I did a Ph.D. on the use of preattentive perception (read "subliminal") on just-in-time memory support. This was the "Memory Glasses" project that got a bunch of media attention a few years ago -- you may have even seen me pitching it to Alan Alada on PBS's Scientific American Frontiers "you can make it on your own" episode.
The long and short of it is that, yes -- properly encoded, "subliminal messages" can jog your memory, but no, they don't otherwise work as sug,gestions or influence your behavior. If you're curious, you can actually read my dissertation on the Memory Glasses and find out more.
There was a lot of hype in the 70's and 80's about the evils of subliminal marketing, but it was all based on junk science with forged data.
One limit is round-trip signal time, which for a clocked architecture (rather than asynchronous) imposes a maximum speed-of-light limit on clock rates in the GHz for reasonable sized chips (or presumably collections of molecular processors). Asynchronous logic isn't limited this way, but has its own problems.
Massive parallelism doesn't beat Moore's law, because Moore's law is an exponential increase, and parallelism gives you only a linear increase with # of processors at best. Once we run into the miniaturization wall (and we will, eventually) we would need to build exponentially bigger and more power-hungry computers to keep up with Moore's law.
Quantum computers are one solution, since they operate in an exponentially larger mathematical space than their physical resources (n quantum bits = 2^n classical bits), but no-one has yet built a practical quantum computer, and this may turn out to be hard. Great for cryptanalysis, though...:)
As far as I can tell crash/virus immunity is pure nonsense -- In a formal sense, a computer that can't crash would need to solve the halting problem (or else how would it know that the computer had crashed or was simply taking a long time to execute an algorithm) which is provably impossible. Likewise, a virus is simply another type of program. In order to make a machine completely immune to viruses you have to make it immune to programming, and you don't have a computer any more.
I'm guessing that they meant that if you have millions of processors, crashing/hanging a few doesn't make a difference. I don't buy this, because having all the processors in the world doesn't make a difference if your code won't run reliably on one. Put another way, sucky software still sucks no matter how many computers you run it on. Likewise, I could easily imagine a virus spreading through the entire molecular computer and rendering all of those millions of processors useless.
The small size and low-power consumption of molecular computers would be great, but there is a long way to go from building a few molecular logic gates in a laboratory and constructing a real machine capable of doing interesting work. My guess is that etched chips are here to stay for the next ten years or so, but if someone hands me a working molecular computer tomorrow I'll start work on the Linux port right away.:)
There will be another Linux-based multimedia wearable demoed at SIGGRAPH this year. It will do location-based context awareness, real-time video processing, and hardware-accelerated 3D graphics. Cooler still, it fits into a nicely tailored vest/jacket combo, and will use an embroidered fabric keypad (conductive thread/capacitive coupling) for input.
Unlike the wearable in this story, it isn't currently set up to broadcast video back to a base station. Instead, the video input is used in conjunction with a small head-mounted camera to do object recognition for annotation (assuming we get the vision code ported in time). Since the demo will run in-doors, we are using IR beacons rather than GPS for the location data, since the chance of getting a clean GPS signal inside the LA convention center is zero.
As the user wanders around the exhibition floor, the wearable will annotate the environment with 3D and 2D content, and relay information back to a base-station using 10Mbit wireless networking. Annotation will include web pages, which the user will be able to brows and navigate using the fabric keyboard. The system weighs slightly more than a laptop, but the weight is distributed throughout the ergonomically design vest; the only obvious sign that the user is wearing a computer is the HMD, which in this case is full-color VGA resolution.
Attribution time:
The demo application is "City of News" by Flavia Sparacino, much of the hardware hacking is being done by Steve Schwartz, and I'm working on the localization system so that the wearable knows where it is (and hopefully what it's looking at). Sandy Pentland heads up the Vision and Modeling Group at the MIT Media Lab where this work is being done, and we are also getting help from Thad Starner who was one of the early wearable pioneers at the lab before becoming a professor at Georgia Tech. Numerous other MIT students are also contributing to this project.
