No tech company I worked for ever gave me such instructions: Hazeltine, Litton, Harman, Alesis. We were aware of patents in our fields are worked with patent-holders when appropriate.
Grad students and professors do cutting-edge research, taking the risks.
What risks? Professors are just doing their jobs, keeping busy so that they'll be paid. Creating something patentable is a bonus; there's no risk involved. Grad students are there for an education and a degree, and if they publish they also gain a reputation. Again, creating something patentable is a bonus; there's no risk involved.
Nobody trawls through the existing patents to check for infringement
This is false more often than true. Other than startups, most companies do patent searches as part of research for ways to solve difficult problems. When a patent is found that is by far the best way to implement a solution, there is a decision to be made: do we license the patent, wait for it to expire, design around it, or use it and hope nobody notices? Engineers usually want to make a design that does almost the same thing but avoids infringement. The choice between licensing and use without permission shows the ethical properties of management.
Theoretically, any perfectly band-limited signal, perfectly sampled at least the number of times specified by Nyquist, can be reconstructed perfectly. The samples need not be uniform in time, and the signal may exceed the range of the ADC at times when it is not being sampled. The "only" problems are practical ones: the calculation burden is immense, and errors in timing and voltage measurement are greatly magnified if the samples are clumped so that relatively large periods are unmeasured.
This would provide the same benefits they're claiming, and it's only slightly more impractical: perfect band limiting requires infinite time.
My hidden point is that their great claim (unlimited dynamic range) requires unrealistically precise measurements and severely limited bandwidth. They do quantify the tradeoff, but it's hidden in the details.
There are several problems. A counter is required for each pixel, and each counter will be incremented anywhere from hundreds to tens of thousands of times per exposure. That's a lot of activity on the sensor chip, which means heat and noise. For a 4000x3000 image sensor, there'd be 12 million wires running from pixels to counters. That's a chip construction problem, serious enough that it might make such a chip impossible. Conventional image sensors use analog shifting or multiplexing, resulting in wire counts in the thousands, not millions. (By wires I mean the aluminum traces on the chip.)
There are ways around the wiring problem. One is to build each counter right next to its pixel, but this means valuable image sensing area is instead used by counter logic. Another is to have a chip with multiple active layers, pixel sensors on the surface and digital logic below. I'm not familiar with whether this is currently practical IC technology, but the complexity suggests high cost. Also, at about 200 transistors per counter, this would be a 2.4 billion transistor IC, comparable to many current production CPUs. Not cheap.
Maybe for a smaller video sensor it would be feasible, perhaps at 640x480 the complexity isn't excessive. But we already have 1024x768 shutterless desk cams for under $20, so it looks like there's nothing to be gained for the consumer market. Perhaps your suggestion would make possible a higher video sample rate or more bits of precision per pixel, but that's beyond my current ability to estimate.
Modern semiconductors are very good, and at least for audio it's possible to get components and design circuits with less noise than any microphone.
The subject of noise in tubes and semiconductors is interesting. Tubes have several factors that contribute to their noise, one of which is the heat of the cathode, which yields an effective "noise temperature". The cathode-caused noise temperature is reduced by the first grid (the space charge between the cathode and the first grid tends to smooth the electron flow). A second grid (tetrode) tends to make things worse by collecting high energy electrons that bounce off the plate, while a third (suppressor) grid in a pentode may reduce noise by repelling the bounced electrons back to the plate.
Unlike semiconductors, a tube can't be cooled down to make it quieter.
Their approach reads as if they're effectively (for example) taking a 12 bit conversion and throwing away the upper 4 bits, then inferring the discarded bits when doing reconstruction. There's something more complicated than just that going on, because they say the process runs away (diverges) if the inference is mistaken. The mistake is then corrected and the process would converge with the proper inference. That does, however, mean that several samples worth of output have to be buffered for error correction.
Keeping the example of 8-bit data storage, let's limit them to an 8 bit ADC. It seems now that if the converter saturates, they're allowed to offset the input voltage and try again, repeatedly until they get it right, up to a total of PI * e times. That would seem to be cheating.
