There are commercially available solar street lights in the US. 5-day battery backup, resistant to 150MPH winds. "During the 2004 hurricane season in Florida, SLV models withstood ground zero wind conditions from category 5 hurricanes and typhoons." Just what's needed to provide light during emergencies.
A truly relevant shared agent would filter out all ads and click-through trap sites, and totally mess up the dynamic of the ad-supported Internet.
That's a feature, not a bug. We're working on the problem. So are others.
"Adblock" is just the beginning. There's Customize Google, which will remove Google text ads. It's a Firefox extension. Also removes Google ad tracking.
We have SiteTruth, which is a form of "intelligent agent" that rates sites for legitimacy, digging in various data sources and reading through the site for business addresses to find out who's behind the site. (No clear business location on a commercial site yields a bad rating.) We mostly use Yahoo search, but we also have a front end for Google which leaves the ads in, then rates both the organic search results and the ads for legitimacy.
As a general rule, advertised sites rate lower than organic search results. We see that with our system, and systems that rate by other criteria (user ratings, hostile code scanning, etc.) see similar results. This makes sense; if you're getting good positioning in organic search results, why run ads in the search engine? There's a clear "bottom-feeder effect" in search engine ads.
#1 hdtvs in stores (and sometimes in people's houses) OFTEN are displaying SD material,
Of course. The DRM system, HDCP, won't let you run multiple monitors from the same protected source. The player and monitor do a cryptographic key exchange to authenticate the monitor, then exchange session keys. So each player-to-monitor session has a unique key. You can't just split the output.
There are multiple-output HDCP-compliant splitters., and they're not cheap. $750 for a 5-output unit is typical. These are certified DRM devices that do the cryptographic handshake on both sides, decrypting and re-encrypting within a single IC, as required by the HDCP consortium. You might find these in use at a high-end video store, but the local discount electronics retailer won't bother. Our local Fry's and Costco are still piping analog S-Video into most of their HTDV monitors.
There's been some concern about this over at the Anti-Phishing Working Group. Much phishing seems to come from domains held for very short periods. But it turns out that's not "domain tasting". It's phishers buying domains with stolen credit card numbers, using retail domain registrars. After a few days, the credit card number is detected as stolen, the transaction is reversed by the bank, and the registrar deletes the domain.
This seems to be a separate problem from "domain tasting". But the "grace period" loophole that makes "domain tasting" possible also enables this scam. If registrars couldn't return domains to the TLD registry without paying, they'd have to raise their standards of customer validation.
nor would they get to feel all special ordering a 10 word drink out loud.
Exactly. That's the entire point of Starbucks. It's an ego boost. It's not about coffee.
That's why expresso machines are made to look complex and to require so much manual attention. The job could be done better by a microprocessor with a few sensors and actuators, and often it is; the manual stuff is mostly for drama.
That business is about experience, not product. Faster order processing isn't the point.
If this wasn't piracy, it would be straightforward to distribute the entire output of the RIAA via NNTP. The bandwidth consumption would be far smaller, because no file traverses a link more than once. The "p2p" approach is a horribly inefficient way of distributing data.
For those of you who don't follow this mess, SCO voluntarily went into bankruptcy in September, the day before the SCO vs. Novell trial was to start. This stopped that case. For about two months. Novell asked the bankruptcy judge to let their case continue. After frantic maneuvers from SCO, which went nowhere, the bankruptcy judge let the Novell case proceed. It will go to trial in January or February. The only remaining issue there is how much money SCO owes Novell. SCO already lost on the "who owns UNIX copyrights" issue; Novell owns them.
The IBM and Red Hat cases are still pending, so if anybody wants to buy up SCO (the company's entire market cap is only $2.4 million today), they'd be stuck with those potential liabilities, on top of the Novell liability. That's why nobody is snapping up the remains of the company.
Plan on visiting the liquidation auction some time next summer.
If you've ever worked with high-precision GPS gear, you know how frustrating it is.
I've used such gear on robot vehicles. Three satellites give you approximate latitude and longitude, but not elevation. With four, you get elevation. With five, Omnistar corrections work and you can get 15cm accuracy. There are many times when you can't see five satellites on land; some may be too low in the sky and blocked by terrain.
Plus, some GPS satellites may be down. They're not always operational.
