Re:Lisp is cool...
on
Land of Lisp
·
· Score: 4, Interesting
Nothing will beat the Symbolics Lisp machine.
Ever use one? I've used the refrigerator-sized Symbolics 3600. 45 minute garbage collections. Flaky electronics. An arrogant service organization. Those things lost out to general purpose UNIX workstations for good reasons. Even for running LISP, a SUN 2 was better than a Symbolics 3600.
(Symbolics was also tied in strongly to the 1980s expert systems crowd, the "strong AI Real Soon Now" people", like Ed Feigenbaum. I went through Stanford CS when those guys were running the department, just as it was becoming clear that expert systems really couldn't do all that much. Not a happy time in academic computer science. Stanford had to move computer sciences from Arts and Sciences to Engineering, and put in adult supervision.)
Many TV shows are now distributed through Flash player programs which load the video and play ads. The CW and CBS do this. These players are designed to prevent the viewer from viewing the show without ads. Will those work through this conversion mechanism? If not, there's not going to be consumer acceptance of this.
(Those players do not play well with high security settings. Some won't run if they can't store cookies. Others will show the same ad over and over. With Flashblock, about half the CBS ads don't appear, but there's a countdown timer you have to wait out before the show resumes.)
Microcomputers were developed based on needs by NASA to have computers that were light enough to be on a spacecraft because you couldn't fit a room-sized mainframe on an Apollo spacecraft or on the Lunar Excursion Module. So, let's see. We have this little space race thing that ends in the 1970s with NASA pouring money into little teeny solid state computing devices and you get the Apple ][ computer in 1977. And the IBM PC four years later. The last Apollo spacecraft was designed around 1967 more or less so I have to ask the naysayers what they're expecting to see in about ten years now that the ISS is complete. because everybody knows NASA science doesn't contribute to anything down here on earth.
No, they weren't. Microcomputers were developed because of a desire to get the parts count down for electronic desk calculators. That's why the Intel 4004 was built. The Intel 8008 was built to get the parts count down for the Datapoint 2200 terminal, although it was late and Datapoint had to use another approach.
The Apollo Guidance Computer wasn't even close to being a single-chip computer. It was a nice piece of electronics packaging, but it had 5600 gates in 2800 packages. Integration was at the level of a single-chip dual NOR gate. It did the job, but was a technological dead end.
Database engines have to be highly reliable, and MySQL has some good ones. They're very difficult to write and get correct. There's considerable theory and formalism involved. Real Q/A is needed. This is not a place for "crowdsourcing".
This has come up in the context of domain names, where a long, painful set of rules has been devised to try to prevent having two domain names which look similar but are different to DNS. If exact equality of text matters, it's helpful to have a limited character set for identifiers.
There's currently a debate underway on Wikipedia over whether user names with unusual characters should be allowed. This isn't a language question; the issue is willful obfuscation by users who choose names with hard-to-type characters.
As for having more operators, it's probably not worth it. It's been tried; both MIT and Stanford had, at one time, custom character sets, with most of the standard mathematical operators on the keys. This never caught on. In fact, operator overloading is usually a lose. Python ran into this. "+" was overloaded for concatenation. Then somebody decided that "*" should be overloaded, so that "a" + "a" was equivalent to 2*"a". The result is thus "aa". This leads to results like 2*"10" being "1010". The big mistake was defining a mixed-mode overload.
In C++, mixed-mode overloads are fully supported by the template system and a nightmare when reading code.
In Mathematica, the standard representation for math uses long names for functions, completely avoiding the macho terseness the math community has historically embraced.
The math people really need to survive in a very dynamic society involves probability, statistics, and estimation. Schools rarely teach how to estimate something within 10-20%, yet that's an enormously valuable skill. Being able to decide what to throw out of an estimation without losing too much accuracy is essential.
Kids should know enough probability to estimate the odds on the local lottery. They should know what an "expectation" is, and what zero-sum and negative-sum games are and how to recognize them. They should be able to calculate the odds of dying in a terrorist attack and in an auto accident. They should know the risk/reward calculation for various career choices. They need to understand the concept of exposure to interest rate variations in loans and investments.
Plane geometry, Euclid proof style, could probably be dropped with no loss. (I've done animation physics engines and GPS calculations, and I didn't use that stuff. Analytical geometry, yes; straightedge and compass proofs, no.)
It is not the direct application of mathematics on everyday life that is most beneficial, but the analytical and conceptual skill set gained by learning higher level math.
Nah. That claim was once made for teaching Latin in public schools. It's still made for teaching Euclidean plane geometry.