Why not just do (for lack of a better name) delta-sampling? At each sample, convert the difference between the current sample and the previous one, and record that data. There are problems with accumulated error that have to be taken care of, but with the type of signal they claim to be handling (erratic pulses), techniques to remove accumulated error shouldn't be too tough.
Perhaps I'm not understanding what they're claiming, but it looks like they're claiming a peculiar solution to a problem better handled in other ways.
The 21st century recessions were due mostly to federal government pressure to loan to insolvent home buyers. That pressure was a result of Democratic party policy then in place, that Republicans were unable to end. The result was inevitable and had nothing to do with your jealousy of rich people.
The people who gained insurance coverage under Obamacare were mostly people who didn't want insurance, and with the end of Obamacare will be free to once again do as they please. You, however, think it's OK to point government guns at their heads and proclaim "Your money or your life."
The AI abilities that would make Asimov's 3 laws relevant are far beyond the realistic goals of AI today. Putting into law things that nobody can currently understand is foolish.
Most tapes were unlabeled and the contents unexamined. The archivist knows nothing about the content of the tapes, not even if they were blank. At best, he's guessing. At worst, he's too lazy to do his job.
Yes, the USPS is subject to stupid laws that make it less efficient. But it also receives huge indirect subsidies which (of course) it's not going to mention on its web site.
Why would you trust the USPS's claims about itself?
The headline and summary are deliberately misleading. Amazon takes advantage of discounts for presorting and local delivery that any entity with enough shipment volume can also take advantage of.
Apparently, those discounts are excessive. As are the discounts for junk mail.
Slashdot Mirror
No tech company I worked for ever gave me such instructions: Hazeltine, Litton, Harman, Alesis. We were aware of patents in our fields are worked with patent-holders when appropriate.
What risks? Professors are just doing their jobs, keeping busy so that they'll be paid. Creating something patentable is a bonus; there's no risk involved. Grad students are there for an education and a degree, and if they publish they also gain a reputation. Again, creating something patentable is a bonus; there's no risk involved.
This is false more often than true. Other than startups, most companies do patent searches as part of research for ways to solve difficult problems. When a patent is found that is by far the best way to implement a solution, there is a decision to be made: do we license the patent, wait for it to expire, design around it, or use it and hope nobody notices? Engineers usually want to make a design that does almost the same thing but avoids infringement. The choice between licensing and use without permission shows the ethical properties of management.
Would you react the same way if someone were killed? Perhaps a member of your family?
Theoretically, any perfectly band-limited signal, perfectly sampled at least the number of times specified by Nyquist, can be reconstructed perfectly. The samples need not be uniform in time, and the signal may exceed the range of the ADC at times when it is not being sampled. The "only" problems are practical ones: the calculation burden is immense, and errors in timing and voltage measurement are greatly magnified if the samples are clumped so that relatively large periods are unmeasured.
This would provide the same benefits they're claiming, and it's only slightly more impractical: perfect band limiting requires infinite time.
My hidden point is that their great claim (unlimited dynamic range) requires unrealistically precise measurements and severely limited bandwidth. They do quantify the tradeoff, but it's hidden in the details.
There are several problems. A counter is required for each pixel, and each counter will be incremented anywhere from hundreds to tens of thousands of times per exposure. That's a lot of activity on the sensor chip, which means heat and noise. For a 4000x3000 image sensor, there'd be 12 million wires running from pixels to counters. That's a chip construction problem, serious enough that it might make such a chip impossible. Conventional image sensors use analog shifting or multiplexing, resulting in wire counts in the thousands, not millions. (By wires I mean the aluminum traces on the chip.)
There are ways around the wiring problem. One is to build each counter right next to its pixel, but this means valuable image sensing area is instead used by counter logic. Another is to have a chip with multiple active layers, pixel sensors on the surface and digital logic below. I'm not familiar with whether this is currently practical IC technology, but the complexity suggests high cost. Also, at about 200 transistors per counter, this would be a 2.4 billion transistor IC, comparable to many current production CPUs. Not cheap.
