This is the next step up from the "gimme, gimme" level of Kickstarter. You can still do Kickstarter projects, but now there's a tier above that, for when you need $1M or so.
OK, we have yet another mesh of processors, an idea that comes back again and again. The details of how processors communicate really matter. Is this is a totally non-shared-memory machine? Is there some shared memory, but it's slow? If there's shared memory, what are the cache consistency rules?
Historically, meshes of processors without shared memory have been painful to program. There's a long line of machines, from the nCube to the Cell, where the hardware worked but the thing was too much of a pain to program. Most designs have suffered from having too little local memory per CPU. If there's enough memory per CPU to, well, run at least a minimal OS and some jobs, then the mesh can be treated as a cluster of intercommunicating peers. That's something for which useful software exists. If all the CPUs have to be treated as slaves of a control machine, then you need all-new software architectures to handle them. This usually results in one-off software that never becomes mature.
Basic truth: we only have three successful multiprocessor architectures that are general purpose - shared-memory multiprocessors, clusters, and GPUs. Everything other than that has been almost useless except for very specialized problems fitted to the hardware. Yet this problem needs to be cracked - single CPUs are not getting much faster.
Right. But a big problem is preventing some hostile app on the user's phone from obtaining the user's secret key used in the challenge/response algorithm. With the code reverse engineered, atttackers now know where to go looking for that key and what to do with it when they have it.
Smartphones are not a secure platform. The carrier and Google (for Android) or Apple (for their phones) have total backdoor access. So does anyone who has their signing keys.
It's not that the US lacks rare earth metal resources. It's that, until recently, China was a cheaper supplier. Then the goverment of China tried to keep the price up and insisted that Chinese companies sell motors and other completed products, not raw materials. Some rare earth metal prices shot up by a factor of 20. So the Mountain Pass mine, closed in 2002, was cranked up again, this time with new equjpiment better pollution controls.
Pollution controls for a rare earth mine are a big deal. "Rare earths" are present in low concentrations, which means that a mine generates a small amount of product and huge amounts of toxic sludge.
The big rare earths mine in China has the world's largest sludge pond, and it leaks. This created an environmental disaster area for tens of kilometers around. Villages have had to be evacuated because of sludge pond leaks.
The Mountain Pass, California mine is less than a mile from I-15 between Barstow and Las Vegas. The US EPA, California regulatory authorities, and the Sierra Club all had to be satisfied that this project wouldn't create a big mess. That was done.
Now Molycorp complains that smuggling of rare earths out of China is pushing the price down, but they're digging them up, processing, and shipping them. Problem solved.
The complaint makes a key point. Facebook lied in their privacy policy. See page 19 of the complaint,
"Facebook Fails to Disclose That Its Private Message Processes Read, Acquire, and Use Private Message Content, in Violation of Its Express Agreements With Facebook Users." This looks like a clear ECPA violation.
One NSA director in the 1960s said "I want a thousand-megacycle machine. I'll get you the money!" There's a book, "IBM's Early Computers", which shows much of NSA's exotic hardware from the 1950s through the early 1970s. High-density tape drives, the first automatic-changing tape library (TRACTOR), the first superscalar machine (STRETCH, which, for NSA, had a special crypto processor instead of an FPU), and a number of cyrogenic machines.
NSA tried hard to get cyrogenic computing to work, from the 1960s onward. They had some successes with getting devices to work fast in the 1960s, but the early superconducting devices were gated magnetically, which meant coils and discrite devices, not ICs. So they could be made fast, but not small, which means speed of light lag within the processor becomes a bottleneck. Mainstream CMOS IC technology eventually beat out the superconducting Josephson junction stuff on both price and speed. Some time in the 1980s, IBM and NSA gave up on that. It just wasn't a win over Moore's Law.
This is yet another lame altcoin. It barely has a mention on bitcointalk. It's not really associated with someone famous. But it points the way to a potential success.
This could be a way to monetize fame. What if some major performer came out with an alt coin? They could anchor the coin value by making it exchangeable for concert tickets, downloaded tracks, and promotional merchandise. None of those things cost much to make, so they're not too vulnerable to price swings. The coin client can be combined with a music player/store client program, distributed with a few free tracks.
