Maxwell's Demon Faces Heat

[Monte Davis]

Nice catch. There were hints of the connections between energy transfers,
thermodynamics, and information in Von Neumann's approach to quantum
theory, in Shannon. Turing and others, but as the summary notes, it was
Rolf Landauer in 1961 who really got traction on Maxwell's Demon. And
despite the stumbles Pynchon has acknowledged in "Entropy" and the Nefastis
machine in CoL49, his instinct that the connections mattered was sound.

Maxwell posited that the demon ( = any mechanism) could (1) sense the
velocity of each gas molecule approaching its gate, and (2) open and close
the gate accordingly to create warmer and cooler reservoirs. He assumed,
knowing it for an assumption, that it could do so "for free" -- or at least
use less energy than you could get from the newly created temperature
difference between the reservoirs. Hey presto -- entropy lowered in a
closed system, free work (in the thermodynamic sense), perpetual motion.
Maxwell knew that if basic thermodynamic principles -- which he'd helped
formulate -- were correct, something was hinky. The Demon was his challenge
to his peers to figure out what.

Nobody then or for the next century could come close to *building* such a
sensor or such a gate, so nearly all the discussions were "principle vs.
practice": OK, any real sensor would need light or other radiation to work,
any real physical gate would involve friction and energy loss, and maybe
those real numbers would always make the Demon's success impossible. But by
1960 we were making or at least planning microelectronics that flipped and
flopped at energy scales not so far from that of single gas molecules. The
numbers in the discussions were getting more realistic, and they were not
comforting: it began to appear that an actual hardware Demon *might* yield
more energy than it consumed.

The importance of Landauer 1961 (it is no accident, comrades, that he was
at IBM) was that he included the energy/entropy cost not only of sensing
and acting, but of computing... call it "processing" or "deciding" if you
prefer. The Demon would use additional energy for its internal housekeeping
-- writing and moving and erasing bits -- even if it left the gate open or
closed for many molecules in succession, and every time it cleared its data
registers to make room for a decision on the next molecule. You wouldn't
have thought about that before Shannon and Turing taught us the
relationship between "bits" of information and "bits" as physical states of
switches or tubes or transistors -- but once you included that, it began to
look like the Demon couldn't produce net energy after all.

That's been the driving insight from 1961 up to and including this
experiment, and IMHO the only truly new and important element since
Maxwell. Behind the Demon there had always been another tacit, hidden
assumption: that some built-in, passive "pure logic" guided its operations.
But a real brain, or real computer, or real physically instantiated
operating system of *any* kind, is as much an energy-consuming,
entropy-producing "device" as the box-and-piston apparatus Nefastis shows
to Oedipa. Even if the Demon could sense and act for free, it can't decide
what to do for free.

Which just might have something to do with Pynchon's concerns.

<https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=oa-2115-v2-b>
Virus-free.
www.avast.com
<https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=oa-2115-v2-b>
<#DDB4FAA8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

On Mon, Mar 28, 2016 at 6:27 AM, Allan Balliett <allan.balliett at gmail.com>
wrote:

> My apologies it this was shared earlier
>
> https://www.sciencenews.org/article/maxwells-demon-faces-heat
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://waste.org/pipermail/pynchon-l/attachments/20160328/4f65d1fa/attachment.html>