Maybe for a smaller video sensor it would be feasible, perhaps at 640x480 the complexity isn't excessive. But we already have 1024x768 shutterless desk cams for under $20, so it looks like there's nothing to be gained for the consumer market. Perhaps your suggestion would make possible a higher video sample rate or more bits of precision per pixel, but that's beyond my current ability to estimate.
Modern semiconductors are very good, and at least for audio it's possible to get components and design circuits with less noise than any microphone.
The subject of noise in tubes and semiconductors is interesting. Tubes have several factors that contribute to their noise, one of which is the heat of the cathode, which yields an effective "noise temperature". The cathode-caused noise temperature is reduced by the first grid (the space charge between the cathode and the first grid tends to smooth the electron flow). A second grid (tetrode) tends to make things worse by collecting high energy electrons that bounce off the plate, while a third (suppressor) grid in a pentode may reduce noise by repelling the bounced electrons back to the plate.
Unlike semiconductors, a tube can't be cooled down to make it quieter.
Read up on delta-sigma converters.
Digital in this context means that a signal is represented by an integer. Please stop trying to confuse people.
The article isn't explicit enough.
Their approach reads as if they're effectively (for example) taking a 12 bit conversion and throwing away the upper 4 bits, then inferring the discarded bits when doing reconstruction. There's something more complicated than just that going on, because they say the process runs away (diverges) if the inference is mistaken. The mistake is then corrected and the process would converge with the proper inference. That does, however, mean that several samples worth of output have to be buffered for error correction.
Keeping the example of 8-bit data storage, let's limit them to an 8 bit ADC. It seems now that if the converter saturates, they're allowed to offset the input voltage and try again, repeatedly until they get it right, up to a total of PI * e times. That would seem to be cheating.
Why not just do (for lack of a better name) delta-sampling? At each sample, convert the difference between the current sample and the previous one, and record that data. There are problems with accumulated error that have to be taken care of, but with the type of signal they claim to be handling (erratic pulses), techniques to remove accumulated error shouldn't be too tough.
Perhaps I'm not understanding what they're claiming, but it looks like they're claiming a peculiar solution to a problem better handled in other ways.
The weather radar path is
radar transmitter -> propagation to cloud -> bounce off cloud -> propagation to radar receiver -> radar receiver.
The wifi router path is
wifi transmitter -> propagation to radar receiver -> radar receiver
The losses to the cloud and reflecting off the cloud can both be substantial. That 200 mW is not negligible.
And here's the pronunciation: https://www.youtube.com/watch?v=GuRQl4rAajs
Clutch Cargo is 58 years old now.
Fuses? Does the electrical code even allow fuses now?
Why don't you ask the billions of people who would die during those few years?
They pay their own way by the benefits they bring.
The 21st century recessions were due mostly to federal government pressure to loan to insolvent home buyers. That pressure was a result of Democratic party policy then in place, that Republicans were unable to end. The result was inevitable and had nothing to do with your jealousy of rich people.
The people who gained insurance coverage under Obamacare were mostly people who didn't want insurance, and with the end of Obamacare will be free to once again do as they please. You, however, think it's OK to point government guns at their heads and proclaim "Your money or your life."
Another factor is time. Most companies take about a year to recognize that somebody is working really hard and taking on extra responsibility.
Unlike Musk, Hughes had the decency to become a recluse.
Number 5 contradicts the combined effect of 1 & 2.
The AI abilities that would make Asimov's 3 laws relevant are far beyond the realistic goals of AI today. Putting into law things that nobody can currently understand is foolish.
Most tapes were unlabeled and the contents unexamined. The archivist knows nothing about the content of the tapes, not even if they were blank. At best, he's guessing. At worst, he's too lazy to do his job.
Yes, the USPS is subject to stupid laws that make it less efficient. But it also receives huge indirect subsidies which (of course) it's not going to mention on its web site.
Why would you trust the USPS's claims about itself?
See this: http://fortune.com/2015/03/27/us-postal-service/. Among other things, the USPS is exempt from state and local taxes.
The headline and summary are deliberately misleading. Amazon takes advantage of discounts for presorting and local delivery that any entity with enough shipment volume can also take advantage of.
Apparently, those discounts are excessive. As are the discounts for junk mail.