The problem with most of these alt coins is that you can't buy anything with them. If they were at least guaranteed to be tradeable for some music tracks and a T-shirt, there'd be some backing behind them.
It's good to have solar-powered things spread around, in case of major power grid problems. As LED street lights are installed, some of them should be solar powered. Especially in areas with a history of floods, hurricanes, earthquakes, etc.
Many USPS locations already have a kiosk with a scale and a vending machine type arrangement to do that, without the need for a postal employee. Or you can get a USPS account (which is free) and print your own bar-coded package labels with postage. Just like FedEx. There's even a discount for that, and you get free tracking.
When you use either of those methods, no postal employee has to do any data entry.
The big problem with Massive Open Online Courses is that, in most cases, the content is recycled lectures with no quality control. Stanford's machine learning course is mostly watching Andrew Ng at a blackboard, with bad handwriting. I watched a Khan Academy course on moments of inertia, and it was full of basic errors - clockwise and counterclockwise reversed, no distinction between a free body and a pinned one - errors likely to confuse anybody new to the subject.
Where's the post-production? Where's the production value? Where's the checking and Q/A? Of course students are dropping out and failing. The product quality sucks.
We have all this compute power and aren't using it to help with the process. Most of these "courses" are just streaming video with some textual material to go with it. We're not seeing systems where users solve problems and, when they get the wrong answer, the system tries to figure out what they did wrong and coach them. The Plato system did that in the 1960s. There have been systems for teaching programming which did that. But no, we just have lectures and texts.
if you want to see how to train people, look at the US military. The military has to train huge numbers of not-super-bright people in complex technical skills, and they've been doing it for decades with good success. Their approach isn't cheap; there are lots of visual aids, simulators, and setups for practicing skills. "Tell, then show, then do" is the mantra of military training.
The real big news in robotics.
on
The Year In Robotics
·
· Score: 4, Informative
More significant events:
Hon Hai Precision Industries, parent of Foxconn, has installed the first 20,000 of the "million robot army" they plan to use in their factories. (Hon Hai makes the iPhone. Apple just does the design, marketing, and some of the software. Hon Hai also makes Sony's PlayStation 3, the Nintendo Wii, Amazon's Kindle Fire, and lots of other stuff.)
Amazon bought Kiva Robotics. All those new warehouses Amazon is building will have many robots and few people. Jim Bezos has another robotics company working robots to replace the remaining people.
Most of the high-end car makers have demonstrated at least semi-automatic driving. Cadillac, BMW, Volkswagen, Mercedes, Nissan, and even Ford have demos. Tesla is still just talk.
The Baxter robot, from iRobot, may bring robotics to short-run production. Cost is low, and it's supposed to be easy to teach.
First, the author is trying to make his case look good by framing the issue in terms of "open robots". The paper could equally well be titled "Let's Legalize Killer Robots!". What he wants to to is provide legal immunity for manufacturers against harm caused by their robots. His justification for this is a law Congress passed, at the urging of the pro-gun crowd, to immunize manufacturers against suits by people injured by their guns. Even that immunity is quite limited - if a criminal shoots you, you can't sue the manufacturer. But if your gun blows up when fired, you can.
Second, robotics is open now. You can buy lots of devices you can program. At the hobbyist level, there are companies like Lynxmotion. Most of the hobbyist robots tend to be on the wimpy side, but you can buy industrial robot arms if you want.
Third, the main reason consumer robotics hasn't taken off is because the devices don't work very well. None of the robotic vacuums are very good vacuum cleaners. Even the expensive Willow Robotics robot the article mentions isn't capable of doing very much. Progress is being made, but slowly.
I suspect this guy saw the DARPA robotics challenge video (probably the jazzed-up edited version for popular consumption, not the raw videos of painfully slow teleoperation) and started pontificating.
They've built a big R/C car. All they did was put R/C servos on steering and throttle. (They don't show how they actuated the brakes.) Running this through WiFi and an iPad instead of just using a regular R/C transmitter adds lag.
Their setup looks dangerously flaky. They have an R/C servo on the throttle, with nothing to force a closed throttle on failure.