More information on wearables at MIT can be found at The MIT Wearable Computing Web Page, although this project does not yet have a link.
I should say to begin with that I'm not a specialist in the field of human genetics or biotechnology, but part of this story strikes me as a little implausible. Ordinarily I trust the BBC a great deal, so perhaps there is a qualified biochemist/biotechnologist out there who can confirm the feasibility of this for me.
What worries me is the use of a cow egg and human DNA. This seems a little like trying to use an iMac to run SGI binaries; it is my understanding there are significant differences between the internal "hardware" (organels and biochemistry) of cells from different species, and our evolutionary lineage split off from cattle many millions of years ago.
This is part of what makes the "Jurassic Park" scenario unlikely; even if we could extract and reconstruct a dinosaur's DNA, there's no guarantee it would "execute" properly on a modern reptile or bird cell.
It's interesting that Micro$oft is able to sign big contracts with European governments. You'd think Europe would be a little more selective about choosing which elements of American culture to import (along with the required McDonald's French fries and Hollywood movies).
It's good to see that free software advocacy is alive and well in Deutschland.
Each bit added to a number makes it exponentially harder to factor using a classical computer. Factoring techniques will improve, but there is no reason to believe anyone will ever develop a polynomial time factoring algorithm for a classical (non-quantum) machine.
Factoring large numbers in polynomial time is one of only two known algorithms for quantum computers. A modest quantum computer, if it could be built, would crack every prime-factorization cryptosystem currently in use (which is everything, essentially).
Researchers here at the Media Lab have already built a two quantum-bit bulk spin-resonance quantum computer. There are significant technical challenges to building a quantum computer large enough for cryptanalysis, but there is currently no reason to believe that this is impossible. If it can be done, it will almost certainly be done in the next five to ten years; major governments and private companies all over the world are pouring lots of money into this effort.
RSA is good, but we need to start developing cryptography algorithms which aren't factoring based (or reducible to prime factoring), and we need to start NOW. If there isn't a strong, widely-available non-factoring crypto implementation if/when the quantum computing breakthrough happens, all hell is going to break loose.
You're worried about the banks dropping a few transactions on Y2K? How about someone spoofing the Federal Reserve Banking Netowrk and bringing the global economy to its knees.
It's amazing that something as revolutionary as the single chip computer could come out of an engineer staring at thirteen separate schematics and saying, "Ok, but what about doing this with one chip?" And then being in the right circumstances to do it.
The single-chip CPU is arguably the most important development of late 20th century, and it's exponential improvement (Moore's Law) is what drives the information economy. So what happens when Moore's law runs out?
If current trends are projected forward, by 2020 a bit of memory will be a single electron transistor, traces will be one molecule wide, and the cost of the fabrication plant will be the GNP of the planet. The speed of light imposes practical limits on how large you can make a chip and how fast you can clock one. This is why we'll have GHz chips, but fundamental physical laws prevent THz chips.
More importantly, the physical limits that shut down THz electronic computers apply to _any_ classical computing architecture; optical computing and other exotic technology can't beat the speed of light, or single-particle storage problems.
You can't win by going to SMP, because at best you get a linear increase with each processor; exponential increases in power require exponential increases in processor number, which require exponential increases in space and power consumption.
The only basis in physics for continuing Moore's law past classical computing is quantum computing. In a quantum computer N quantum bits (qbits) equals 2^N classical bits. This allows you to build a computer which scales exponentially with the physical resources of the computer. Quantum computing isn't a solved problem, but if and when it is it will be a revolution as big as the first single-chip CPU.
Hot fusion is a hard problem, and will probably stay that way for many years to come. It will change the world when it's economically feasible, but it will always be a big, expensive thing to build and maintain (keeping the sun in a jar is not cheap). Forget about the developing world sharing in the hot fusion boom (unless someone there gets a hold of a fusion bomb and threatens the energy rich nations of the world with a different kind of boom).
