CSS uses DMCA to protect license, not encryption
on
DeCSS, From the Beginning
·
· Score: 3, Interesting
Upon reading of the 100+ page license for CSS, I
had a thought- DMCA isn't protecting the encryption; it's protecting the license for CSS.
Wrap a weak encryption around a product, and only allow legal decryption if you agree to an onerous
license. It doesn't matter how weak the encryption
is, that's not the point. The point is to force
agreement to the terms of the license. This seems to have legal ramifications, since if the purpose of the encryption is not to encrypt, but to activate the DMCA and thereby force the licensing terms, then it's not really encryption; it's a
licensing ploy. So perchance then it doesn't fall under DMCA anymore, since the intention of the scheme
isn't really encryption but licensing?
It's a great idea (and ameliorates the
excess
cow problem as well). Unfortunately, in some
studies we performed for BARPA (Bovine Advanced
Research Projects Agency) a while back we encountered some serious technical problems:
Coefficient of Drag: The average cow's
drag coefficient (even in the most aerodynamically
efficient "ass-backwards" posture) is
approximately 0.6, as compared to 0.02 for a
well-designed streamlined warhead. Since terminal
velocity scales with square root of drag coefficient in the high-velocity limit, and kinetic kill energy scales with velocity squared,
this yields a 30-fold reduction in energy on target.
Density: The density of a typical
Hereford is 1.1 grams per cubic centimeter. In
contrast, the density of depleted uranium is closer to 19 grams per cubic centimeter. The 19-fold reduction in gravitation force then reduces terminal velocity by 19 times, with a consequent 19-fold reduction in kinetic energy on-target (assuming drag
force proportional to velocity squared).
Ablation: The real killer here is
premature ablation of the bovine carapace.
Reentry from 400,000 feet can raise the temperature of the cow's exterior surfaces to
3000 K. First off, the water evaporates, then
the fats burn off, leaving a dessicated cinder.
Even worse the density of the resulting cow cinder
is greatly reduced, reducing terminal velocity
further.
The only advantage here is the nice sizzling
barbeque smell that permeates the stratosphere
on reentry.
That said, of course we recommended further
study.
Is it good thing for people to spend
money on hardware they don't necessarily need?
Spending in itself is not a net benefit to society, only spending on stuff that improves
quality of life. Not spending extra $$ on computer
upgrades could be a good thing for the
economy as a whole- that's more resources for
something else that might be more important to
quality of life.
Micromachines are currently in practical use,
at least very simple ones. The most common
application (to my limited knowledge) is airbag
acceleration sensors. A small silicon
cantilever bends under acceleration and a
resulting electrical signal sets off the air bag.
I very simple micromachine.
Other applications are in the works, but progress
on complex machines is difficult since most devices are built with lithography, and lithography naturally produces two-dimensional
structures, not three. For example,
microfluidics is a potential
application, where small tubes and pumps analyze
microscopic amounts of fluid for various
e.g. chemical/biological assays, etc.
It's a long way from "I can build a gear" to
"I can build a gear and a motor and a shaft and
connect them to a couple hundred other parts with
minimal frictional losses, a power supply, etc. in
three dimensions."
Interesting things will be coming, but it'll be
very incremental as researchers built up a toolset.
Isotopically pure Si could be grown strained
just as well as natural silicon. Should be
no problem to integrate the two technologies,
assuming each was economically feasible on
its own.
Interesting. Maybe any 'random' file of finite
length is not entirely random- there is
information in its length itself. Perhaps this is
what allows for your loophole of the
(extraordinarily unlikely, but not impossible)
random file which is all zeros. This is only
possible with a finite length file. An infinite
length random file will demonstrably have no
patterns.
Doesn't quite work that way, but the
difficulty is very subtle. You have to tell
the decompressor where in the file the string
of zeros starts. A string of 1000 zeroes happens
once in 2^1000 bits. So you need a 1000-bit
pointer to say where to find it. The key to the
impossibility of compressing the truly random
file is the true randomness. Any pattern and
yes it can be compressed. No pattern at all, and
no compression. Cutting the file in two is
sneaky, but it's information hiding, not compression. There's information in where you
choose to cut the 2^1000 bit file.
Isotopic thermal conductivity enhancement? Cool!
on
What 1.7Ghz Is Like
·
· Score: 1
This warms the heart of a solid-state
physicist... isotopically homogeneous silicon
processors would be very cool.
I would be amazed
if they can do it as cheaply as claimed- isotopic
enrichment is usually very expensive (think
Manhattan project- a big chunk of the cost was
isotopic enrichment of the uranium), but fundamentally there is no barrier to someone
finding a cheap way to make it, since the
starting material is very cheap (just Si).
Rarely do I see the words "electron-phonon
coupling" in the press:-). I love it.