Plugging something into a USB port is only effective as an attack if autorun is turned on in Windows. You can turn it off for all pluggable devices. A file system device is still recognized as having a file system, but something has to go to the device and get a file before anything happens.
Running Windows on an ATM is lame, but common. Running a desktop version of windows, instead of Windows Embedded (which allows removing all the stuff that shouldn't be there) is just stupid.
Maybe we just need a Android app that lets you run Windows applications. You know, for those times you need to run some ancient CRM app from the corporate network on your tablet. That's probably more useful than the other way round.
Personally I find that Facebook has too many features. It sort of reminds me of Microsoft Office with this endless parade of new tiny and mostly useless features.
It's not that Facebook is complicated. It's that most of the new features involve either advertising or collecting data about you. They have value for Facebook, not the user. Facebook is pulling a Myspace. Worse, they're doing it in the phone era, where ads are more annoying due to the limited screen real estate.
Snapchat is still in the "no ads, no revenue" phase, when it's fun to use. Originally, Google didn't have ads. Originally, Facebook didn't have ads. Until recently, Twitter did not have ads. Once the ads appear, the downward spiral begins.
It would be amusing, and perhaps useful, to create a social network system that looks to the user like Twitter/Snapchat/WhatsApp, but uses XMPP/email/IRC/SMS for transport and doesn't need servers of its own. Sell the app once for $5 or so. No ads. Phone providers usually give you a mail account and an SMS number. That's all you really need. WhatsApp comes close, but they have servers and an overreaching EULA, like everybody else. The trick is to make it spam-free, which probably means you have a friends list and only they, and maybe friends once removed, can reach you.
Building cinder-block walls around transformers in the transmission power grid might not be a bad idea. Cheap, and if concrete-filled, will stop most ammo. After a decade of anti-terrorism hype, it's surprising this hasn't been done yet. Most anti-terrorism studies of electric power grids mention transformers in the transmission system as a vulnerable point.
It's not necessary to heavily protect the whole switchyard. Switchgear is easier and cheaper to replace than transformers, and less vulnerable. The transformers occupy only a small fraction of substation area.
Transformer substations are something that people, even in the utility industry, don't think about much. They're very reliable, need little attention, and are usually unmanned. So they tend to be ignored unless there's a problem.
It's embarrassing that PG&E has such poor surveillance of a major substation. The video, grainy analog black and white with slow VHS-type artifacts, means they haven't upgraded since the 1980s or 1990s. It's not like color HD cameras are expensive any more.
I don't think it would fit into the decor of your house as well as the Honeywell Round.
That was an important design consideration with the Honeywell Round. The outer plastic ring was originally available in many colors, back when Making Everything Match was considered very important in home decorating. It's still available ($26.98 at Home Depot for the heat-only model), only in beige. But you can remove the plastic ring and paint it to match the wall if you like.
Honeywell has a touch-screen, WiFi, Android/IOS enabled thermostat which also measures humidity and can decide when to run in fan-only mode and save energy. True to their color history, Honeywell lets you change the touch-screen's background color to match your decor.
The Nest isn't a bad product, but it is overhyped for what it does.
I'm not sure the article is saying anything. For example, #1 on the list is the Nest thermostat. It has a lot of words talking about Nest, but nowhere in there does it explain why nest is so amazing. It has a pretty picture, but it's hard to see why it is qualitatively better than the old fashioned thermostat.
The "old fashioned thermostat" shown is the famous Honeywell Round, usually credited to Henry Dreyfuss. It's one of the iconic objects of 20th century industrial design. The Nest thermostat copies that design. That's it's big selling point. There are other thermostats with Internet connections.
All it does is turn the HVAC on and off. It's not for use with systems where outside air intake is controllable with a damper or fan. It doesn't control fans separately from heating and cooling. It doesn't sense CO2 and humidity, and increase the air change rate when more people are present. (That last feature is a huge win for classrooms, conference rooms, and hotel function rooms.) Newer commercial building systems do all that. The Nest could have brought that technology to the home. But it didn't. It mostly just looks cool, and performs like other semi-intelligent thermostats.
Slashdot Mirror
Mod parent up. He's right. Slashdot edit staff FAIL.
