The music industry has a long history of trying
to get the consumer to pay more for less. Sometimes it works (price premium for CDs), sometimes it doesn't (CD singles).
I have a feeling that this time it won't work; consumers seem to be getting tired of getting ripped off.
Sure, Junkyard Wars is cool, but the new season of Battlebots is cooler.
Since the machines on Battlebots aren't even true bots (they are remote-controlled, not self-controlled) I think it is considerably less cool than Junkyard Wars, where at least the people actually use the right words.
The reason that quantam encryption isn't used everywehere, is that it's so darn hard to detect the spin of single photons. I think it's extremley unlikley that they have figured out how to discern the spin of a stream of photons, over 10 kilomiters, with a 0% error rate (otherwise you've got a bad encryption key) when it can barely be done over inches.
Did you read the article? The Los Alamos people are doing an experiment to attempt to do this. They don't know how to do it yet. That's why it's called an experiment.
If Eve can't intercept the stream and recreate it verbatim back to Bob, then what kind of technology will this satelite have that can violate the laws of physics?
The satellite doesn't break the laws of physics; it doesn't make any measurements, so it doesn't have to recreate the stream of photons.
The initial experiments had an error bar that overlapped zero, but was mostly negative. Later experiments had an error bar that lay entirely below zero. That is, within experimental error, they had measured a negative squared rest mass for the neutrino, implying an imaginary rest mass for the neutrino, which would mean that neutrinos are in fact tachyons.
As one of the authors of the result to which you refer, I can authoritatively say that Cramer managed to get it almost completely wrong.
In fact, there was a paper written by Stephenson that showed that the result could not come from tachyonic neutrinos. In that case, we would have seen a completely different signature.
The most likely physical explanation for the result would have been another unknown particle. Lobashev still believes in that, but I think the evidence has accumulated that there is no significant excess at the endpoint of the tritium spectrum.
Of course, it is worth pointing out that nobody has ever found any error in the original data, and we spent something like 2 years trying to find problems before we published it in the first place!
Since (as was pointed out by another poster) the neutrinos arrive before the main light, this can't be done exactly as you describe.
However, an upper limit on the neutrino mass could be made based on the spread in arrival times for the neutrinos. If the higher-energy neutrinos arrive before the lower-energy ones, then there might be a mass.
Of course, all this is complicated by the supernova itself, which might eject higher-energy neutrinos first, but, given that our models of supernovae are correct, then the spread in arrival times put a limit on the electron neutrino mass of about 19 eV or so.
Terrestrial experiments have put an upper limit on the electron neutrino mass of about 3 eV.
The best guess for the electron neutrino mass from the SNO results is much, much smaller; probably it is less than.01 eV.
About 18 percent of all the "dark matter" in the universe may now be made up of neutrinos.
The value of 18% is an upper limit on the neutrino mass contribution to the total mass of the Universe, not to the dark matter. The lower limit quoted in the paper is 0.1%, so I think it is inaccurate to say that neutrinos make up "about" 18% of the Universe's mass.
sales of singles are down something like 10% in a strengthening economy, which begs the question, "was it Napster?" (my personal opinion : Yes it was).
You haven't happened to go to a record store and look at the prices they charge for singles, have you? There is no way in hell I am paying $6.99 for a single song!
Maybe the music industry should look somewhere else for an explanation of the sudden dip.
Dumbing down technical content for the clueless is on thing, but actual misinformation is another. The article says:
Besides water, the other main cooling option for heavy-duty overclockers is a "peltier," or an electrically charged heat sink that acts as a thermal transfer device. The top of the peltier can reach temperatures up to 600 degrees, while the bottom side that touches the processor is frosty cold.
An "electrically charged heat sink." Is that hilarious or what?
And proving that they have mass would help us account for a lot of the 'missing' or 'dark' matter that we think should exist.
Um, no. In order to make up the "missing mass," neutrinos would have to have a mass of at least 10 eV, which they don't. The upper limit on the mass of the electron neutrino is less than a few eV, as determined by tritium beta-decay experiments.
The masses claimed by neutrino oscillation experiments are much, much smaller, like 1e-5 eV or less. Of course, the oscillation experiments don't observe the mass directly; they can see the difference in masses between two neutrino flavors. In any case, neutrinos cannot make up the missing mass of the Universe.
Neutrinos are incredibly important cosmologically, anyway. The density of neutrinos in our galaxy is about 1e9 per cc. That includes inside you! Unlike many other particles, the probability of a neutrino interacting goes down as the neutrino energy goes down, so these primordial neutrinos are undetectable.