Btw, isotopically enriched diamond (one up
from Si on the periodic table) is the best
thermal conductor known. The lattice vibrates
at higher frquencies than silicon (since
carbon is lighter and the carbon-carbon
bonds are stronger than the Si-Si bonds) which
speeds the transport of heat. At low temperatures
(like 20 Kelvins) the increase in thermal
conductivity with isotopic enrichment is
enormous- 10 or 100 fold. At room temperature
this drops to roughly two-fold, but that's
still a huge win.
There was not so much a rush to judgement as a rush to confirm. I was at Berkeley physics at the time in
grad school and *alot* of people tried very hard
to reproduce these results. A colleague of mine,
top of the field, performed high-powered
first principles
calculations of hydrogen-loaded palladium.
Others scattered across the country were working
hard trying to reproduce the result.
No careful experiment
showed anything remotely like fusion. However, a literature search did reveal some interesting
palladium electrochemistry done many decades ago
(I unfortunately don't have the reference) which
showed a possible
heat gain due to electrochemical effects.
Nothing one could extract useful power from.
The comparison to superconductivity given
in the original writeup is spurious. Yes, it's taking many years to understand the mechanism of high temperature
superconductivity. But everyone who tried
could clearly
reproduced the
phenomenom almost immediately. That didn't happen
with cold fusion.
Plenty of people with expertise in calorimetry
tried to reproduce this experiment. Nothing that
required fusion as an explanation was observed.
Palladium has some interesting
electrochemistry, but there's no evidence of fusion, unfortunately. I wish there had been- it
would have been wonderful.
I've seen these issues from many sides:
author, reviewer, editor. Totally free access
and open review falls down on one important
count- the prime currency of academic research reputation (not money; that's not
the main impediment here). People in
academia could generally make more $$ in industry
or business. $$ is not their main motive (except as
$$ in the form of grants is a means to research). The
motives are to advance knowledge, have fun doing
something intersting, and
gain the respect of one's peers.
Most of the value in a research journal is
in the brand: how difficult is it to publish there, how much prestige is due to the author
whose work appears there, how much attention
busy readers pay to publications in
one journal over another.
Without a strong publication record, nearly all
researchers will have trouble getting funding,
tenure, etc.
An open publication system needs to somehow
maintain this means to rate papers based on
quality. Otherwise, researchers will continue
to submit their best results to Science and
Nature, which will then hold copyright.
This could be done: design a semi-open
peer review
system whereby authors can only publish in
the archive if they also provide quality peer reviews,
and manage it all through community-wide feedback
and also possibly the efforts of professional editors skilled in the field (who would have
to be paid). And rate papers based on their
importance to the community and the judgement of
the referees and readers.
In the end, it sounds like a variant of
slashdot itself- self-organized publication
ratings with mechanisms in place to minimize
abuse. Resistance to abuse is critical-
junior researchers under a tenure cloud will
have great incentive to work the system.
The page charges at the APS are voluntary. If
you don't want to pay, you don't have to.
Elsevier might not have page charges, but
their subscription prices are typically much, much higher
than those of APS.
Many specialized western journals also cost
thousands of dollars per year. Of two main
journals on Schizophrenia, the one published by
the NIH is under $100/yr and the one from a
commercial publisher is over $2000/yr.
Journal publsihing is a strange market: the
author does not pay for publication so she or he
is as likely to submit to an expensive journal
as a cheap one. Since the journal then holds the
copyright, libraries are forced to pay the
exorbitant subscription rate of some commercial
journals or else they have a hole in their
collection. Commercial publishers figured this
out several years ago- their academic publishing
arms make a quite high return on investment.
Every individual manuscript published is like
a little monopoly for the journal on a slice of knowledge.
I agree with these comments: long-term applications of quantum computing are more likely
to come from solid-state based devices, possibly using electron spin or nuclear spins of
embedded impurity atoms as qubits. Optical
atom traps (and NMR magnets) are very
expensive and likely to remain so for
quite a while. But in the end
it's really too early to say, of course, since
we're looking ahead a long way.
This work is impressive, but NMR-based quantum
computers have some special scaling problems. The net
spin polarization of these liquids at room
temperature is very small- about one part in
a thousand roughly. This means that the quantum
computation is done with only the slightest of
distinction between 0 and 1. That can be handled
for small computations, but the signal to
noise problem becomes severe for larger systems.
The current quantum error correction codes
impose a large bit overhead (factor of three
or so), which compounds the signal/noise difficulty of the NMR technique.
The nice thing about this method is that the
nuclei are well-isolated inside the electron
clouds and decohere slowly. However, my guess is
that NMR-based quantum computing will be first
out of the gates, but will lag behind other
implementations before reaching the finish line.
Nice points, all. I've thought a little bit
about GPS-based
traffic systems and I think there's a
feasible way to implement it, but
it's a different philosophy than speed governors.
Keep all the map db info on the server, where it
should be. The GPS in the car simply records a
time record of the travel path of the vehicle.