Cold fusion would be fantastic, because you could do that in your basement, filtering the heavy hydrogen out of your tap water. This would be an even bigger win, because the whole world could afford it. Of course, cold fusion seems even less likely than hot right now.
Short run we should focus on moving to cleaner burning fossil fuels like natural gas, and slightly longer run we should think about converting to a fuel-cell based hydrogen economy. There are big efficiency and environmental wins to be had, without trying to contain a solar furnace in a magnetic bottle.
The problem is that there is a significant chance that an extinction-class impact will happen within our lifetimes. The chance isn't large, but it has been calculated by astronomers to be approximately equal to the chance of the average person dying in a commercial airline crash.
The number of people actively engaged in looking for these killer rocks is fewer than the number of people who work in your local McDonald's. That's the _global_ total.
The reason for this is a lack of funding. It is pure coincidence that this particular rock was observed; the number of known-trajectory asteroids is small compared to the total number in the solar system, and most of the sky is unwatched most of the time.
That's why the chance of this asteroid hitting us on a near approach is about the same as being hit by one we haven't observed.
Lots of governments spend lots of money ensuring airline safety. I don't have the numbers at hand, but it is my understanding that a fairly comprehensive sky survey would cost significantly less.
Personally, I'd feel awfully silly being killed by a rock we could have nudged in a different direction with a few years of warning. If we discover one with only a few months of lead time, NASA says we're pretty much screwed.
I'm writing a thesis, so I'll try to make this short.:)
There are essentially two ways of expressing an idea: formal and informal language.
Formal language is mathematical notation, computer code, or any other extremely well defined, syntactically precise expression of an idea. An expression in formal language is intended to precisely express an idea or set of relationships in a way that is both rigorous and unambiguous. A C program, Maxwell's equations, and the axioms of set theory are all examples of formal language.
In some fields of human endeavor, I can get away without using formal language to convey an idea -- the study of French literature or poetry, for instance. In other areas, such as number theory or cryptography I have no choice but to use formal language to express formal ideas about which I must make very specific, technical claims.
I can talk about encryption in an informal way, but the only way I can say exactly what the blowfish cipher is doing is by providing an algorithmic description of its operation. Such a description is absolutely necessary, because any informal description could lead to multiple interpretations of its operation, which could in turn lead to differing analyses of its cryptographic strength.
Therefor, I cannot talk about cryptography (in the sense of a conversation between experts, or a conversation between teacher and student) without resorting to formal language if my ideas are to be understood clearly. Source code or an algorithmic description _by_itself_ may not be speech, but speech on the subject of mathematics, science, or computer science _is_not_possible_ without formal language.
Unless the government seeks to ban education in the field of cryptography, number theory, or the foundations of mathematics and computer science, I must be able to freely publish formal and informal descriptions of my work. Simply because someone could translate those descriptions into a functioning cryptographic system, for instance, doesn't change the necessity of this type of exchange if education and informed discussion are to take place.
A lot of the claims made in the article are misleading or overblown. The idea of using very short pulses for data transmission is not new, and as someone has already pointed out this is merely a special case of spread spectrum encoding.
First: An extremely short pulse approximates a delta function, which has infinite frequency content; "DC to daylight." This is still a form of RF transmission, it just happens that you are dumping energy into a very wide range of frequencies.
Second: Transmissions using this technique _do_ interfere with other RF transmissions. In fact, they interfere with _all_ other transmissions, but that interference is spread over the entire spectrum so it does not interfere with any one frequency very strongly (this raises FCC regulatory questions). In addition, a time-domain spread spectrum encoding makes the likelihood of interfering with another pulsed time-domain spread-spectrum transmission very small, if a good spreading algorithm is chosen.
Third: This is not a new idea (we were looking at this a few months ago for a data transmission application) and there is a reason why this hasn't been widely implemented: timing. In order to receive a pulsed time-domain spread-spectrum signal, you must synchronize your receiver's spread-spectrum decoder to the transmitter's encoder. The shorter the pulses, the more exact the timing and the more difficult this synchronization becomes.