Keep in mind that in order to connect, you need to get through a firewall, which means you need an explicit allow entry, and if youre running NAT (which tbqh I dont see why the workstations WOULDNT be natted) you also need an explicit forward rule (all this precludes uPnP, but really who would have THAT enabled;) ).
Not necessarily. The attack could be staged by using a cross-site scripting vulnerability to allow Javascript on a web page to make a call via XMLHttpRequest to the local machine. You can't quite exploit this by talking to "localhost", because the input has to come in over the LAN to reach the management controller hardware in the network interface, but you might be able to get the local router to reflect packets back to the local host.
Against a server farm, if you can get something running on any server on the farm, you can usually talk to other servers in the same farm, and try to take them over via this attack.
I get the feeling they had a big budget cut in the early 2000s. They were way ahead in 2000, but Asimo hasn't made much progress in the last decade. Compare BigDog, which has robust locomotion and real running. Asimo has been seen to fall in public demos in a benign environment. Nor can it get up if it falls, something most of the better hobbyist humanoid robots, and even toys, now do. Asimo's running has such a short flight phase (both feet off the ground) that they have to take high-speed photos from ground level to show that there's a moment when both feet are just barely off the ground.
What dould possibly go wrong?
on
USB 'Dead Drops'
·
· Score: 3, Interesting
Could be worse. In 1969, the Museum of Modern Art in New York deployed Pulsa, an exhibit which included many strobe lights arranged to flash in sequence. There was a long line of strobes not only on the museum, but extending to adjacent buildings.
Pilots reported runway lighting in midtown Manhattan. The "moving ball of light" strobe system for runways was chosen because, even in cluttered urban areas with many parallel lines of light, there's nothing which looks like that. The FAA made them retime the strobes so that it didn't look like a runway.
The Internet Archive should pick this up, if they haven't already. I'll talk to some people.
Archiving is getting easier. I had a minor part in preserving the archives of the Stanford AI lab. That required weeks of loading 6250bpi 2400 foot open reel tapes.
The intel business has changed. It used to be that the US intelligence community was focused on the capabilities of the USSR, which was a big, slow-moving, closed society. Moving to today's targets is tough. The CIA and NSA had all that expertise focused on what the USSR was doing. They were looking for big stuff like missile launchers that are visible from orbit, and communications between a very centralized bureaucracy in Moscow with outlying subordinate stations. It was reasonably clear how to approach that. All that capability was ill-matched to the many post-USSR threats.
Trying to get intel on a terrorist group is tough. First, the target is tiny. Remember, 9/11 only involved about 25 people, and only a few of them knew the plan more than a day in advance. Second, the groups aren't that connected. Islamic terrorism is an ideology, not an organization. Al-Queda ("The Base") is maybe 200 people at this point, and not doing much. The terrorist incidents in recent years haven't been very connected. Third, intel on terrorist groups has a short useful life.
Where bin Laden was last month is only of historical interest. US intelligence used to be strategic. Now it's mostly tactical. The US used to obsess over Soviet bomber production rates. There's nothing like that to track now.
Then there are the messes in Afghanistan and Iraq. That's an intel problem; insurgents are hard to find but easy to kill. The dumb insurgents are already dead. The remaining ones know how to keep quiet. There's no centralized control of either insurgency. If the insurgents establish a "stronghold", they become vulnerable. That, by the way, is why the war with the Taliban is stalemated. If the Taliban concentrates enough combat power to do anything big, they become vulnerable to modern firepower. If they operate in the background, they can survive, but can't take over, unless they can wear out their opposition. (This frustrates the US military. "Marine doctrine demands a decision." - FMFM-1. Insurgent doctrine does not.
"The enemy advances, we retreat; the enemy camps, we harass; the enemy tires, we attack; the enemy retreats, we pursue" - Mao Zedong.)
Coming up next: Mexico. Arguably, northern Mexico is already a "failed state". Drug lords are more vulnerable to intel operations than religiously-motivated insurgents, though. They can't hide too much and still do business, they have to deal and communicate, and the members mistrust each other.
That confusion is why the US now has such a confused intel establishment. That's no excuse for it being as big as it is, though. Or, really, as secretive. Most of the targets today have insignificant capabilities to infiltrate or eavesdrop on the US intel establishment. It's not like going up against Moscow Center, which would devote huge resources and years of time to getting inside some US establishment. The secrecy can get in the way of getting things done.