This information is occasionally fed to local
(or satellite) receivers scattered about. The
information is then fed into an algorithm that
calculates a monthly bill for moving violations,
etc. The central system need not even store the
bits for the actual traffic pattern- it could
process the traffic pattern (the overpass
which-road-am-i-on ambiguities
issues raised above can be covered through a
time average of speed or by removing the
overpass regions from the processing) and
then throw
it away, storing only the monthly bill.
If the driver wants an itemization or to
challenge the bill, they go to
their car, which
is the only device that actually retains the
privacy-sensitive info on their traffic patterns.
Of course, the system is open to hacking and
backdoors, as would any wide-spread IT system.
But a good (open-source?) implementation could
help ameliorate these concerns. The personal
info must go to the central server, but it need
not be stored to disk there. Just converted
to XX $$ bill. Ain't perfect, but possibly
workable? I'd be interested to
hear reactions from those more tech-savvy than
me.
Many of the disadvantages of such a system
have already been posted here, but also keep in
mind the advantages. Saving life. Saving fuel.
Saving the environment. Saving time (yes- see
next paragaph). Greater freedom (yes- see next
paragraph).
One nice irony of the automobile as a symbol
of freedom (in U.S.) is that the road system
is a nearly perfect example of communism. It's
build and managed by a central governmental
authority, and access to resources are allocated
by queue (ever been in a traffic jam:-) ? ).
Soviet bread line, anyone? Road congestion
to a large extent results because a driver
does not pay the very real cost that (s)he imposes
by slowing down the traffic for everyone else.
That economic failure encourages excess driving.
A GPS-based pricing system could go some way to alleviating that problem. (This line of thought
now splits into multiple interesting threads
on social implications, fairness to the poor, etc.
but only finite time to type). Freedom to
some extent is freedom from the unfair "free-ride"
impositions of others...
I am a slashdotter myself occasionally.
I am also currently performing research on
nanotubes at a major research unversity, with
publications in Science, Nature, and
other places on nanoscience.
Perhaps the slashdot readership is a bit
more diverse than you though. I find it an
interesting place to visit, with some
very intelligent conversation.
(And thanks much to the moderators).
Buckyballs don't have any important current
application that I know of (there was some
work in nonlinear optical materials, but I
don't know if it panned out).
Carbon nanotubes have a pretty good chance of
appearing as field emitters in flat-panel
displays before too long. They have great
conductivity, they're very pointy and extremely
robust under the relevant electrical conditions. They are already at
the simple prototype stage. Batteries might also
pan out (they store lithium quite well) but
tubes are still much too expensive for that.
The tubes will very likely find many more
commercial applications than buckyballs.
Oh- and there's one application out there
right now. Mass-produced (and very low-quality)
multiwalled tubes are currently sold to mix into
plastic parts to add conductivity to the plastic
so that they can be charged up
uniformly and electrostatically painted. Nice
shiny plastic car bumpers. Not quite an
elevator to geosynchronous, but it makes (a little) money.
One can estimate theoretically the ultimate
strength of a nanotube be examining the microscopic failure modes, i.e. the ways in
which atoms rearrange in response to an external
stress (i.e. stretching).
In the case of perfect, defect-free
nanotubes, there are two modes that seem to be
important. First, the rotation of a single
carbon-carbon bond by 90 degrees, which converts
a patch of 4 hexagons (remember that carbon atoms
are arranged in a chicken-wire or honeycomb
pattern on the tube wall) into two pentagons
and two heptagons (relevant references are
Zhang & Crespi from Penn State
in Physical Review Letters and
work by Bernholc at NC State and
Yacobson at Rice I think, but the exact
journal escapes me at the moment). This mode
is a plastic distortion of the tube; the tube
with the bonds rearranged is a bit longer than
it was before.
The second failure mode is for one of the
hexagonal rings of carbon atoms to break open, i.e. for a carbon-carbon bond to break. This is
a more catastrophic event, in that the tube then
quickly breaks near the point of failure.
Which way a tube fails may actually depend on
how the honeycomb pattern is rolled into a
tube shape.
Now that's just the microscopic theory on
the ideal, defect-free system. In a real tube,
one expects there to be pre-existing defects
in the structure. The failure under tension
will then be at the defective points
But, since nanotubes are so small, it's plausible that a single tube or bunch of tubes
might grow entirely defect-free, in which case
one can access the ultimate theoretical
failure strength. Experiments on trying to
stretch and break single bundles of nanotubes
(Lieber's group at Harvard) show that one can
extend a nanotube by about 6% of it's length
before it breaks. This is in good agreement
with the theoretical predictions mentioned
above (and it's a legit prediction- the
theory came first!). So it appears that in
small enough systems, one can attain the
theoretical mechanical strength.