Another cool and interesting factoid: when a supernova explodes, only about 1% of the total energy is emitted as light and kinetic energy. The other 99% is neutrinos. So even though a supernova can be optically brighter than the rest of its galaxy put together, what you see pales in comparison to the neutrino flux!
The whole story will indeed have to wait a few years, though, since nailing down the details of the neutral current vs. charged current measurement involves adding a lot of salt to the heavy water (the better to capture loose neutrons resulting from the interactions).
Actually, the detection or neutrons from the neutral-current interactions at SNO will use helium-3 tubes, not salt. The two methods were competing, but the 3He tubes leave the D2O pure, which helps keep the background down.
The applications that are deployed on these types of systems are "embarassingly parallel", i.e.
specifically designed to be largely independent of inter-process synchronization, which makes them immune to Amdahl's law.
You are completely wrong. The problems solved by these computers tend to be extremely difficult to parallelize. A good example is a 3D hydro code with radiation, in which every cell interacts with every other cell at every iteration.
For an example of an "embarrassingly parallel" problem, consider Monte Carlo radiation transport, such as is solved by the Peregrine project, which is located at LLNL, the same lab as ASCII White.
It's too bad this article came out this week; just two weeks ago was LLNL's Family Days, when (if you had a friend who works there) you could get in to see ASCII White.
Disclaimer: I work at LLNL and I worked on Peregrine
Generally speaking, watermarking techniques are designed so that re-creation of A from A' is not possible without A. That's because the generation of A' from A is via a non-invertible process.
However: if the watermark in A' can be detected, then it can be altered (in this case, made undetectable) in such a way that only the parts of A that were altered to form A' are again altered to transform A' to A".
In that case, one would expect the additional degradation that comes from making A" to be of the same magnitude as the original degradation of A'. Actually, it should be less; a naive analysis would expect the degradation of A" to be about sqrt(2) times the degradation of A'. So if the watermarking system is designed so that A' is perceptually indistinguishable from A (for large values of "indistinguishable"), then A" should be pretty good, too.
In other words: if the watermarking schemes have been broken at all then it is possible to break them in such a way that the music is not significantly degraded.
No- actually it cannot get the _corners_ of the square wave but it can get the vertical part substantially more vertical than DVD audio.
You're right that the described format has a much higher frequency response; that response does not have a sharp falloff, as one would see in a PCM-encoding format, but the frequency response becomes less and less accurate for higher frequencies.
The problem with a 1-bit stream is the accuracy, not the frequency response. For a given signal amplitude, the accuracy of a 1-bit stream goes like the square root of the number of samples. So to get a 1% accuracy, the player needs to read 10,000 samples. At a sampling rate of 2.82 MHz, that takes about 3 ms!
So there is a tradeoff -- you gain frequency response at the expense of signal accuracy. In CD audio, you have 16 bits, or an accuracy of about 1 in 64,000, but in this new format, the relative amplitude of the waveform is only good to about 2% averaged over a millisecond.
By the way, you can use GPLed software, privately, any way you want to. You just can't sell the result. As I read it, DC is telling me how to use the CueCat privately.
"People can't expect to take one of our devices and run it through their own engine," he said. "There are boundaries.... It must still do what it was built for,but they can extend its functionality."
He really doesn't get it. The CueCat is <Cartman> mine </Cartman>, not his. They gave it to me. I can run its output through whatever I want.
It's the attitude that DC has that they somehow still own the CueCat that really irks me.
If i understand the watermarking correctly, it is based on the fact that some parts (frequencies?) of the audio signal are inaudible to human listeners, so a watermark signal there wouldn't degrade the sound. Isn't it quite trivial to generate some noise/random signals at the *modulated* frequencies *of* the watermark signal, or just read the watermark signal, invert it and put it back? What am i missing?
Well, you don't understand understand watermarking correctly. The watermarking schemes used by SDMI are all spread-spectrum techniques in which the watermark is detected by correlation of the signal with the (known) watermark vector. The watermark, then, goes into every frequency, including audible ones. The idea is to make the changes in the amplitudes of the frequencies small enough you can't hear them but large enough so that, even if the signal were altered as you suggest, a correlation detector would still see the watermark.
But that's not to say that breaking watermarking schemes isn't pretty straightforward. If you can discover the key that was used to embed the watermark, then you can either (a) remove it, or (b) fake your own.
Even if you don't know the key, I claim that if the watermark is robust against frequency-domain quantization (which it needs to be for MP3) then you can still determine it by a so-called "hill-climbing" technique. As far as I can tell, there is no mathematical way around that conundrum.
I don't understand the SDMI scheme very well. Either it won't let you play songs with watermarks, or it won't let you play songs without watermarks. Either way, it's not too tough to break.