During WWII, and for decades thereafter, it didn't take a pass to get into the Pentagon. Gen. Marshall decided that any competent intelligence service would figure out a way to get into the building, and so only the really important stuff would be secured. Trying to secure the whole building would be security theater. We need more of that kind of thinking.
What's worried me for some time are the various "remote maintenance" schemes built into network controllers. See, for example, Intel's "Active Management Technology". This is Intel's successor to the Intelligent Platform Management Interface. These have a protocol stack built into the network board, with connections to other parts of the system strong enough to power the machine on and off, patch the disk, and do other drastic system changes. AMT is easier to attack from a distance than IPMI; it uses SOAP, HTTP, and TCP (on ports 16992 through 16995, which had better be blocked at your firewall), while IPMI used its own specialized protocol over UDP.
All that prevents taking over a machine with this mechanism is that the network controller is supposed to ship with no keys loaded. A "backdoor" would simply consist of pre-loading some crypto keys at the factory, or somewhere else in the supply chain. Considering the amount of hostile junk that routinely shows up on new USB sticks, that probably wouldn't be hard to accomplish.
A true "hardware level" attack for IPMI or AMT would be to ship a network controller which had keys pre-installed and enabled, but reported that remote management was disabled. There would be no way to find such a "backdoor", short of grinding open the network controller chip and reverse engineering it with a scanning electron microscope. There are special purpose systems for doing exactly that, used for reverse engineering IC designs, but this is e difficult and expensive process.
There's been considerable improvement.
Google still has some holes in dealing with "malware", phishing, etc. But these are mostly obscure tricks used to get around Google's malware reporting. You can report the sites below over and over, but nothing happens, because Google's reporting system doesn't understand that these Google features are exploitable.
I'm pleased to notice that, at last, Google is no longer running ads for software for spamming Craigslist. Search for "craigslist auto poster tool". There used to be ads for programs for spamming Craigslist, and some of them even accepted payment through Google Checkout. (That last could lead to legal problems, since Google was not only advertising an legally questionable product, but taking a cut of the revenue.) That seems to have stopped. There are still ads for offshored services which manually spam Craigslist.
Some numbers from their YouTube video slide show. (That's a bad sign. I'd much rather see a technical paper.) Here's what they claim:
"300 Wh/Kg"
"2500 charge cycles with no degradation".
"6 minute charge time for 100KWh."
Those are impressive numbers, if real. 2500 charge cycles at 500 Km per charge is 750,000 Km. Typical car life today is around 250,000 Km, so the battery will outlast the car.
300 Wh/Kg is very high, but not unheard of. They say it's a lithium chemistry. Lithium tetrachloroaluminate batteries get numbers like that. Unfortunately, it has hazard problems. "Reacts violently with water or humid air to give off corrosive fumes of hydrocloric acid and sulfur oxide." (A basic problem with battery chemistry is that the further you go out on the electromotive series, the higher the cell voltage, but the more reactive the material. Sodium-sulfur batteries have very good energy density, but sodium burns on contact with water.)
It's possible that this battery does everything they say. But they never mention safety or flammability.
Seriously, a whole lot of embedded devices - sensors, microcontrollers, machinery, vehicles, booths - use RS232 (as simple, universal and VASTLY easier to program than USB)
Yes, when you do embedded work, you often find yourself going back 20 years in technology. There's progress, though.
The trend in the embedded world is to put sensors and controllers on 10baseT. The traditional alternatives were either huge numbers of serial ports, or nonstandard proprietary networks. Both suck. 10baseT is quite robust electrically; it's noise-immune, balanced, and AC-coupled. This matters when you have heavy machinery around.
USB is making some headway in the embedded world, but there's a problem - the standard USB connector has no retention mechanism. Ethernet cables latch in place, but USB connectors do not. There are now "high retention" USB connectors (they're orange) for industrial use, and at least three incompatible latching mechanisms. This is not happy-making for embedded system designers, who would like to use USB more, but can't tolerate plugs falling out.
OK, for all you whiners about the evils of software patents, this is what you get - secret algorithms.
US law used to disfavor trade secrets, encouraging patents. Patents have a limited life, and disclose the technology. Trade secrets have a potentially unlimited life, and no disclosure requirement.
Note that in dealing with Microsoft's technology, patents were a minor headache for interoperability, but secret APIs have been a huge obstacle to interoperability.
Yes, US cable TV sales are flat. In fact, they peaked in 2001. That's before video over the Internet was popular, but when digital over-the-air HDTV was ramping up.
We'll know that cable TV is in trouble when the sports networks start expressing concern over declining viewing. So far, that hasn't happened. What has happened is fragmentation. Network and per-channel market share is declining.