Now if one wants to make a space elevator,
one's material has to also be resistant to
radiation damage, etc. I think a back of the
envelope estimate shows carbon nanotubes or
diamond nanowires as being in the right
ballpark, so long as one allows the structure
to taper, but once one factors in the necessary
engineering margins and the need to be resistant
to damage over long periods (don't want it
to fall apart in a year or two:-) it's much
less clear if it's really possible. It's all
in the very very long run, of course.
I should admit- I have not yet read this
specific article (New Scientist website is
crashing on me) so I can't comment specifically
on this current experimental result. My guess is
they did a larger-scale version of Lieber's
experiment and found that the resulting thread
was alot weaker (not surpirising- their
structure likely has lots more defects and
possibly single tubes don't extend throughout
the entire length- they overlap).
OK, the definition of a decoherence-free subspace:
Quantum mechanical wavefunctions are described
in terms of their projection onto a set of basis
functions. This is exactly analogous to a Fourier
transform of a function (i.e. the projection of
the function onto a set of sine and cosine function).
As the wavefunction of say an electron
evolves with time, the
weights of the various basis functions will
typically change. If the wavefunction is
coupled to other systems (i.e. other electrons,
surrounding atoms, molecules, etc.), then the wavefunction becomes very complicated as the
pieces that describe the single electron mix
up with the pieces describing the
other parts of the system.
This is termed decoherence.
The subspace referred to in the posting
is a subset of the full set of basis functions
(like taking a finite bandpass of the Fourier
space). For wavefunctions that can be described
completely in terms of sums of the basis functions
in this subspace, the wavefunctions will maintain
their coherence, which means that
as they evolve in time, they don't get mixed
up with the wavefunctions describing the
surrounding environment
This isolation of a subset of the degrees of
freedom describing a system
(i.e. the decoherence-free subspace) is essential for
quantum computing, as a quantum computer uses
the subtle correlations within a wavefunction
to perform what are essentially massively
parallel computations. Should the system
decohere, the subtle structures in
the wavefunction are lost.
This is a major outstanding problem with
nanoelectronics nowadays- the third terminal. It's
easy (well, if you have access to about a
million dollars worth of e-beam equipment,
or a hundred K of scanning probe equipment)
to get two contacts onto a nanometer-scale device.
The tricky part is the third contact. Typically,
the third contact is underneath, the e.g. field-effect lead, and it's very big compared to the rest
of the device. One
can also do alot with only two contacts, if the device
is hysteretic (i.e. has memory of its history).
Many of the molecular electronics groups,
realizing the limits to our current ability in
contacting and integrating electronic devices
on this scale, are concentrating first on implementing very simple memory architectures.
Various aspects and approaches to these ideas
are being pursued at many places: several
groups at Berkeley (mostly physics),
Penn State University (chemistry, elec eng, physics), Harvard (chemistry), Notre Dame (physics), Delft (in Netherlands), HP, Rice (chemistry),
IBM, and others. No-one knows yet what if any approach will pan out.
Being in the top full row of the periodic
table, it is rather small, which means that the
electrons feel a strong electrostatic potential
from the nuclear charge, and therefore can be
tightly bound into molecules and more extended
structures. In addition, an accident of the
fundamental constants means that carbon (and nearby elements on the periodic table) can form
multiple bonds to neighboring atoms. One consequence of this is the ability to form continuously bonded atomically thin structures
such as graphite. Graphitic structures can then
be curved into long thin nanotubes. (Another
consequence is the richness of organic
chemistry, and biology).
Graphite sheets are also rather unusual in
being what are called semi-metals: on the edge
between a metal and a semiconductor. Rolling them
into a tubular shape is actually enough to push
them one way or another into real metals or
real semiconductors in carbon nanotubes.
Both types can even be combined
within a single tube to make diode, etc. on-tube.
The strong in-plane bonding of carbon in
graphitic strucures (see point 1 above) means
that it is an excellent material for high strength-to-mass applications (if the fibers can
be embedded in an appropriate matrix). This also
means that the structures could be very thermally stable.
All that said, there are of course many other
issues of science and engineering to be
tackled before many applications envisioned
become plausible. Though some (e.g. field emission
displays, better batteried) might be coming
relatively soon.
The operating temperature of a single-electron
transistor is set to a large extent by the
capacitive charging of the dot where the
electron resides. The first single electron
transistors (SET) where fabricated with e.g. scanning
electron microscope (SEM) e-beam writing
technology, and the dots where
consequently quite a bit larger than a single C60
molecule. A big dot has a large capacitance,
therefore the charge of a single electron
produces a very small voltage. Voltage, multiplied
by the electron charge, yields an energy. Converting this energy to temperature, one obtains
a very low temperature for an SEM-defined SET.
However, the charging energy for a single electron
on a buckyball is quite a bit higher. Therefore
it is possible to envision devices that could
operate at room temperature (e.g. the nanotube-based transistors work fine at room
temperature). That said, Paul McEuen's
experiment here
is performed at 1.5 Kelvins (very cold) since they
want to resolve very fine detail in the electron
current/voltage characteristics that are
associated with the vibrations of the buckyball.