In cities, it's hard to see five sats most of the time. Combination GPS/GLONASS units will be a big help for high-precision ground work.
The big problem with the original GLONASS sats was a short design life. They were only intended to work for a few years before replacement. The newer sats have a longer design life, so Russia should be able to keep the constellation running now. GPS sats have a useful life of about 15 years. #57 was launched on December 20, 2007, and replaces #37, launched in 1993.
That kind of toughness makes real sense in expensive mobile devices. I was surprised that the iPhone didn't come with a sapphire or diamond screen.
This isn't exotic technology today. The typical supermarket checkout scanner uses sapphire or diamond coating on the glass. That's why it can survive years of canned goods (and, for Home Depot, hand tools) being scraped across the scanner. In the checkout scanner world, plain glass lasts 2-4 weeks. For diamond, the makers claim 9 years. The sapphire vendor offers a lifetime warranty.
Those are quite expensive in anything higher than 640x480 resolution, and their smaller model already has 1024x768.
Actually, Unibrain "consumer" (about $100) and "industrial" (about $400) FireWire cameras are the same electronics in different packaging. Their industrial camera has the voltage regulator further from the imager, so its heat doesn't add noise to the image. That's about the only difference in the electronics.
Synchronizing two FireWire cameras is straightforward, too, FireWire cameras running in isochronous mode on the same FireWire tree are all running off the same clock (the "isochronous master"). If you start them in sync, they'll stay in sync. The Linux driver doesn't support multiple cameras (did that get fixed in the 2.6 kernel FireWire re-implementation?), but I once wrote a QNX driver that did, and could run multiple cameras in sync. It doesn't take any extra hardware.
All you need is solid mechanical and optical alignment between the two cameras. Yes, you can try to correct for angular misalignment in software, but if you can get the cameras aligned so that the rows on both cameras are parallel, the stereo processing is much easier.
That's heavy for what's essentially a laptop with wheels.
Apparently its main sensors are just little IR ranging devices. Those things are basically non-contact bumpers. Not too impressive. It really is a rehash of 1980s technology. I don't see much use for a 55 pound dumbbot. Robotics is way beyond that point.
This thing ought to have at least two cameras, stereo vision, and SLAM software. Wouldn't add that much to the cost, and they have the needed CPU power onboard. A pair of webcam chips mounted rigidly to the same frame, so that they stay aligned within a pixel, would make stereo vision work. You can buy stereo camera pairs for robotics, but they cost too much because they're made in tiny quantities. Made by a toy manufacturer, they'd be no more expensive than two standard webcams.
If enough houses got together, they could make a great set of fake runway lights for aircraft.
The Museum of Modern Art in New York once did that by accident. They had a piece called "Pulsa", by Patrick Clancy, in 1969-1970. This involved strobe lights in the MOMA courtyard and on adjacent buildings. Speakers played clicks as the strobes flashed, and the display was timed to produce the look and sound of something moving at hundreds of miles per hour.
The FAA made them shut it down, after a pilot reported seeing it from the air. They'd accidentally reproduced the "meatball" which identifies airport runways. In midtown Manhattan. Oops. The work was brought back up with some mods so it didn't look anything like a runway.
The site has too much junk on it. No wonder the server is overloaded.
There are several.swf objects. Some are movies. There's a Javascript picture rotator. There seems to be server-side Java; if you try vote.jsp on the site, you get a Java backtrace.
The "vote" script is amusing. The web designer seems to have copied a "suggestion box" script from somewhere, then commented out the "vote" capability. It's so PRC. The government is terrified of their people voting on anything.
Incidentally, if you've never tried gloves-and-goggles VR, it's cool for about ten minutes. Trying to do things by making gestures in the air is a huge pain. Without tactile feedback, it's tiring and inaccurate. I tried most of the VR systems in the first round, including Jaron Lainer's original system. No good.
It might not suck if the system had an end to end lag of under 10ms. "Turn head, wait for view to catch up" systems drive the user nuts. That problem was solved by "cave" type systems, where the user is surrounded by screens. Bulky, but tolerable.
Take a look at Sprint PCS coverage for Northern California. See those huge grey areas? That's "Analog Roam" territory. We're not talking about Nevada desert here, far from civilization. These are areas within fifteen miles of Silicon Valley.