Meanwhile, Univision just passed The CW in market share, and is about to move into the #3 position. Univision has little competition in the US Spanish-language sector, so it's become the main network for a big chunk of the population. The original author has no clue. SF has a minor market share. The action is in telenovelas.
"Microsoft Sues Motorola Over Android Patent Infringement"
"Microsoft says Motorola is violating nine patents "that are essential to the smartphone user experience, including synchronizing email, calendars and contacts, scheduling meetings, and notifying applications of changes in signal strength and battery power."
So this has nothing to do with Linux. Those are features of Android. And, from some other patent agreements, the "synchronizing mail" thing applies only to synchronizing with Microsoft Outlook and Exchange.
All ASUS has to do is remove Microsoft Outlook and Exchange compatibility from their version of Android. Encourage users to use Google's "cloud" apps instead. Or ordinary IMAP. Microsoft will love that.
Arguably, we should have some emergency controls for the Internet. I'd suggest that the following emergency systems be implemented:
Mail servers forward only text email, stripping all MIME content. Useful in case of serious virus trouble.
Cell phone switches handle voice and SMS messages only. Maybe raw pictures on some platforms. No downloads, no "apps", no tethering, no IP.
Under severe overload conditions during a cyber-attack, the FCC should be able to order an advertising shutdown. All advertising servers must go offline until the emergency is over.
All this should be publicly tested occasionally, like the Emergency Broadcast System.
This would be enough to deal with serious overloads, outages, or viruses, but doesn't have censorship implications.
This is very clever. Vacuum pickers have been around for most of a century; they first appeared for paper handling in printing presses But they're usually flat, or at best, they have a foam or sponge front, so they can deal with some irregularities in the object being lifted. This is the first one I've seen that can grip around something. The clever part is that the flexible vacuum bag is filled with small objects that keep the bag size almost constant even when vacuum is applied. In operation, I presume it is used by pushing the gripper into wrapping around the object.
The usual vacuum picker problems apply. If only part of the bag (which has a pattern of small holes) is in contact with the object, the rest of the bag leaks. So the vacuum system has to extract a lot of wasted air to keep the pressure inside the system low. This limits the strength of the grip. It's also going to be noisy, probably about as noisy as the business end of a vacuum cleaner.
This definitely has applications in industrial automation where soft objects are being handled. It may be useful for fruit picking and clothing assembly, which are still too labor-intensive.
Seeing your comment, Animats, that the the person who makes the decision to abort, and the person who actually makes the abort are in different places, it makes sense to me now how that failure occurred.
That's part of it. One guy was frantically trying to release the payload and launch, trying to yank on a strap. Coordination was by walkie-talkies, so, since he had his hands full, he probably wasn't on the air. So the guy with the abort button didn't act until it was too late. The range safety mistake was that they had a safe zone within which the launch truck was supposed to stay, but the perimeter wasn't clearly marked. They left that zone without aborting the launch. (Next time, buy a few dozen traffic cones.) From the video, the audience was tiny; there were four or five cars in the viewing area. They were in Alice Springs, after all, at the edge of the Australian outback. It's not like a launch at Canaveral, with a hundred thousand spectators. They had a chance of still launching the balloon. The odds of hitting anything were low. So I can see why they persisted.
The report is mostly bureaucratic blithering.
NTSB aviation crash reports are much more coherent. One would expect a transcript of who said what when (you'd think they'd record the walkie talkie traffic.) Also, the NTSB names names; NASA doesn't.
All they really need to fix is 1) bigger bolts on the crane's attachment plate, 2) establish policy that if the launch truck leaves the safety zone, immediately abort and cut the balloon loose, 3) come up with a release mechanism with a design that wasn't borrowed from link-and-pin railroad couplers circa 1870, and 4) get the spectators upwind from the balloon. Identifying the people who designed the attachment plate and the release mechanism, and firing them, wouldn't be a bad idea, either. Those guys flunked basic mechanical engineering.
That's a worst-case analysis. A neat one, though. If you don't insist that the algorithm guarantee> a solution within some error bound, you can get much faster results. There are, however, pathological cases that will take longer. Non-deterministic algorithms usually come in at O(N^2). See "A Survey on Travelling Salesman Problem".
Watched the video. Fox didn't get any pictures of actual scans taken while they were there. The pictures shown are from AS&E's usual set of demo pictures.
They should have had scans of the reporters. But the mobile system isn't certified for personnel scans.