(I must admit- I only skimmed the article, so
I might've missed something).
I'm leaving out some details here- the
spacing between quantum mechanical electron
energy levels is also important (it becomes
bigger as the electrons become more confined in
smaller devices).
Short answer- a sufficiently small single-electron device can operate at room
temperature, if properly designed. The real trick
(as mentioned in an earlier comment) is to
integrate more than one device (say, oh a couple
billion) into a useful device.
It turned out that the buckyball, although
nicely round, is too sticky to work well
as a lubricant on the macroscopic scale.
They tend to collect adsorbates from the
air, which then dirties their surfaces.
But a collection of clean buckyballs, when
carefully prepared, forms an ordered
close-packed crystal in which, above about
260 Kelvins (room temperature is 300 K), the
individual balls spin freely about their
lattice positions. So they are very smooth, when
clean.
I did a bit of research on buckyballs (aka C60)
in my grad student and postdoc days at Berkeley.
Slashdot Mirror
Upon reading of the 100+ page license for CSS, I had a thought- DMCA isn't protecting the encryption; it's protecting the license for CSS. Wrap a weak encryption around a product, and only allow legal decryption if you agree to an onerous license. It doesn't matter how weak the encryption is, that's not the point. The point is to force agreement to the terms of the license. This seems to have legal ramifications, since if the purpose of the encryption is not to encrypt, but to activate the DMCA and thereby force the licensing terms, then it's not really encryption; it's a licensing ploy. So perchance then it doesn't fall under DMCA anymore, since the intention of the scheme isn't really encryption but licensing?
Ooops- I meant square root of 19 times for terminal velocity in part 2.
It's a great idea (and ameliorates the excess cow problem as well). Unfortunately, in some studies we performed for BARPA (Bovine Advanced Research Projects Agency) a while back we encountered some serious technical problems:
That said, of course we recommended further study.
Is it good thing for people to spend money on hardware they don't necessarily need? Spending in itself is not a net benefit to society, only spending on stuff that improves quality of life. Not spending extra $$ on computer upgrades could be a good thing for the economy as a whole- that's more resources for something else that might be more important to quality of life.
Micromachines are currently in practical use, at least very simple ones. The most common application (to my limited knowledge) is airbag acceleration sensors. A small silicon cantilever bends under acceleration and a resulting electrical signal sets off the air bag. I very simple micromachine.
Other applications are in the works, but progress on complex machines is difficult since most devices are built with lithography, and lithography naturally produces two-dimensional structures, not three. For example, microfluidics is a potential application, where small tubes and pumps analyze microscopic amounts of fluid for various e.g. chemical/biological assays, etc.
It's a long way from "I can build a gear" to "I can build a gear and a motor and a shaft and connect them to a couple hundred other parts with minimal frictional losses, a power supply, etc. in three dimensions." Interesting things will be coming, but it'll be very incremental as researchers built up a toolset.
Isotopically pure Si could be grown strained just as well as natural silicon. Should be no problem to integrate the two technologies, assuming each was economically feasible on its own.
Interesting. Maybe any 'random' file of finite length is not entirely random- there is information in its length itself. Perhaps this is what allows for your loophole of the (extraordinarily unlikely, but not impossible) random file which is all zeros. This is only possible with a finite length file. An infinite length random file will demonstrably have no patterns.
Doesn't quite work that way, but the difficulty is very subtle. You have to tell the decompressor where in the file the string of zeros starts. A string of 1000 zeroes happens once in 2^1000 bits. So you need a 1000-bit pointer to say where to find it. The key to the impossibility of compressing the truly random file is the true randomness. Any pattern and yes it can be compressed. No pattern at all, and no compression. Cutting the file in two is sneaky, but it's information hiding, not compression. There's information in where you choose to cut the 2^1000 bit file.
This warms the heart of a solid-state physicist... isotopically homogeneous silicon processors would be very cool. I would be amazed if they can do it as cheaply as claimed- isotopic enrichment is usually very expensive (think Manhattan project- a big chunk of the cost was isotopic enrichment of the uranium), but fundamentally there is no barrier to someone finding a cheap way to make it, since the starting material is very cheap (just Si). Rarely do I see the words "electron-phonon coupling" in the press :-). I love it.
Btw, isotopically enriched diamond (one up from Si on the periodic table) is the best thermal conductor known. The lattice vibrates at higher frquencies than silicon (since carbon is lighter and the carbon-carbon bonds are stronger than the Si-Si bonds) which speeds the transport of heat. At low temperatures (like 20 Kelvins) the increase in thermal conductivity with isotopic enrichment is enormous- 10 or 100 fold. At room temperature this drops to roughly two-fold, but that's still a huge win.