Here is an analogy:
Imagine transmitting a signal by encoding it as a time-varying sequence of baseballs being fed to a pitching machine. The receiver catches the balls, decodes the sequence and reconstructs the signal. If the transmitter is the only one pitching, the task of decoding is easy.
The problem is, the transmitter is not the only one feeding the pitching machine -- the noise in the environment is also feeding balls in. The best way to encode the signal to avoid any particular noise source (and to avoid interfering with anyone else) is to make the encoding look as random as possible, which is what spread-spectrum encoding is all about.
The resulting stream of baseballs looks random, since it is a combination of a spread-spectrum signal and random interference. In order to decode the signal, you want to catch only the balls that represent the signal.
In order to do this, you install a shutter in front of the receiver -- the spread spectrum decoder -- which will only let the "signal" balls through. This requires the decoder driving the shutter to be exactly synchronized with the encoder.
As the pulses become narrower, the "balls" are coming faster and timing the shutter must become more exact to exclude non-signal balls. If a non-signal ball passes through the shutter (or a signal ball is missed), the error will break the syncronization between the tranmitter and receiver. Narrow pulses also make it more difficult to lock the receiver's decoder to the transmitter's encoder in the first place. Once the pulses become short enough, maintinaing synchronization becomes almost impossible without an additional, non-spread communication channel. If an additional, non-spread chanel is used, then you are back to the problems of ordinary RF transmission.
There is great potential in this technology, but the technical chalenges (and regulatory hurdles) are large.
They got it mostly right -- interesting that the somewhat tricky ideas of open source (value != $$, competitive advantage to using non-proprietary code and giving stuff away) are finally permeating the main-stream media. Somewhat ironic to have a mag titled "the economist" touting the value of free software...
Slashdot Mirror
No, there is no cause for concern.
I did a Ph.D. on the use of preattentive perception (read "subliminal") on just-in-time memory support. This was the "Memory Glasses" project that got a bunch of media attention a few years ago -- you may have even seen me pitching it to Alan Alada on PBS's Scientific American Frontiers "you can make it on your own" episode.
The long and short of it is that, yes -- properly encoded, "subliminal messages" can jog your memory, but no, they don't otherwise work as sug,gestions or influence your behavior. If you're curious, you can actually read my dissertation on the Memory Glasses and find out more.
There was a lot of hype in the 70's and 80's about the evils of subliminal marketing, but it was all based on junk science with forged data.
One limit is round-trip signal time, which for a clocked architecture (rather than asynchronous) imposes a maximum speed-of-light limit on clock rates in the GHz for reasonable sized chips (or presumably collections of molecular processors). Asynchronous logic isn't limited this way, but has its own problems.
:)
Massive parallelism doesn't beat Moore's law, because Moore's law is an exponential increase, and parallelism gives you only a linear increase with # of processors at best. Once we run into the miniaturization wall (and we will, eventually) we would need to build exponentially bigger and more power-hungry computers to keep up with Moore's law.
Quantum computers are one solution, since they operate in an exponentially larger mathematical space than their physical resources (n quantum bits = 2^n classical bits), but no-one has yet built a practical quantum computer, and this may turn out to be hard. Great for cryptanalysis, though...
As far as I can tell crash/virus immunity is pure nonsense -- In a formal sense, a computer that can't crash would need to solve the halting problem (or else how would it know that the computer had crashed or was simply taking a long time to execute an algorithm) which is provably impossible. Likewise, a virus is simply another type of program. In order to make a machine completely immune to viruses you have to make it immune to programming, and you don't have a computer any more.
:)
I'm guessing that they meant that if you have millions of processors, crashing/hanging a few doesn't make a difference. I don't buy this, because having all the processors in the world doesn't make a difference if your code won't run reliably on one. Put another way, sucky software still sucks no matter how many computers you run it on. Likewise, I could easily imagine a virus spreading through the entire molecular computer and rendering all of those millions of processors useless.