It's not a significant radiation hazard beyond the recommended 9 foot approach limit. But there are reasonable questions about someone close to the scanning vehicle. The Hickam AFB study did not measure the exposure directly in line with the scanning fan closer than 7' from the vehicle. At that distance, the exposure is 35mRem/hour. Driving by a parked car equipped with standard dosimeters returned a zero dose reading.
Slashdot Mirror
Nothing will beat the Symbolics Lisp machine.
Ever use one? I've used the refrigerator-sized Symbolics 3600. 45 minute garbage collections. Flaky electronics. An arrogant service organization. Those things lost out to general purpose UNIX workstations for good reasons. Even for running LISP, a SUN 2 was better than a Symbolics 3600.
(Symbolics was also tied in strongly to the 1980s expert systems crowd, the "strong AI Real Soon Now" people", like Ed Feigenbaum. I went through Stanford CS when those guys were running the department, just as it was becoming clear that expert systems really couldn't do all that much. Not a happy time in academic computer science. Stanford had to move computer sciences from Arts and Sciences to Engineering, and put in adult supervision.)
Many TV shows are now distributed through Flash player programs which load the video and play ads. The CW and CBS do this. These players are designed to prevent the viewer from viewing the show without ads. Will those work through this conversion mechanism? If not, there's not going to be consumer acceptance of this.
(Those players do not play well with high security settings. Some won't run if they can't store cookies. Others will show the same ad over and over. With Flashblock, about half the CBS ads don't appear, but there's a countdown timer you have to wait out before the show resumes.)
Microcomputers were developed based on needs by NASA to have computers that were light enough to be on a spacecraft because you couldn't fit a room-sized mainframe on an Apollo spacecraft or on the Lunar Excursion Module. So, let's see. We have this little space race thing that ends in the 1970s with NASA pouring money into little teeny solid state computing devices and you get the Apple ][ computer in 1977. And the IBM PC four years later. The last Apollo spacecraft was designed around 1967 more or less so I have to ask the naysayers what they're expecting to see in about ten years now that the ISS is complete. because everybody knows NASA science doesn't contribute to anything down here on earth.
No, they weren't. Microcomputers were developed because of a desire to get the parts count down for electronic desk calculators. That's why the Intel 4004 was built. The Intel 8008 was built to get the parts count down for the Datapoint 2200 terminal, although it was late and Datapoint had to use another approach.
The Apollo Guidance Computer wasn't even close to being a single-chip computer. It was a nice piece of electronics packaging, but it had 5600 gates in 2800 packages. Integration was at the level of a single-chip dual NOR gate. It did the job, but was a technological dead end.
Database engines have to be highly reliable, and MySQL has some good ones. They're very difficult to write and get correct. There's considerable theory and formalism involved. Real Q/A is needed. This is not a place for "crowdsourcing".
This has come up in the context of domain names, where a long, painful set of rules has been devised to try to prevent having two domain names which look similar but are different to DNS. If exact equality of text matters, it's helpful to have a limited character set for identifiers.
There's currently a debate underway on Wikipedia over whether user names with unusual characters should be allowed. This isn't a language question; the issue is willful obfuscation by users who choose names with hard-to-type characters.
As for having more operators, it's probably not worth it. It's been tried; both MIT and Stanford had, at one time, custom character sets, with most of the standard mathematical operators on the keys. This never caught on. In fact, operator overloading is usually a lose. Python ran into this. "+" was overloaded for concatenation. Then somebody decided that "*" should be overloaded, so that "a" + "a" was equivalent to 2*"a". The result is thus "aa". This leads to results like 2*"10" being "1010". The big mistake was defining a mixed-mode overload.
In C++, mixed-mode overloads are fully supported by the template system and a nightmare when reading code.
In Mathematica, the standard representation for math uses long names for functions, completely avoiding the macho terseness the math community has historically embraced.
The math people really need to survive in a very dynamic society involves probability, statistics, and estimation. Schools rarely teach how to estimate something within 10-20%, yet that's an enormously valuable skill. Being able to decide what to throw out of an estimation without losing too much accuracy is essential.
Kids should know enough probability to estimate the odds on the local lottery. They should know what an "expectation" is, and what zero-sum and negative-sum games are and how to recognize them. They should be able to calculate the odds of dying in a terrorist attack and in an auto accident. They should know the risk/reward calculation for various career choices. They need to understand the concept of exposure to interest rate variations in loans and investments.
Plane geometry, Euclid proof style, could probably be dropped with no loss. (I've done animation physics engines and GPS calculations, and I didn't use that stuff. Analytical geometry, yes; straightedge and compass proofs, no.)