There was not so much a rush to judgement as a rush to confirm. I was at Berkeley physics at the time in grad school and *alot* of people tried very hard to reproduce these results. A colleague of mine, top of the field, performed high-powered first principles calculations of hydrogen-loaded palladium. Others scattered across the country were working hard trying to reproduce the result. No careful experiment showed anything remotely like fusion. However, a literature search did reveal some interesting palladium electrochemistry done many decades ago (I unfortunately don't have the reference) which showed a possible heat gain due to electrochemical effects. Nothing one could extract useful power from.
The comparison to superconductivity given in the original writeup is spurious. Yes, it's taking many years to understand the mechanism of high temperature superconductivity. But everyone who tried could clearly reproduced the phenomenom almost immediately. That didn't happen with cold fusion.
Plenty of people with expertise in calorimetry tried to reproduce this experiment. Nothing that required fusion as an explanation was observed. Palladium has some interesting electrochemistry, but there's no evidence of fusion, unfortunately. I wish there had been- it would have been wonderful.
I've seen these issues from many sides: author, reviewer, editor. Totally free access and open review falls down on one important count- the prime currency of academic research reputation (not money; that's not the main impediment here). People in academia could generally make more $$ in industry or business. $$ is not their main motive (except as $$ in the form of grants is a means to research). The motives are to advance knowledge, have fun doing something intersting, and gain the respect of one's peers.
Most of the value in a research journal is in the brand: how difficult is it to publish there, how much prestige is due to the author whose work appears there, how much attention busy readers pay to publications in one journal over another. Without a strong publication record, nearly all researchers will have trouble getting funding, tenure, etc.
An open publication system needs to somehow maintain this means to rate papers based on quality. Otherwise, researchers will continue to submit their best results to Science and Nature, which will then hold copyright.
This could be done: design a semi-open peer review system whereby authors can only publish in the archive if they also provide quality peer reviews, and manage it all through community-wide feedback and also possibly the efforts of professional editors skilled in the field (who would have to be paid). And rate papers based on their importance to the community and the judgement of the referees and readers.
In the end, it sounds like a variant of slashdot itself- self-organized publication ratings with mechanisms in place to minimize abuse. Resistance to abuse is critical- junior researchers under a tenure cloud will have great incentive to work the system.
The page charges at the APS are voluntary. If you don't want to pay, you don't have to. Elsevier might not have page charges, but their subscription prices are typically much, much higher than those of APS.
Many specialized western journals also cost thousands of dollars per year. Of two main journals on Schizophrenia, the one published by the NIH is under $100/yr and the one from a commercial publisher is over $2000/yr.
Journal publsihing is a strange market: the author does not pay for publication so she or he is as likely to submit to an expensive journal as a cheap one. Since the journal then holds the copyright, libraries are forced to pay the exorbitant subscription rate of some commercial journals or else they have a hole in their collection. Commercial publishers figured this out several years ago- their academic publishing arms make a quite high return on investment.
Every individual manuscript published is like a little monopoly for the journal on a slice of knowledge.
I agree with these comments: long-term applications of quantum computing are more likely to come from solid-state based devices, possibly using electron spin or nuclear spins of embedded impurity atoms as qubits. Optical atom traps (and NMR magnets) are very expensive and likely to remain so for quite a while. But in the end it's really too early to say, of course, since we're looking ahead a long way.
This work is impressive, but NMR-based quantum computers have some special scaling problems. The net spin polarization of these liquids at room temperature is very small- about one part in a thousand roughly. This means that the quantum computation is done with only the slightest of distinction between 0 and 1. That can be handled for small computations, but the signal to noise problem becomes severe for larger systems.
The current quantum error correction codes impose a large bit overhead (factor of three or so), which compounds the signal/noise difficulty of the NMR technique.
The nice thing about this method is that the nuclei are well-isolated inside the electron clouds and decohere slowly. However, my guess is that NMR-based quantum computing will be first out of the gates, but will lag behind other implementations before reaching the finish line.
Spam is commercial speech; the political candidate to which you refer is engaging in political speech. Commercial speech in the US is not afforded the same degree of first-amendment protection as is political speech.
Nice points, all. I've thought a little bit about GPS-based traffic systems and I think there's a feasible way to implement it, but it's a different philosophy than speed governors.
Keep all the map db info on the server, where it should be. The GPS in the car simply records a time record of the travel path of the vehicle. This information is occasionally fed to local (or satellite) receivers scattered about. The information is then fed into an algorithm that calculates a monthly bill for moving violations, etc. The central system need not even store the bits for the actual traffic pattern- it could process the traffic pattern (the overpass which-road-am-i-on ambiguities issues raised above can be covered through a time average of speed or by removing the overpass regions from the processing) and then throw it away, storing only the monthly bill. If the driver wants an itemization or to challenge the bill, they go to their car, which is the only device that actually retains the privacy-sensitive info on their traffic patterns.