This is the next step up from the "gimme, gimme" level of Kickstarter. You can still do Kickstarter projects, but now there's a tier above that, for when you need $1M or so.
Followup - He's back in prison. "Lompoc Escaped convict turns himself in".
He's probably not in the minimum-security camp any more.
OK, we have yet another mesh of processors, an idea that comes back again and again. The details of how processors communicate really matter. Is this is a totally non-shared-memory machine? Is there some shared memory, but it's slow? If there's shared memory, what are the cache consistency rules?
Historically, meshes of processors without shared memory have been painful to program. There's a long line of machines, from the nCube to the Cell, where the hardware worked but the thing was too much of a pain to program. Most designs have suffered from having too little local memory per CPU. If there's enough memory per CPU to, well, run at least a minimal OS and some jobs, then the mesh can be treated as a cluster of intercommunicating peers. That's something for which useful software exists. If all the CPUs have to be treated as slaves of a control machine, then you need all-new software architectures to handle them. This usually results in one-off software that never becomes mature.
Basic truth: we only have three successful multiprocessor architectures that are general purpose - shared-memory multiprocessors, clusters, and GPUs. Everything other than that has been almost useless except for very specialized problems fitted to the hardware. Yet this problem needs to be cracked - single CPUs are not getting much faster.
Right. But a big problem is preventing some hostile app on the user's phone from obtaining the user's secret key used in the challenge/response algorithm. With the code reverse engineered, atttackers now know where to go looking for that key and what to do with it when they have it.
Smartphones are not a secure platform. The carrier and Google (for Android) or Apple (for their phones) have total backdoor access. So does anyone who has their signing keys.
On December 19, 2013, Molycorp started up their rare earths separation plant. It's in Mountain Pass, California. So now there's a US source.
It's not that the US lacks rare earth metal resources. It's that, until recently, China was a cheaper supplier. Then the goverment of China tried to keep the price up and insisted that Chinese companies sell motors and other completed products, not raw materials. Some rare earth metal prices shot up by a factor of 20. So the Mountain Pass mine, closed in 2002, was cranked up again, this time with new equjpiment better pollution controls.
Pollution controls for a rare earth mine are a big deal. "Rare earths" are present in low concentrations, which means that a mine generates a small amount of product and huge amounts of toxic sludge. The big rare earths mine in China has the world's largest sludge pond, and it leaks. This created an environmental disaster area for tens of kilometers around. Villages have had to be evacuated because of sludge pond leaks. The Mountain Pass, California mine is less than a mile from I-15 between Barstow and Las Vegas. The US EPA, California regulatory authorities, and the Sierra Club all had to be satisfied that this project wouldn't create a big mess. That was done.
Now Molycorp complains that smuggling of rare earths out of China is pushing the price down, but they're digging them up, processing, and shipping them. Problem solved.
The complaint makes a key point. Facebook lied in their privacy policy. See page 19 of the complaint, "Facebook Fails to Disclose That Its Private Message Processes Read, Acquire, and Use Private Message Content, in Violation of Its Express Agreements With Facebook Users." This looks like a clear ECPA violation.
One NSA director in the 1960s said "I want a thousand-megacycle machine. I'll get you the money!" There's a book, "IBM's Early Computers", which shows much of NSA's exotic hardware from the 1950s through the early 1970s. High-density tape drives, the first automatic-changing tape library (TRACTOR), the first superscalar machine (STRETCH, which, for NSA, had a special crypto processor instead of an FPU), and a number of cyrogenic machines.
NSA tried hard to get cyrogenic computing to work, from the 1960s onward. They had some successes with getting devices to work fast in the 1960s, but the early superconducting devices were gated magnetically, which meant coils and discrite devices, not ICs. So they could be made fast, but not small, which means speed of light lag within the processor becomes a bottleneck. Mainstream CMOS IC technology eventually beat out the superconducting Josephson junction stuff on both price and speed. Some time in the 1980s, IBM and NSA gave up on that. It just wasn't a win over Moore's Law.
Quantum computing, though... Just maybe.
This is yet another lame altcoin. It barely has a mention on bitcointalk. It's not really associated with someone famous. But it points the way to a potential success.