The small size and low-power consumption of molecular computers would be great, but there is a long way to go from building a few molecular logic gates in a laboratory and constructing a real machine capable of doing interesting work. My guess is that etched chips are here to stay for the next ten years or so, but if someone hands me a working molecular computer tomorrow I'll start work on the Linux port right away.
There will be another Linux-based multimedia wearable demoed at
SIGGRAPH this year. It will do location-based context awareness,
real-time video processing, and hardware-accelerated 3D graphics. Cooler still,
it fits into a nicely tailored vest/jacket combo, and will use an
embroidered fabric keypad (conductive thread/capacitive coupling)
for input.
Unlike the wearable in this story, it isn't currently set up to
broadcast video back to a base station. Instead, the video input is
used in conjunction with a small head-mounted camera to do object
recognition for annotation (assuming we get the vision code ported in
time). Since the demo will run in-doors, we are using IR beacons
rather than GPS for the location data, since the chance of getting a
clean GPS signal inside the LA convention center is zero.
As the user wanders around the exhibition floor, the wearable will
annotate the environment with 3D and 2D content, and relay information
back to a base-station using 10Mbit wireless networking. Annotation
will include web pages, which the user will be able to brows and
navigate using the fabric keyboard. The system weighs slightly more
than a laptop, but the weight is distributed throughout the
ergonomically design vest; the only obvious sign that the user is
wearing a computer is the HMD, which in this case is full-color VGA
resolution.
Attribution time:
The demo application is "City of News" by Flavia Sparacino, much of
the hardware hacking is being done by Steve Schwartz, and I'm working
on the localization system so that the wearable knows where it is (and
hopefully what it's looking at). Sandy Pentland heads up the Vision
and Modeling Group at the MIT Media Lab where this work is being done,
and we are also getting help from Thad Starner who was one of the early wearable pioneers at the lab before becoming a professor at Georgia
Tech. Numerous other MIT students are also contributing to this
project.
More information on wearables at MIT can be found at The MIT Wearable
Computing Web Page, although this project does not yet have a
link.
I should say to begin with that I'm not a specialist in the field of human genetics or biotechnology, but part of this story strikes me as a little implausible. Ordinarily I trust the BBC a great deal, so perhaps there is a qualified biochemist/biotechnologist out there who can confirm the feasibility of this for me.
What worries me is the use of a cow egg and human DNA. This seems a little like trying to use an iMac to run SGI binaries; it is my understanding there are significant differences between the internal "hardware" (organels and biochemistry) of cells from different species, and our evolutionary lineage split off from cattle many millions of years ago.
This is part of what makes the "Jurassic Park" scenario unlikely; even if we could extract and reconstruct a dinosaur's DNA, there's no guarantee it would "execute" properly on a modern reptile or bird cell.
It's interesting that Micro$oft is able to sign big contracts with European governments. You'd think Europe would be a little more selective about choosing which elements of American culture to import (along with the required McDonald's French fries and Hollywood movies).
It's good to see that free software advocacy is alive and well in Deutschland.
Each bit added to a number makes it exponentially harder to factor using a classical computer. Factoring techniques will improve, but there is no reason to believe anyone will ever develop a polynomial time factoring algorithm for a classical (non-quantum) machine.
Factoring large numbers in polynomial time is one of only two known algorithms for quantum computers. A modest quantum computer, if it could be built, would crack every prime-factorization cryptosystem currently in use (which is everything, essentially).
Researchers here at the Media Lab have already built a two quantum-bit bulk spin-resonance quantum computer. There are significant technical challenges to building a quantum computer large enough for cryptanalysis, but there is currently no reason to believe that this is impossible. If it can be done, it will almost certainly be done in the next five to ten years; major governments and private companies all over the world are pouring lots of money into this effort.