Of course, the system is open to hacking and backdoors, as would any wide-spread IT system. But a good (open-source?) implementation could help ameliorate these concerns. The personal info must go to the central server, but it need not be stored to disk there. Just converted to XX $$ bill. Ain't perfect, but possibly workable? I'd be interested to hear reactions from those more tech-savvy than me.
Many of the disadvantages of such a system have already been posted here, but also keep in mind the advantages. Saving life. Saving fuel. Saving the environment. Saving time (yes- see next paragaph). Greater freedom (yes- see next paragraph).
One nice irony of the automobile as a symbol of freedom (in U.S.) is that the road system is a nearly perfect example of communism. It's build and managed by a central governmental authority, and access to resources are allocated by queue (ever been in a traffic jam :-) ? ).
Soviet bread line, anyone? Road congestion
to a large extent results because a driver
does not pay the very real cost that (s)he imposes
by slowing down the traffic for everyone else.
That economic failure encourages excess driving.
A GPS-based pricing system could go some way to alleviating that problem. (This line of thought
now splits into multiple interesting threads
on social implications, fairness to the poor, etc.
but only finite time to type). Freedom to
some extent is freedom from the unfair "free-ride"
impositions of others...
Hi,
I am a slashdotter myself occasionally. I am also currently performing research on nanotubes at a major research unversity, with publications in Science, Nature, and other places on nanoscience.
Perhaps the slashdot readership is a bit more diverse than you though. I find it an interesting place to visit, with some very intelligent conversation. (And thanks much to the moderators).
Buckyballs don't have any important current application that I know of (there was some work in nonlinear optical materials, but I don't know if it panned out).
Carbon nanotubes have a pretty good chance of appearing as field emitters in flat-panel displays before too long. They have great conductivity, they're very pointy and extremely robust under the relevant electrical conditions. They are already at the simple prototype stage. Batteries might also pan out (they store lithium quite well) but tubes are still much too expensive for that.
The tubes will very likely find many more commercial applications than buckyballs.
Oh- and there's one application out there right now. Mass-produced (and very low-quality) multiwalled tubes are currently sold to mix into plastic parts to add conductivity to the plastic so that they can be charged up uniformly and electrostatically painted. Nice shiny plastic car bumpers. Not quite an elevator to geosynchronous, but it makes (a little) money.
One can estimate theoretically the ultimate strength of a nanotube be examining the microscopic failure modes, i.e. the ways in which atoms rearrange in response to an external stress (i.e. stretching).
In the case of perfect, defect-free nanotubes, there are two modes that seem to be important. First, the rotation of a single carbon-carbon bond by 90 degrees, which converts a patch of 4 hexagons (remember that carbon atoms are arranged in a chicken-wire or honeycomb pattern on the tube wall) into two pentagons and two heptagons (relevant references are Zhang & Crespi from Penn State in Physical Review Letters and work by Bernholc at NC State and Yacobson at Rice I think, but the exact journal escapes me at the moment). This mode is a plastic distortion of the tube; the tube with the bonds rearranged is a bit longer than it was before.
The second failure mode is for one of the hexagonal rings of carbon atoms to break open, i.e. for a carbon-carbon bond to break. This is a more catastrophic event, in that the tube then quickly breaks near the point of failure. Which way a tube fails may actually depend on how the honeycomb pattern is rolled into a tube shape.
Now that's just the microscopic theory on the ideal, defect-free system. In a real tube, one expects there to be pre-existing defects in the structure. The failure under tension will then be at the defective points
But, since nanotubes are so small, it's plausible that a single tube or bunch of tubes might grow entirely defect-free, in which case one can access the ultimate theoretical failure strength. Experiments on trying to stretch and break single bundles of nanotubes (Lieber's group at Harvard) show that one can extend a nanotube by about 6% of it's length before it breaks. This is in good agreement with the theoretical predictions mentioned above (and it's a legit prediction- the theory came first!). So it appears that in small enough systems, one can attain the theoretical mechanical strength.
Now if one wants to make a space elevator, one's material has to also be resistant to radiation damage, etc. I think a back of the envelope estimate shows carbon nanotubes or diamond nanowires as being in the right ballpark, so long as one allows the structure to taper, but once one factors in the necessary engineering margins and the need to be resistant to damage over long periods (don't want it to fall apart in a year or two :-) it's much
less clear if it's really possible. It's all
in the very very long run, of course.
I should admit- I have not yet read this specific article (New Scientist website is crashing on me) so I can't comment specifically on this current experimental result. My guess is they did a larger-scale version of Lieber's experiment and found that the resulting thread was alot weaker (not surpirising- their structure likely has lots more defects and possibly single tubes don't extend throughout the entire length- they overlap).
OK, the definition of a decoherence-free subspace:
Quantum mechanical wavefunctions are described in terms of their projection onto a set of basis functions. This is exactly analogous to a Fourier transform of a function (i.e. the projection of the function onto a set of sine and cosine function).