It is not the direct application of mathematics on everyday life that is most beneficial, but the analytical and conceptual skill set gained by learning higher level math.
Nah. That claim was once made for teaching Latin in public schools. It's still made for teaching Euclidean plane geometry.
Keep in mind that in order to connect, you need to get through a firewall, which means you need an explicit allow entry, and if youre running NAT (which tbqh I dont see why the workstations WOULDNT be natted) you also need an explicit forward rule (all this precludes uPnP, but really who would have THAT enabled ;) ).
Not necessarily. The attack could be staged by using a cross-site scripting vulnerability to allow Javascript on a web page to make a call via XMLHttpRequest to the local machine. You can't quite exploit this by talking to "localhost", because the input has to come in over the LAN to reach the management controller hardware in the network interface, but you might be able to get the local router to reflect packets back to the local host.
Against a server farm, if you can get something running on any server on the farm, you can usually talk to other servers in the same farm, and try to take them over via this attack.
I get the feeling they had a big budget cut in the early 2000s. They were way ahead in 2000, but Asimo hasn't made much progress in the last decade. Compare BigDog, which has robust locomotion and real running. Asimo has been seen to fall in public demos in a benign environment. Nor can it get up if it falls, something most of the better hobbyist humanoid robots, and even toys, now do. Asimo's running has such a short flight phase (both feet off the ground) that they have to take high-speed photos from ground level to show that there's a moment when both feet are just barely off the ground.
Could be worse. In 1969, the Museum of Modern Art in New York deployed Pulsa, an exhibit which included many strobe lights arranged to flash in sequence. There was a long line of strobes not only on the museum, but extending to adjacent buildings.
Pilots reported runway lighting in midtown Manhattan. The "moving ball of light" strobe system for runways was chosen because, even in cluttered urban areas with many parallel lines of light, there's nothing which looks like that. The FAA made them retime the strobes so that it didn't look like a runway.
Good.
The Internet Archive should pick this up, if they haven't already. I'll talk to some people.
Archiving is getting easier. I had a minor part in preserving the archives of the Stanford AI lab. That required weeks of loading 6250bpi 2400 foot open reel tapes.
The intel business has changed. It used to be that the US intelligence community was focused on the capabilities of the USSR, which was a big, slow-moving, closed society. Moving to today's targets is tough. The CIA and NSA had all that expertise focused on what the USSR was doing. They were looking for big stuff like missile launchers that are visible from orbit, and communications between a very centralized bureaucracy in Moscow with outlying subordinate stations. It was reasonably clear how to approach that. All that capability was ill-matched to the many post-USSR threats.
Trying to get intel on a terrorist group is tough. First, the target is tiny. Remember, 9/11 only involved about 25 people, and only a few of them knew the plan more than a day in advance. Second, the groups aren't that connected. Islamic terrorism is an ideology, not an organization. Al-Queda ("The Base") is maybe 200 people at this point, and not doing much. The terrorist incidents in recent years haven't been very connected. Third, intel on terrorist groups has a short useful life. Where bin Laden was last month is only of historical interest. US intelligence used to be strategic. Now it's mostly tactical. The US used to obsess over Soviet bomber production rates. There's nothing like that to track now.
Then there are the messes in Afghanistan and Iraq. That's an intel problem; insurgents are hard to find but easy to kill. The dumb insurgents are already dead. The remaining ones know how to keep quiet. There's no centralized control of either insurgency. If the insurgents establish a "stronghold", they become vulnerable. That, by the way, is why the war with the Taliban is stalemated. If the Taliban concentrates enough combat power to do anything big, they become vulnerable to modern firepower. If they operate in the background, they can survive, but can't take over, unless they can wear out their opposition. (This frustrates the US military. "Marine doctrine demands a decision." - FMFM-1. Insurgent doctrine does not. "The enemy advances, we retreat; the enemy camps, we harass; the enemy tires, we attack; the enemy retreats, we pursue" - Mao Zedong.)
Coming up next: Mexico. Arguably, northern Mexico is already a "failed state". Drug lords are more vulnerable to intel operations than religiously-motivated insurgents, though. They can't hide too much and still do business, they have to deal and communicate, and the members mistrust each other.
That confusion is why the US now has such a confused intel establishment. That's no excuse for it being as big as it is, though. Or, really, as secretive. Most of the targets today have insignificant capabilities to infiltrate or eavesdrop on the US intel establishment. It's not like going up against Moscow Center, which would devote huge resources and years of time to getting inside some US establishment. The secrecy can get in the way of getting things done.