This could be a way to monetize fame. What if some major performer came out with an alt coin? They could anchor the coin value by making it exchangeable for concert tickets, downloaded tracks, and promotional merchandise. None of those things cost much to make, so they're not too vulnerable to price swings. The coin client can be combined with a music player/store client program, distributed with a few free tracks.
The problem with most of these alt coins is that you can't buy anything with them. If they were at least guaranteed to be tradeable for some music tracks and a T-shirt, there'd be some backing behind them.
It's good to have solar-powered things spread around, in case of major power grid problems. As LED street lights are installed, some of them should be solar powered. Especially in areas with a history of floods, hurricanes, earthquakes, etc.
Many USPS locations already have a kiosk with a scale and a vending machine type arrangement to do that, without the need for a postal employee. Or you can get a USPS account (which is free) and print your own bar-coded package labels with postage. Just like FedEx. There's even a discount for that, and you get free tracking.
When you use either of those methods, no postal employee has to do any data entry.
The big problem with Massive Open Online Courses is that, in most cases, the content is recycled lectures with no quality control. Stanford's machine learning course is mostly watching Andrew Ng at a blackboard, with bad handwriting. I watched a Khan Academy course on moments of inertia, and it was full of basic errors - clockwise and counterclockwise reversed, no distinction between a free body and a pinned one - errors likely to confuse anybody new to the subject.
Where's the post-production? Where's the production value? Where's the checking and Q/A? Of course students are dropping out and failing. The product quality sucks.
We have all this compute power and aren't using it to help with the process. Most of these "courses" are just streaming video with some textual material to go with it. We're not seeing systems where users solve problems and, when they get the wrong answer, the system tries to figure out what they did wrong and coach them. The Plato system did that in the 1960s. There have been systems for teaching programming which did that. But no, we just have lectures and texts.
if you want to see how to train people, look at the US military. The military has to train huge numbers of not-super-bright people in complex technical skills, and they've been doing it for decades with good success. Their approach isn't cheap; there are lots of visual aids, simulators, and setups for practicing skills. "Tell, then show, then do" is the mantra of military training.
More significant events:
I just read the article. I'm not impressed.
First, the author is trying to make his case look good by framing the issue in terms of "open robots". The paper could equally well be titled "Let's Legalize Killer Robots!". What he wants to to is provide legal immunity for manufacturers against harm caused by their robots. His justification for this is a law Congress passed, at the urging of the pro-gun crowd, to immunize manufacturers against suits by people injured by their guns. Even that immunity is quite limited - if a criminal shoots you, you can't sue the manufacturer. But if your gun blows up when fired, you can.
Second, robotics is open now. You can buy lots of devices you can program. At the hobbyist level, there are companies like Lynxmotion. Most of the hobbyist robots tend to be on the wimpy side, but you can buy industrial robot arms if you want.
Third, the main reason consumer robotics hasn't taken off is because the devices don't work very well. None of the robotic vacuums are very good vacuum cleaners. Even the expensive Willow Robotics robot the article mentions isn't capable of doing very much. Progress is being made, but slowly.
I suspect this guy saw the DARPA robotics challenge video (probably the jazzed-up edited version for popular consumption, not the raw videos of painfully slow teleoperation) and started pontificating.
They've built a big R/C car. All they did was put R/C servos on steering and throttle. (They don't show how they actuated the brakes.) Running this through WiFi and an iPad instead of just using a regular R/C transmitter adds lag.
Their setup looks dangerously flaky. They have an R/C servo on the throttle, with nothing to force a closed throttle on failure.
Google has managed to come up with something even more intrusive than this classic evil scheme from The Phone Company.
Plugging something into a USB port is only effective as an attack if autorun is turned on in Windows. You can turn it off for all pluggable devices. A file system device is still recognized as having a file system, but something has to go to the device and get a file before anything happens.
Running Windows on an ATM is lame, but common. Running a desktop version of windows, instead of Windows Embedded (which allows removing all the stuff that shouldn't be there) is just stupid.
It burned up before it hit the ground, so it probably wasn't anything very big or very solid. Ice or rock, probably. Yawn.