RSA is good, but we need to start developing cryptography algorithms which aren't factoring based (or reducible to prime factoring), and we need to start NOW. If there isn't a strong, widely-available non-factoring crypto implementation if/when the quantum computing breakthrough happens, all hell is going to break loose.
You're worried about the banks dropping a few transactions on Y2K? How about someone spoofing the Federal Reserve Banking Netowrk and bringing the global economy to its knees.
parallax
It's amazing that something as revolutionary as the single chip computer could come out of an engineer staring at thirteen separate schematics and saying, "Ok, but what about doing this with one chip?" And then being in the right circumstances to do it.
The single-chip CPU is arguably the most important development of late 20th century, and it's exponential improvement (Moore's Law) is what drives the information economy. So what happens when Moore's law runs out?
If current trends are projected forward, by 2020 a bit of memory will be a single electron transistor, traces will be one molecule wide, and the cost of the fabrication plant will be the GNP of the planet. The speed of light imposes practical limits on how large you can make a chip and how fast you can clock one. This is why we'll have GHz chips, but fundamental physical laws prevent THz chips.
More importantly, the physical limits that shut down THz electronic computers apply to _any_ classical computing architecture; optical computing and other exotic technology can't beat the speed of light, or single-particle storage problems.
You can't win by going to SMP, because at best you get a linear increase with each processor; exponential increases in power require exponential increases in processor number, which require exponential increases in space and power consumption.
The only basis in physics for continuing Moore's law past classical computing is quantum computing. In a quantum computer N quantum bits (qbits) equals 2^N classical bits. This allows you to build a computer which scales exponentially with the physical resources of the computer. Quantum computing isn't a solved problem, but if and when it is it will be a revolution as big as the first single-chip CPU.
parallax
What about other clean alternatives?
Hot fusion is a hard problem, and will probably stay that way for many years to come. It will change the world when it's economically feasible, but it will always be a big, expensive thing to build and maintain (keeping the sun in a jar is not cheap). Forget about the developing world sharing in the hot fusion boom (unless someone there gets a hold of a fusion bomb and threatens the energy rich nations of the world with a different kind of boom).
Cold fusion would be fantastic, because you could do that in your basement, filtering the heavy hydrogen out of your tap water. This would be an even bigger win, because the whole world could afford it. Of course, cold fusion seems even less likely than hot right now.
Short run we should focus on moving to cleaner burning fossil fuels like natural gas, and slightly longer run we should think about converting to a fuel-cell based hydrogen economy. There are big efficiency and environmental wins to be had, without trying to contain a solar furnace in a magnetic bottle.
The problem is that there is a significant chance that an extinction-class impact will happen within our lifetimes. The chance isn't large, but it has been calculated by astronomers to be approximately equal to the chance of the average person dying in a commercial airline crash.
The number of people actively engaged in looking for these killer rocks is fewer than the number of people who work in your local McDonald's. That's the _global_ total.
The reason for this is a lack of funding. It is pure coincidence that this particular rock was observed; the number of known-trajectory asteroids is small compared to the total number in the solar system, and most of the sky is unwatched most of the time.
That's why the chance of this asteroid hitting us on a near approach is about the same as being hit by one we haven't observed.
Lots of governments spend lots of money ensuring airline safety. I don't have the numbers at hand, but it is my understanding that a fairly comprehensive sky survey would cost significantly less.
Personally, I'd feel awfully silly being killed by a rock we could have nudged in a different direction with a few years of warning. If we discover one with only a few months of lead time, NASA says we're pretty much screwed.
I'm writing a thesis, so I'll try to make this short. :)
There are essentially two ways of expressing an idea: formal and informal language.
Formal language is mathematical notation, computer code, or any other extremely well defined, syntactically precise expression of an idea. An expression in formal language is intended to precisely express an idea or set of relationships in a way that is both rigorous and unambiguous. A C program, Maxwell's equations, and the axioms of set theory are all examples of formal language.
In some fields of human endeavor, I can get away without using formal language to convey an idea -- the study of French literature or poetry, for instance. In other areas, such as number theory or cryptography I have no choice but to use formal language to express formal ideas about which I must make very specific, technical claims.
