environment.

Another scenario would be in an experiment where the temperature is kept at a constant temperature (sometimes referred to as an isothermal "bath") an MD device should keep being able to do work.  Which is why Philip and the Quencoites destroy pentodes in toaster ovens.I think so.   So MD machines have been looked at in a number of ways.   One proof was done which stated (loosely) that whatever mechanism is doing the sorting has to be making a decision, in order to make a decision information  must be, at least temporarily stored.  Then at some point erased.   Another proof showed that erasing data has an entropic cost.

I used the prisim only to indicate that not all MD's are locked under information theory.
No measurement of the energy needs to be made to achieve the sorting in that the measurement itself has already sorted it.
The electrons sort themselves from random energies, like photons through the prisim. Though this could be considered counting or information at this point, it is a measurement after the fact.

Quote from: lumen on January 03, 2013, 11:31:41 AM
I used the prisim only to indicate that not all MD's are locked under information theory.

Oh, that's what you're getting at.   As far as I knew you were talking about a MD which was "like" a prism is some way.  The only aspect that you appear to mention was sorting by energy level which is done by number of proposed MD's.

If you want to use a prism in that way.  You first need to prove (in the mathematical sense) that it *is* an MD.  Which it isn't but feel free to try.

While you're at it, why don't you build a machine that can deterministically tell you if an arbitrary computer program will end.

Quote from: sarkeizen on January 03, 2013, 02:08:42 PM
Oh, that's what you're getting at.   As far as I knew you were talking about a MD which was "like" a prism is some way.  The only aspect that you appear to mention was sorting by energy level which is done by number of proposed MD's.

Though a number of MD's sort by energy level, they first determine the level which now falls into information theory and these MD's can never work.
For a MD to work, it must not perform any test on the quantum objects or the test will alter the object.
Through the prisim, the photons sort because of their individual properties, in a quenco, the electrons sort by their accumulated kinetic energy and no test is performed to select the electrons.

Quote from: sarkeizen on January 03, 2013, 02:08:42 PM
If you want to use a prism in that way.  You first need to prove (in the mathematical sense) that it *is* an MD.  Which it isn't but feel free to try.

It may be possible to prove a prisim as a MD but the point is that sorting is performed at little or no cost and yet there is information produced.

Quote from: sarkeizen on January 03, 2013, 02:08:42 PM
While you're at it, why don't you build a machine that can deterministically tell you if an arbitrary computer program will end.

In a realistic term this may not be as difficult as you suggest.
Suppose you take a large number of arbitrary computer programs (by this I'm thinking of just a pile of random binary numbers) and run them with the computer.
It would seem that some would end very fast and others would take longer and longer and some would never end from internal loops.
Just logging that information would let you plot a curve and estimate the probability of a particular program ending on that CPU.

Due to the instruction set of the CPU used in the test, the random data would develop a pattern of predictability based on the size of the arbitrary program and you could know the probability of any size program ending in a given time.

Seems like a waste of time to me, being a computer programmer for 30 years.

Quote from: lumen on January 03, 2013, 08:44:26 PM
It may be possible to prove a prisim as a MD but the point is that sorting is performed at little or no cost and yet there is information produced.

If a prism is NOT an MD then it's irrelevant to what information theory says about MD's.  If you would like to *construct* an MD using a prisim then please explain exactly what it is you are constructing.

Sorry for not responding to your other comments they were way too vague.

Quote
Suppose you take a large number of arbitrary computer programs (by this I'm thinking of just a pile of random binary numbers) and run them with the computer. It would seem that some would end very fast and others would take longer and longer and some would never end from internal loops. Just logging that information would let you plot a curve and estimate the probability of a particular program ending on that CPU.

So you fail solving the problem in a number of ways there.  One particular glaring way is that you "log the information" for programs that don't end.  Exactly how are you determining that the program doesn't end so that you can log it's information so you can plot your curve? You've essentially embedded the same problem into your solution.  I hope this isn't standard engineering practice.