Here's a video. This is still a small-movement actuator. Piezoelectric devices are usually good for a maximum strain of about 0.1%, and this stuff gets up to 0.45%. This has minor uses, but it's a long way from being an "artificial muscle", which requires
strain values around maybe 20%. This won't replace the hydraulic cylinder any time soon.
There are other materials with more promise for artificial muscles. See Artificial Muscle, Inc., which has a polymer-based material which changes length when electricity is applied. This is being used in auto-focus cameras, and they're working towards valve actuators in appliances, automotive fuel pumps, and similar short-travel applications.
Oh, please, not more Transformers. Those transformations looked so stupid. At least do something that's geometrically possible.
In the early days of morphing, I was visiting Pacific Data Images, and I stopped by a workstation where an artist had a pictures of a tiger's face and a truck front on screen. She was trying to find control points that would morph to the other without looking totally stupid. It just wasn't working. Yes, you can morph anything to anything, and you'll get a smooth transition, but that doesn't mean it's a good idea artistically.
(Incidentally, Hollywood just made a very good movie: "Charlie Wilson's War". It's a totally outrageous story that happens to be quite true. Written by Aaron Sorkin, who did most of "The West Wing".)
Cute idea. What they're trying to do, it seems, is mooch a little power from the electrostatic field gradient around the wire. This is quite feasible if you have a wire with a few KV to ground. The classic demo is to light up a fluorescent lamp by placing it vertically below a high tension line. This works partly because air is not a perfect insulator. There's an electrical path to ground; it just has a high resistance.
If the thing lands on an 11KV power line that's 10m above ground, and has a conductive part that dips 10cm below the line, it should see a voltage difference of about 90 volts. You can't draw very much current before the voltage difference disappears, but you can draw a little.
It's also possible to extract some energy magnetically. See U.S. Patent #3,202,963, "Apparatus for Illuminating Power Lines". But that approach requires heavier parts than an electrostatic approach.
Interesting precedent. Now we can ask what would happen if a repair tech found evidence of tax evasion, like two sets of books. They call it into the IRS Tax Evasion Hotline (1-800-829-0433), and apply for the reward.
Questions:
Is this an unreasonable search?
Would a court rule differently if it wasn't about "evil child pornography"?
Who is entitled to the IRS reward, the employer or the employee?
This may be a hoax. Looking in Google, the story is all over the crap tech sites and blogs, but there's nothing from Toshiba or in the mainstream press.
Small packaged fission power reactors are certainly possible. The U.S. Army had several designs working in the 1950s and 1960s. But they were one-offs, and the infrastructure to support them, including training facilities, cost more then they were worth.
Gartner sees no easy way out of this dilemma unless e-mail providers have incentives to invest in solutions to keep phishing e-mails from reaching consumers in the first place, and unless advertising networks and other "infection point" providers (which theoretically can be any legitimate Web site or service) have incentives to keep malware from being planted on their Web sites to reach unsuspecting consumers.
In practice, only a small minority of "legitimate Web sites or services" are "infection point providers". We have a little list. Right now, there are 166 major sites known to be providing material support to phishing attacks. There were 171 when The Register covered this last week, so publicity is having some effect. Most sites on the list only stay there for a few days, until somebody fixes the problem. A few sites stay on the list, and may need a clue stick applied.
These are exploits of open redirectors, DSL lines with zombies, sites that let hostile content be uploaded (uploading a hostile ".swf" file to Photobucket, for example), and out and out break-ins. These aren't sites that are cooperating with phishers; they're innocent, but often clueless, victims.
We blacklist the entire second-level domain if there's any phishing activity anywhere in the domain. This is far more effective than blacklisting by URL. Phishing sites change URLs and subdomains constantly now, so blacklisting by URL is as useless as virus scanning by signature. Yes, there's some collateral damage. It's all to sites on that list. We make the list public, and provide links to the actual phishing information (which is from PhishTank.), so major sites can fix their problems.
This part of the problem can be fixed. It just takes a hard-line approach.
Slashdot Mirror
YouTube needs some kind of flagging system:
There are commercially available solar street lights in the US. 5-day battery backup, resistant to 150MPH winds. "During the 2004 hurricane season in Florida, SLV models withstood ground zero wind conditions from category 5 hurricanes and typhoons." Just what's needed to provide light during emergencies.
The "Solar Tree" is more of an art project.