During WWII, and for decades thereafter, it didn't take a pass to get into the Pentagon. Gen. Marshall decided that any competent intelligence service would figure out a way to get into the building, and so only the really important stuff would be secured. Trying to secure the whole building would be security theater. We need more of that kind of thinking.
What's worried me for some time are the various "remote maintenance" schemes built into network controllers. See, for example, Intel's "Active Management Technology". This is Intel's successor to the Intelligent Platform Management Interface. These have a protocol stack built into the network board, with connections to other parts of the system strong enough to power the machine on and off, patch the disk, and do other drastic system changes. AMT is easier to attack from a distance than IPMI; it uses SOAP, HTTP, and TCP (on ports 16992 through 16995, which had better be blocked at your firewall), while IPMI used its own specialized protocol over UDP.
All that prevents taking over a machine with this mechanism is that the network controller is supposed to ship with no keys loaded. A "backdoor" would simply consist of pre-loading some crypto keys at the factory, or somewhere else in the supply chain. Considering the amount of hostile junk that routinely shows up on new USB sticks, that probably wouldn't be hard to accomplish.
A true "hardware level" attack for IPMI or AMT would be to ship a network controller which had keys pre-installed and enabled, but reported that remote management was disabled. There would be no way to find such a "backdoor", short of grinding open the network controller chip and reverse engineering it with a scanning electron microscope. There are special purpose systems for doing exactly that, used for reverse engineering IC designs, but this is e difficult and expensive process.
There's been considerable improvement. Google still has some holes in dealing with "malware", phishing, etc. But these are mostly obscure tricks used to get around Google's malware reporting. You can report the sites below over and over, but nothing happens, because Google's reporting system doesn't understand that these Google features are exploitable.
I'm pleased to notice that, at last, Google is no longer running ads for software for spamming Craigslist. Search for "craigslist auto poster tool". There used to be ads for programs for spamming Craigslist, and some of them even accepted payment through Google Checkout. (That last could lead to legal problems, since Google was not only advertising an legally questionable product, but taking a cut of the revenue.) That seems to have stopped. There are still ads for offshored services which manually spam Craigslist.
Some numbers from their YouTube video slide show. (That's a bad sign. I'd much rather see a technical paper.) Here's what they claim:
Those are impressive numbers, if real. 2500 charge cycles at 500 Km per charge is 750,000 Km. Typical car life today is around 250,000 Km, so the battery will outlast the car.
300 Wh/Kg is very high, but not unheard of. They say it's a lithium chemistry. Lithium tetrachloroaluminate batteries get numbers like that. Unfortunately, it has hazard problems. "Reacts violently with water or humid air to give off corrosive fumes of hydrocloric acid and sulfur oxide." (A basic problem with battery chemistry is that the further you go out on the electromotive series, the higher the cell voltage, but the more reactive the material. Sodium-sulfur batteries have very good energy density, but sodium burns on contact with water.)
It's possible that this battery does everything they say. But they never mention safety or flammability.
Seriously, a whole lot of embedded devices - sensors, microcontrollers, machinery, vehicles, booths - use RS232 (as simple, universal and VASTLY easier to program than USB)
Yes, when you do embedded work, you often find yourself going back 20 years in technology. There's progress, though. The trend in the embedded world is to put sensors and controllers on 10baseT. The traditional alternatives were either huge numbers of serial ports, or nonstandard proprietary networks. Both suck. 10baseT is quite robust electrically; it's noise-immune, balanced, and AC-coupled. This matters when you have heavy machinery around.
USB is making some headway in the embedded world, but there's a problem - the standard USB connector has no retention mechanism. Ethernet cables latch in place, but USB connectors do not. There are now "high retention" USB connectors (they're orange) for industrial use, and at least three incompatible latching mechanisms. This is not happy-making for embedded system designers, who would like to use USB more, but can't tolerate plugs falling out.
OK, for all you whiners about the evils of software patents, this is what you get - secret algorithms.
US law used to disfavor trade secrets, encouraging patents. Patents have a limited life, and disclose the technology. Trade secrets have a potentially unlimited life, and no disclosure requirement.
Note that in dealing with Microsoft's technology, patents were a minor headache for interoperability, but secret APIs have been a huge obstacle to interoperability.
Yeah, it's got a serial port, with TTL levels, at its external connector. Big deal.
It's also possible to attach USB devices, which is somewhat more useful today. For example, you can plug a real keyboard into an iPad.
Yes, US cable TV sales are flat. In fact, they peaked in 2001. That's before video over the Internet was popular, but when digital over-the-air HDTV was ramping up.