As the wavefunction of say an electron evolves with time, the weights of the various basis functions will typically change. If the wavefunction is coupled to other systems (i.e. other electrons, surrounding atoms, molecules, etc.), then the wavefunction becomes very complicated as the pieces that describe the single electron mix up with the pieces describing the other parts of the system. This is termed decoherence.
The subspace referred to in the posting is a subset of the full set of basis functions (like taking a finite bandpass of the Fourier space). For wavefunctions that can be described completely in terms of sums of the basis functions in this subspace, the wavefunctions will maintain their coherence, which means that as they evolve in time, they don't get mixed up with the wavefunctions describing the surrounding environment
This isolation of a subset of the degrees of freedom describing a system (i.e. the decoherence-free subspace) is essential for quantum computing, as a quantum computer uses the subtle correlations within a wavefunction to perform what are essentially massively parallel computations. Should the system decohere, the subtle structures in the wavefunction are lost.
This is a major outstanding problem with nanoelectronics nowadays- the third terminal. It's easy (well, if you have access to about a million dollars worth of e-beam equipment, or a hundred K of scanning probe equipment) to get two contacts onto a nanometer-scale device. The tricky part is the third contact. Typically, the third contact is underneath, the e.g. field-effect lead, and it's very big compared to the rest of the device. One can also do alot with only two contacts, if the device is hysteretic (i.e. has memory of its history).
Many of the molecular electronics groups, realizing the limits to our current ability in contacting and integrating electronic devices on this scale, are concentrating first on implementing very simple memory architectures. Various aspects and approaches to these ideas are being pursued at many places: several groups at Berkeley (mostly physics), Penn State University (chemistry, elec eng, physics), Harvard (chemistry), Notre Dame (physics), Delft (in Netherlands), HP, Rice (chemistry), IBM, and others. No-one knows yet what if any approach will pan out.
- Being in the top full row of the periodic
table, it is rather small, which means that the
electrons feel a strong electrostatic potential
from the nuclear charge, and therefore can be
tightly bound into molecules and more extended
structures. In addition, an accident of the
fundamental constants means that carbon (and nearby elements on the periodic table) can form
multiple bonds to neighboring atoms. One consequence of this is the ability to form continuously bonded atomically thin structures
such as graphite. Graphitic structures can then
be curved into long thin nanotubes. (Another
consequence is the richness of organic
chemistry, and biology).
- Graphite sheets are also rather unusual in
being what are called semi-metals: on the edge
between a metal and a semiconductor. Rolling them
into a tubular shape is actually enough to push
them one way or another into real metals or
real semiconductors in carbon nanotubes.
Both types can even be combined
within a single tube to make diode, etc. on-tube.
- The strong in-plane bonding of carbon in
graphitic strucures (see point 1 above) means
that it is an excellent material for high strength-to-mass applications (if the fibers can
be embedded in an appropriate matrix). This also
means that the structures could be very thermally stable.
All that said, there are of course many other issues of science and engineering to be tackled before many applications envisioned become plausible. Though some (e.g. field emission displays, better batteried) might be coming relatively soon.The operating temperature of a single-electron transistor is set to a large extent by the capacitive charging of the dot where the electron resides. The first single electron transistors (SET) where fabricated with e.g. scanning electron microscope (SEM) e-beam writing technology, and the dots where consequently quite a bit larger than a single C60 molecule. A big dot has a large capacitance, therefore the charge of a single electron produces a very small voltage. Voltage, multiplied by the electron charge, yields an energy. Converting this energy to temperature, one obtains a very low temperature for an SEM-defined SET. However, the charging energy for a single electron on a buckyball is quite a bit higher. Therefore it is possible to envision devices that could operate at room temperature (e.g. the nanotube-based transistors work fine at room temperature). That said, Paul McEuen's experiment here is performed at 1.5 Kelvins (very cold) since they want to resolve very fine detail in the electron current/voltage characteristics that are associated with the vibrations of the buckyball. (I must admit- I only skimmed the article, so I might've missed something).
I'm leaving out some details here- the spacing between quantum mechanical electron energy levels is also important (it becomes bigger as the electrons become more confined in smaller devices).
Short answer- a sufficiently small single-electron device can operate at room temperature, if properly designed. The real trick (as mentioned in an earlier comment) is to integrate more than one device (say, oh a couple billion) into a useful device.
It turned out that the buckyball, although nicely round, is too sticky to work well as a lubricant on the macroscopic scale. They tend to collect adsorbates from the air, which then dirties their surfaces. But a collection of clean buckyballs, when carefully prepared, forms an ordered close-packed crystal in which, above about 260 Kelvins (room temperature is 300 K), the individual balls spin freely about their lattice positions. So they are very smooth, when clean. I did a bit of research on buckyballs (aka C60) in my grad student and postdoc days at Berkeley.