Some space junk does make it all the way down. Titanium or stainless steel pressure tanks often make it relatively intact. Less solid stuff rarely does.
Maybe we just need a Android app that lets you run Windows applications. You know, for those times you need to run some ancient CRM app from the corporate network on your tablet. That's probably more useful than the other way round.
Fairly obvious statement, but one some newer programmers don't know. Koenig is talking about white-box testing, which is well understood.
Personally I find that Facebook has too many features. It sort of reminds me of Microsoft Office with this endless parade of new tiny and mostly useless features.
It's not that Facebook is complicated. It's that most of the new features involve either advertising or collecting data about you. They have value for Facebook, not the user. Facebook is pulling a Myspace. Worse, they're doing it in the phone era, where ads are more annoying due to the limited screen real estate.
Snapchat is still in the "no ads, no revenue" phase, when it's fun to use. Originally, Google didn't have ads. Originally, Facebook didn't have ads. Until recently, Twitter did not have ads. Once the ads appear, the downward spiral begins.
It would be amusing, and perhaps useful, to create a social network system that looks to the user like Twitter/Snapchat/WhatsApp, but uses XMPP/email/IRC/SMS for transport and doesn't need servers of its own. Sell the app once for $5 or so. No ads. Phone providers usually give you a mail account and an SMS number. That's all you really need. WhatsApp comes close, but they have servers and an overreaching EULA, like everybody else. The trick is to make it spam-free, which probably means you have a friends list and only they, and maybe friends once removed, can reach you.
Building cinder-block walls around transformers in the transmission power grid might not be a bad idea. Cheap, and if concrete-filled, will stop most ammo. After a decade of anti-terrorism hype, it's surprising this hasn't been done yet. Most anti-terrorism studies of electric power grids mention transformers in the transmission system as a vulnerable point. It's not necessary to heavily protect the whole switchyard. Switchgear is easier and cheaper to replace than transformers, and less vulnerable. The transformers occupy only a small fraction of substation area.
Transformer substations are something that people, even in the utility industry, don't think about much. They're very reliable, need little attention, and are usually unmanned. So they tend to be ignored unless there's a problem.
It's embarrassing that PG&E has such poor surveillance of a major substation. The video, grainy analog black and white with slow VHS-type artifacts, means they haven't upgraded since the 1980s or 1990s. It's not like color HD cameras are expensive any more.
They'll catch him no doubt pretty soon...
Probably. In the entire history of the Federal prison system, only about 10 people have escaped and were not eventually caught.
He'd already served 3/4 of a 4-year sentence. Less than a year to go.
I don't think it would fit into the decor of your house as well as the Honeywell Round.
That was an important design consideration with the Honeywell Round. The outer plastic ring was originally available in many colors, back when Making Everything Match was considered very important in home decorating. It's still available ($26.98 at Home Depot for the heat-only model), only in beige. But you can remove the plastic ring and paint it to match the wall if you like.
Honeywell has a touch-screen, WiFi, Android/IOS enabled thermostat which also measures humidity and can decide when to run in fan-only mode and save energy. True to their color history, Honeywell lets you change the touch-screen's background color to match your decor.
The Nest isn't a bad product, but it is overhyped for what it does.
I'm not sure the article is saying anything. For example, #1 on the list is the Nest thermostat. It has a lot of words talking about Nest, but nowhere in there does it explain why nest is so amazing. It has a pretty picture, but it's hard to see why it is qualitatively better than the old fashioned thermostat.
The "old fashioned thermostat" shown is the famous Honeywell Round, usually credited to Henry Dreyfuss. It's one of the iconic objects of 20th century industrial design. The Nest thermostat copies that design. That's it's big selling point. There are other thermostats with Internet connections.
All it does is turn the HVAC on and off. It's not for use with systems where outside air intake is controllable with a damper or fan. It doesn't control fans separately from heating and cooling. It doesn't sense CO2 and humidity, and increase the air change rate when more people are present. (That last feature is a huge win for classrooms, conference rooms, and hotel function rooms.) Newer commercial building systems do all that. The Nest could have brought that technology to the home. But it didn't. It mostly just looks cool, and performs like other semi-intelligent thermostats.