A truly relevant shared agent would filter out all ads and click-through trap sites, and totally mess up the dynamic of the ad-supported Internet.
That's a feature, not a bug. We're working on the problem. So are others.
"Adblock" is just the beginning. There's Customize Google, which will remove Google text ads. It's a Firefox extension. Also removes Google ad tracking.
We have SiteTruth, which is a form of "intelligent agent" that rates sites for legitimacy, digging in various data sources and reading through the site for business addresses to find out who's behind the site. (No clear business location on a commercial site yields a bad rating.) We mostly use Yahoo search, but we also have a front end for Google which leaves the ads in, then rates both the organic search results and the ads for legitimacy.
As a general rule, advertised sites rate lower than organic search results. We see that with our system, and systems that rate by other criteria (user ratings, hostile code scanning, etc.) see similar results. This makes sense; if you're getting good positioning in organic search results, why run ads in the search engine? There's a clear "bottom-feeder effect" in search engine ads.
#1 hdtvs in stores (and sometimes in people's houses) OFTEN are displaying SD material,
Of course. The DRM system, HDCP, won't let you run multiple monitors from the same protected source. The player and monitor do a cryptographic key exchange to authenticate the monitor, then exchange session keys. So each player-to-monitor session has a unique key. You can't just split the output.
There are multiple-output HDCP-compliant splitters., and they're not cheap. $750 for a 5-output unit is typical. These are certified DRM devices that do the cryptographic handshake on both sides, decrypting and re-encrypting within a single IC, as required by the HDCP consortium. You might find these in use at a high-end video store, but the local discount electronics retailer won't bother. Our local Fry's and Costco are still piping analog S-Video into most of their HTDV monitors.
There's been some concern about this over at the Anti-Phishing Working Group. Much phishing seems to come from domains held for very short periods. But it turns out that's not "domain tasting". It's phishers buying domains with stolen credit card numbers, using retail domain registrars. After a few days, the credit card number is detected as stolen, the transaction is reversed by the bank, and the registrar deletes the domain.
This seems to be a separate problem from "domain tasting". But the "grace period" loophole that makes "domain tasting" possible also enables this scam. If registrars couldn't return domains to the TLD registry without paying, they'd have to raise their standards of customer validation.
nor would they get to feel all special ordering a 10 word drink out loud.
Exactly. That's the entire point of Starbucks. It's an ego boost. It's not about coffee.
That's why expresso machines are made to look complex and to require so much manual attention. The job could be done better by a microprocessor with a few sensors and actuators, and often it is; the manual stuff is mostly for drama.
That business is about experience, not product. Faster order processing isn't the point.
If this wasn't piracy, it would be straightforward to distribute the entire output of the RIAA via NNTP. The bandwidth consumption would be far smaller, because no file traverses a link more than once. The "p2p" approach is a horribly inefficient way of distributing data.
The NYPD has used scooters for years; they're great in congested areas. They probably like this thing because it's quiet.
The NYPD doesn't do many car chases. NYC is too crowded. They prefer to get a few units into position ahead of the vehicle being pursued.
For those of you who don't follow this mess, SCO voluntarily went into bankruptcy in September, the day before the SCO vs. Novell trial was to start. This stopped that case. For about two months. Novell asked the bankruptcy judge to let their case continue. After frantic maneuvers from SCO, which went nowhere, the bankruptcy judge let the Novell case proceed. It will go to trial in January or February. The only remaining issue there is how much money SCO owes Novell. SCO already lost on the "who owns UNIX copyrights" issue; Novell owns them.
The IBM and Red Hat cases are still pending, so if anybody wants to buy up SCO (the company's entire market cap is only $2.4 million today), they'd be stuck with those potential liabilities, on top of the Novell liability. That's why nobody is snapping up the remains of the company.
Plan on visiting the liquidation auction some time next summer.
This is going to help.
If you've ever worked with high-precision GPS gear, you know how frustrating it is. I've used such gear on robot vehicles. Three satellites give you approximate latitude and longitude, but not elevation. With four, you get elevation. With five, Omnistar corrections work and you can get 15cm accuracy. There are many times when you can't see five satellites on land; some may be too low in the sky and blocked by terrain. Plus, some GPS satellites may be down. They're not always operational.