I can talk about encryption in an informal way, but the only way I can say exactly what the blowfish cipher is doing is by providing an algorithmic description of its operation. Such a description is absolutely necessary, because any informal description could lead to multiple interpretations of its operation, which could in turn lead to differing analyses of its cryptographic strength.
Therefor, I cannot talk about cryptography (in the sense of a conversation between experts, or a conversation between teacher and student) without resorting to formal language if my ideas are to be understood clearly. Source code or an algorithmic description _by_itself_ may not be speech, but speech on the subject of mathematics, science, or computer science _is_not_possible_ without formal language.
Unless the government seeks to ban education in the field of cryptography, number theory, or the foundations of mathematics and computer science, I must be able to freely publish formal and informal descriptions of my work. Simply because someone could translate those descriptions into a functioning cryptographic system, for instance, doesn't change the necessity of this type of exchange if education and informed discussion are to take place.
Rich
A lot of the claims made in the article are misleading or overblown. The idea of using very short pulses for data transmission is not new, and as someone has already pointed out this is merely a special case of spread spectrum encoding.
First: An extremely short pulse approximates a delta function, which has infinite frequency content; "DC to daylight." This is still a form of RF transmission, it just happens that you are dumping energy into a very wide range of frequencies.
Second: Transmissions using this technique _do_ interfere with other RF transmissions. In fact, they interfere with _all_ other transmissions, but that interference is spread over the entire spectrum so it does not interfere with any one frequency very strongly (this raises FCC regulatory questions). In addition, a time-domain spread spectrum encoding makes the likelihood of interfering with another pulsed time-domain spread-spectrum transmission very small, if a good spreading algorithm is chosen.
Third: This is not a new idea (we were looking at this a few months ago for a data transmission application) and there is a reason why this hasn't been widely implemented: timing. In order to receive a pulsed time-domain spread-spectrum signal, you must synchronize your receiver's spread-spectrum decoder to the transmitter's encoder. The shorter the pulses, the more exact the timing and the more difficult this synchronization becomes.
Here is an analogy:
Imagine transmitting a signal by encoding it as a time-varying sequence of baseballs being fed to a pitching machine. The receiver catches the balls, decodes the sequence and reconstructs the signal.
If the transmitter is the only one pitching, the task of decoding is easy.
The problem is, the transmitter is not the only one feeding the pitching machine -- the noise in the environment is also feeding balls in. The best way to encode the signal to avoid any particular noise source (and to avoid interfering with anyone else) is to make the encoding look as random as possible, which is what spread-spectrum encoding is all about.
The resulting stream of baseballs looks random, since it is a combination of a spread-spectrum signal and random interference. In order to decode the signal, you want to catch only the balls that represent the signal.
In order to do this, you install a shutter in front of the receiver -- the spread spectrum decoder -- which will only let the "signal" balls through. This requires the decoder driving the shutter to be exactly synchronized with the encoder.
As the pulses become narrower, the "balls" are coming faster and timing the shutter must become more exact to exclude non-signal balls. If a non-signal ball passes through the shutter (or a signal ball is missed), the error will break the syncronization between the tranmitter and receiver. Narrow pulses also make it more difficult to lock the receiver's decoder to the transmitter's encoder in the first place. Once the pulses become short enough, maintinaing synchronization becomes almost impossible without an additional, non-spread communication channel. If an additional, non-spread chanel is used, then you are back to the problems of ordinary RF transmission.
There is great potential in this technology, but the technical chalenges (and regulatory hurdles) are large.
Rich
is it a surprise that open source and commecrial ventures are compatible? since red hat, etc. it seems pretty obvious.
They got it mostly right -- interesting that the somewhat tricky ideas of open source (value != $$, competitive advantage to using non-proprietary code and giving stuff away) are finally permeating the main-stream media. Somewhat ironic to have a mag titled "the economist" touting the value of free software...