I tried to suggest my own scheme to SDMI, in which the music would be reversibly degraded and distributed for free, like shareware. It took care of many of those thorny issues about fair use, etc., but apparently it didn't leave enough control in the hands of the record companies, because they didn't go for it.
Didn't matter all that much, anyway, because my scheme could still be broken by capturing the decoded input, just like the existing one can.
p.s. since it's legal to send a cease and desist letter for anything, can we get a slash lawyer to write a cease-and-desist letter to DC that asks them to cease and desist being assholes? Then all of us can mail them a copy...
Actually, isn't harrassment by baseless legal threats barratry? And isn't that enough to get lawyers disciplined by the bar? Perhaps the Texas
bar would be interested to hear about these letters...
I've heard that referred to as "audiojacking". Frankly, I don't see this as a credible solution to the problem: transmitting the signal over an 1/8 inch stereo cable represents conversion to an analog signal, with concomitant signal degredation.
The degradation from the cable should be negligible; the degradation you should worry about is from the ADC and amplifier electronics in the sound card.
College is not about acquiring job skills, people. It's about learning how to think and how to learn.
I was tempted to skip graduate school for a well-paying job, but I stayed with it. Boy, am I glad I did! There's a lot more to a good career than a paycheck, and the more education you have, the more options for fun work you get.
Slashdot Mirror
The music industry has a long history of trying to get the consumer to pay more for less. Sometimes it works (price premium for CDs), sometimes it doesn't (CD singles).
I have a feeling that this time it won't work; consumers seem to be getting tired of getting ripped off.
Really? All the Win2k users you know are running IIS? That seems weird. Maybe your friends are just particularly stupid.
Since the machines on Battlebots aren't even true bots (they are remote-controlled, not self-controlled) I think it is considerably less cool than Junkyard Wars, where at least the people actually use the right words.
Sure, I love Carl Hiassen, and Skink is hilarious, but the original eco-terrorists were invented by Edward Abbey -- The Monkey Wrench Gang.
Did you read the article? The Los Alamos people are doing an experiment to attempt to do this. They don't know how to do it yet. That's why it's called an experiment.
The satellite doesn't break the laws of physics; it doesn't make any measurements, so it doesn't have to recreate the stream of photons.
I don't think so, since the lepton flavor eigenstate has to be a linear combination of the mass eigenstates. Interesting idea, though.
As one of the authors of the result to which you refer, I can authoritatively say that Cramer managed to get it almost completely wrong.
In fact, there was a paper written by Stephenson that showed that the result could not come from tachyonic neutrinos. In that case, we would have seen a completely different signature.
The most likely physical explanation for the result would have been another unknown particle. Lobashev still believes in that, but I think the evidence has accumulated that there is no significant excess at the endpoint of the tritium spectrum.
Of course, it is worth pointing out that nobody has ever found any error in the original data, and we spent something like 2 years trying to find problems before we published it in the first place!
However, an upper limit on the neutrino mass could be made based on the spread in arrival times for the neutrinos. If the higher-energy neutrinos arrive before the lower-energy ones, then there might be a mass.
Of course, all this is complicated by the supernova itself, which might eject higher-energy neutrinos first, but, given that our models of supernovae are correct, then the spread in arrival times put a limit on the electron neutrino mass of about 19 eV or so.
Terrestrial experiments have put an upper limit on the electron neutrino mass of about 3 eV.
The best guess for the electron neutrino mass from the SNO results is much, much smaller; probably it is less than .01 eV.
You haven't happened to go to a record store and look at the prices they charge for singles, have you? There is no way in hell I am paying $6.99 for a single song!
Maybe the music industry should look somewhere else for an explanation of the sudden dip.
Dumbing down technical content for the clueless is on thing, but actual misinformation is another. The article says:
An "electrically charged heat sink." Is that hilarious or what?
Um, no. In order to make up the "missing mass," neutrinos would have to have a mass of at least 10 eV, which they don't. The upper limit on the mass of the electron neutrino is less than a few eV, as determined by tritium beta-decay experiments.
The masses claimed by neutrino oscillation experiments are much, much smaller, like 1e-5 eV or less. Of course, the oscillation experiments don't observe the mass directly; they can see the difference in masses between two neutrino flavors. In any case, neutrinos cannot make up the missing mass of the Universe.
Neutrinos are incredibly important cosmologically, anyway. The density of neutrinos in our galaxy is about 1e9 per cc. That includes inside you! Unlike many other particles, the probability of a neutrino interacting goes down as the neutrino energy goes down, so these primordial neutrinos are undetectable.