We'll know that cable TV is in trouble when the sports networks start expressing concern over declining viewing. So far, that hasn't happened. What has happened is fragmentation. Network and per-channel market share is declining.
Meanwhile, Univision just passed The CW in market share, and is about to move into the #3 position. Univision has little competition in the US Spanish-language sector, so it's become the main network for a big chunk of the population. The original author has no clue. SF has a minor market share. The action is in telenovelas.
"Microsoft Sues Motorola Over Android Patent Infringement" "Microsoft says Motorola is violating nine patents "that are essential to the smartphone user experience, including synchronizing email, calendars and contacts, scheduling meetings, and notifying applications of changes in signal strength and battery power."
So this has nothing to do with Linux. Those are features of Android. And, from some other patent agreements, the "synchronizing mail" thing applies only to synchronizing with Microsoft Outlook and Exchange.
All ASUS has to do is remove Microsoft Outlook and Exchange compatibility from their version of Android. Encourage users to use Google's "cloud" apps instead. Or ordinary IMAP. Microsoft will love that.
Arguably, we should have some emergency controls for the Internet. I'd suggest that the following emergency systems be implemented:
This would be enough to deal with serious overloads, outages, or viruses, but doesn't have censorship implications.
This is very clever. Vacuum pickers have been around for most of a century; they first appeared for paper handling in printing presses But they're usually flat, or at best, they have a foam or sponge front, so they can deal with some irregularities in the object being lifted. This is the first one I've seen that can grip around something. The clever part is that the flexible vacuum bag is filled with small objects that keep the bag size almost constant even when vacuum is applied. In operation, I presume it is used by pushing the gripper into wrapping around the object.
The usual vacuum picker problems apply. If only part of the bag (which has a pattern of small holes) is in contact with the object, the rest of the bag leaks. So the vacuum system has to extract a lot of wasted air to keep the pressure inside the system low. This limits the strength of the grip. It's also going to be noisy, probably about as noisy as the business end of a vacuum cleaner.
This definitely has applications in industrial automation where soft objects are being handled. It may be useful for fruit picking and clothing assembly, which are still too labor-intensive.
Seeing your comment, Animats, that the the person who makes the decision to abort, and the person who actually makes the abort are in different places, it makes sense to me now how that failure occurred.
That's part of it. One guy was frantically trying to release the payload and launch, trying to yank on a strap. Coordination was by walkie-talkies, so, since he had his hands full, he probably wasn't on the air. So the guy with the abort button didn't act until it was too late. The range safety mistake was that they had a safe zone within which the launch truck was supposed to stay, but the perimeter wasn't clearly marked. They left that zone without aborting the launch. (Next time, buy a few dozen traffic cones.) From the video, the audience was tiny; there were four or five cars in the viewing area. They were in Alice Springs, after all, at the edge of the Australian outback. It's not like a launch at Canaveral, with a hundred thousand spectators. They had a chance of still launching the balloon. The odds of hitting anything were low. So I can see why they persisted.
The report is mostly bureaucratic blithering. NTSB aviation crash reports are much more coherent. One would expect a transcript of who said what when (you'd think they'd record the walkie talkie traffic.) Also, the NTSB names names; NASA doesn't.
All they really need to fix is 1) bigger bolts on the crane's attachment plate, 2) establish policy that if the launch truck leaves the safety zone, immediately abort and cut the balloon loose, 3) come up with a release mechanism with a design that wasn't borrowed from link-and-pin railroad couplers circa 1870, and 4) get the spectators upwind from the balloon. Identifying the people who designed the attachment plate and the release mechanism, and firing them, wouldn't be a bad idea, either. Those guys flunked basic mechanical engineering.
That's a worst-case analysis. A neat one, though. If you don't insist that the algorithm guarantee> a solution within some error bound, you can get much faster results. There are, however, pathological cases that will take longer. Non-deterministic algorithms usually come in at O(N^2). See "A Survey on Travelling Salesman Problem".
Watched the video. Fox didn't get any pictures of actual scans taken while they were there. The pictures shown are from AS&E's usual set of demo pictures.
They should have had scans of the reporters. But the mobile system isn't certified for personnel scans.
It's not a significant radiation hazard beyond the recommended 9 foot approach limit. But there are reasonable questions about someone close to the scanning vehicle. The Hickam AFB study did not measure the exposure directly in line with the scanning fan closer than 7' from the vehicle. At that distance, the exposure is 35mRem/hour. Driving by a parked car equipped with standard dosimeters returned a zero dose reading.