In cities, it's hard to see five sats most of the time. Combination GPS/GLONASS units will be a big help for high-precision ground work.
The big problem with the original GLONASS sats was a short design life. They were only intended to work for a few years before replacement. The newer sats have a longer design life, so Russia should be able to keep the constellation running now. GPS sats have a useful life of about 15 years. #57 was launched on December 20, 2007, and replaces #37, launched in 1993.
That kind of toughness makes real sense in expensive mobile devices. I was surprised that the iPhone didn't come with a sapphire or diamond screen.
This isn't exotic technology today. The typical supermarket checkout scanner uses sapphire or diamond coating on the glass. That's why it can survive years of canned goods (and, for Home Depot, hand tools) being scraped across the scanner. In the checkout scanner world, plain glass lasts 2-4 weeks. For diamond, the makers claim 9 years. The sapphire vendor offers a lifetime warranty.
Those are quite expensive in anything higher than 640x480 resolution, and their smaller model already has 1024x768.
Actually, Unibrain "consumer" (about $100) and "industrial" (about $400) FireWire cameras are the same electronics in different packaging. Their industrial camera has the voltage regulator further from the imager, so its heat doesn't add noise to the image. That's about the only difference in the electronics.
Synchronizing two FireWire cameras is straightforward, too, FireWire cameras running in isochronous mode on the same FireWire tree are all running off the same clock (the "isochronous master"). If you start them in sync, they'll stay in sync. The Linux driver doesn't support multiple cameras (did that get fixed in the 2.6 kernel FireWire re-implementation?), but I once wrote a QNX driver that did, and could run multiple cameras in sync. It doesn't take any extra hardware.
All you need is solid mechanical and optical alignment between the two cameras. Yes, you can try to correct for angular misalignment in software, but if you can get the cameras aligned so that the rows on both cameras are parallel, the stereo processing is much easier.
That's heavy for what's essentially a laptop with wheels.
Apparently its main sensors are just little IR ranging devices. Those things are basically non-contact bumpers. Not too impressive. It really is a rehash of 1980s technology. I don't see much use for a 55 pound dumbbot. Robotics is way beyond that point.
This thing ought to have at least two cameras, stereo vision, and SLAM software. Wouldn't add that much to the cost, and they have the needed CPU power onboard. A pair of webcam chips mounted rigidly to the same frame, so that they stay aligned within a pixel, would make stereo vision work. You can buy stereo camera pairs for robotics, but they cost too much because they're made in tiny quantities. Made by a toy manufacturer, they'd be no more expensive than two standard webcams.
If enough houses got together, they could make a great set of fake runway lights for aircraft.
The Museum of Modern Art in New York once did that by accident. They had a piece called "Pulsa", by Patrick Clancy, in 1969-1970. This involved strobe lights in the MOMA courtyard and on adjacent buildings. Speakers played clicks as the strobes flashed, and the display was timed to produce the look and sound of something moving at hundreds of miles per hour.
The FAA made them shut it down, after a pilot reported seeing it from the air. They'd accidentally reproduced the "meatball" which identifies airport runways. In midtown Manhattan. Oops. The work was brought back up with some mods so it didn't look anything like a runway.
The site has too much junk on it. No wonder the server is overloaded.
There are several .swf objects. Some are movies. There's a Javascript picture rotator. There seems to be server-side Java; if you try vote.jsp on the site, you get a Java backtrace.
The "vote" script is amusing. The web designer seems to have copied a "suggestion box" script from somewhere, then commented out the "vote" capability. It's so PRC. The government is terrified of their people voting on anything.
It's been done. Remember the Nintendo PowerGlove?
Incidentally, if you've never tried gloves-and-goggles VR, it's cool for about ten minutes. Trying to do things by making gestures in the air is a huge pain. Without tactile feedback, it's tiring and inaccurate. I tried most of the VR systems in the first round, including Jaron Lainer's original system. No good.
It might not suck if the system had an end to end lag of under 10ms. "Turn head, wait for view to catch up" systems drive the user nuts. That problem was solved by "cave" type systems, where the user is surrounded by screens. Bulky, but tolerable.
Take a look at Sprint PCS coverage for Northern California. See those huge grey areas? That's "Analog Roam" territory. We're not talking about Nevada desert here, far from civilization. These are areas within fifteen miles of Silicon Valley.