Another cool and interesting factoid: when a supernova explodes, only about 1% of the total energy is emitted as light and kinetic energy. The other 99% is neutrinos. So even though a supernova can be optically brighter than the rest of its galaxy put together, what you see pales in comparison to the neutrino flux!
Actually, the detection or neutrons from the neutral-current interactions at SNO will use helium-3 tubes, not salt. The two methods were competing, but the 3He tubes leave the D2O pure, which helps keep the background down.
I think it was Hubbard and Campbell, the famous editor/publisher. And I also believe it is true.
You are completely wrong. The problems solved by these computers tend to be extremely difficult to parallelize. A good example is a 3D hydro code with radiation, in which every cell interacts with every other cell at every iteration.
For an example of an "embarrassingly parallel" problem, consider Monte Carlo radiation transport, such as is solved by the Peregrine project, which is located at LLNL, the same lab as ASCII White.
It's too bad this article came out this week; just two weeks ago was LLNL's Family Days, when (if you had a friend who works there) you could get in to see ASCII White.
Disclaimer: I work at LLNL and I worked on Peregrine
Generally speaking, watermarking techniques are designed so that re-creation of A from A' is not possible without A. That's because the generation of A' from A is via a non-invertible process.
However: if the watermark in A' can be detected, then it can be altered (in this case, made undetectable) in such a way that only the parts of A that were altered to form A' are again altered to transform A' to A".
In that case, one would expect the additional degradation that comes from making A" to be of the same magnitude as the original degradation of A'. Actually, it should be less; a naive analysis would expect the degradation of A" to be about sqrt(2) times the degradation of A'. So if the watermarking system is designed so that A' is perceptually indistinguishable from A (for large values of "indistinguishable"), then A" should be pretty good, too.
In other words: if the watermarking schemes have been broken at all then it is possible to break them in such a way that the music is not significantly degraded.
You're right that the described format has a much higher frequency response; that response does not have a sharp falloff, as one would see in a PCM-encoding format, but the frequency response becomes less and less accurate for higher frequencies.
The problem with a 1-bit stream is the accuracy, not the frequency response. For a given signal amplitude, the accuracy of a 1-bit stream goes like the square root of the number of samples. So to get a 1% accuracy, the player needs to read 10,000 samples. At a sampling rate of 2.82 MHz, that takes about 3 ms!
So there is a tradeoff -- you gain frequency response at the expense of signal accuracy. In CD audio, you have 16 bits, or an accuracy of about 1 in 64,000, but in this new format, the relative amplitude of the waveform is only good to about 2% averaged over a millisecond.
GPLed source == software.
CueCat == hardware.
Duh.
By the way, you can use GPLed software, privately, any way you want to. You just can't sell the result. As I read it, DC is telling me how to use the CueCat privately.
Later on, Davis says:
He really doesn't get it. The CueCat is <Cartman> mine </Cartman>, not his. They gave it to me. I can run its output through whatever I want.
It's the attitude that DC has that they somehow still own the CueCat that really irks me.
Well, you don't understand understand watermarking correctly. The watermarking schemes used by SDMI are all spread-spectrum techniques in which the watermark is detected by correlation of the signal with the (known) watermark vector. The watermark, then, goes into every frequency, including audible ones. The idea is to make the changes in the amplitudes of the frequencies small enough you can't hear them but large enough so that, even if the signal were altered as you suggest, a correlation detector would still see the watermark.
But that's not to say that breaking watermarking schemes isn't pretty straightforward. If you can discover the key that was used to embed the watermark, then you can either (a) remove it, or (b) fake your own.
Even if you don't know the key, I claim that if the watermark is robust against frequency-domain quantization (which it needs to be for MP3) then you can still determine it by a so-called "hill-climbing" technique. As far as I can tell, there is no mathematical way around that conundrum.
I don't understand the SDMI scheme very well. Either it won't let you play songs with watermarks, or it won't let you play songs without watermarks. Either way, it's not too tough to break.
I tried to suggest my own scheme to SDMI, in which the music would be reversibly degraded and distributed for free, like shareware. It took care of many of those thorny issues about fair use, etc., but apparently it didn't leave enough control in the hands of the record companies, because they didn't go for it.
Didn't matter all that much, anyway, because my scheme could still be broken by capturing the decoded input, just like the existing one can.
Given the number of missing features, this release sounds more like an alpha than a beta to me.
College is not about acquiring job skills, people. It's about learning how to think and how to learn.
I was tempted to skip graduate school for a well-paying job, but I stayed with it. Boy, am I glad I did! There's a lot more to a good career than a paycheck, and the more education you have, the more options for fun work you get.