This stuff has been around since at least 2002. You can buy it commercially.
Here's a video. This is still a small-movement actuator. Piezoelectric devices are usually good for a maximum strain of about 0.1%, and this stuff gets up to 0.45%. This has minor uses, but it's a long way from being an "artificial muscle", which requires strain values around maybe 20%. This won't replace the hydraulic cylinder any time soon.
There are other materials with more promise for artificial muscles. See Artificial Muscle, Inc., which has a polymer-based material which changes length when electricity is applied. This is being used in auto-focus cameras, and they're working towards valve actuators in appliances, automotive fuel pumps, and similar short-travel applications.
Oh, please, not more Transformers. Those transformations looked so stupid. At least do something that's geometrically possible.
In the early days of morphing, I was visiting Pacific Data Images, and I stopped by a workstation where an artist had a pictures of a tiger's face and a truck front on screen. She was trying to find control points that would morph to the other without looking totally stupid. It just wasn't working. Yes, you can morph anything to anything, and you'll get a smooth transition, but that doesn't mean it's a good idea artistically.
(Incidentally, Hollywood just made a very good movie: "Charlie Wilson's War". It's a totally outrageous story that happens to be quite true. Written by Aaron Sorkin, who did most of "The West Wing".)
Cute idea. What they're trying to do, it seems, is mooch a little power from the electrostatic field gradient around the wire. This is quite feasible if you have a wire with a few KV to ground. The classic demo is to light up a fluorescent lamp by placing it vertically below a high tension line. This works partly because air is not a perfect insulator. There's an electrical path to ground; it just has a high resistance.
If the thing lands on an 11KV power line that's 10m above ground, and has a conductive part that dips 10cm below the line, it should see a voltage difference of about 90 volts. You can't draw very much current before the voltage difference disappears, but you can draw a little.
It's also possible to extract some energy magnetically. See U.S. Patent #3,202,963, "Apparatus for Illuminating Power Lines". But that approach requires heavier parts than an electrostatic approach.
Interesting precedent. Now we can ask what would happen if a repair tech found evidence of tax evasion, like two sets of books. They call it into the IRS Tax Evasion Hotline (1-800-829-0433), and apply for the reward.
Questions:
FireFox was supposed to be getting a JIT compiler for JavaScript. It's the one from the Flash player, where it runs ActionScript. That's apparently now expected in 2008. Then we'll see some real improvement.
This may be a hoax. Looking in Google, the story is all over the crap tech sites and blogs, but there's nothing from Toshiba or in the mainstream press.
Small packaged fission power reactors are certainly possible. The U.S. Army had several designs working in the 1950s and 1960s. But they were one-offs, and the infrastructure to support them, including training facilities, cost more then they were worth.
We can't actually build a small implantable GPS yet. Passive RFID tags, yes; GPS receiver with uplink, no.
Well, in theory you could build a pacemaker-sized device powered by a nuclear battery, but that would take major surgery to install, and approval from the FDA and DOE.
Gartner sees no easy way out of this dilemma unless e-mail providers have incentives to invest in solutions to keep phishing e-mails from reaching consumers in the first place, and unless advertising networks and other "infection point" providers (which theoretically can be any legitimate Web site or service) have incentives to keep malware from being planted on their Web sites to reach unsuspecting consumers.
In practice, only a small minority of "legitimate Web sites or services" are "infection point providers". We have a little list. Right now, there are 166 major sites known to be providing material support to phishing attacks. There were 171 when The Register covered this last week, so publicity is having some effect. Most sites on the list only stay there for a few days, until somebody fixes the problem. A few sites stay on the list, and may need a clue stick applied.
These are exploits of open redirectors, DSL lines with zombies, sites that let hostile content be uploaded (uploading a hostile ".swf" file to Photobucket, for example), and out and out break-ins. These aren't sites that are cooperating with phishers; they're innocent, but often clueless, victims.
We blacklist the entire second-level domain if there's any phishing activity anywhere in the domain. This is far more effective than blacklisting by URL. Phishing sites change URLs and subdomains constantly now, so blacklisting by URL is as useless as virus scanning by signature. Yes, there's some collateral damage. It's all to sites on that list. We make the list public, and provide links to the actual phishing information (which is from PhishTank.), so major sites can fix their problems.
This part of the problem can be fixed. It just takes a hard-line approach.