There's a fair amount of fake physics involved. Properly, the parts of the character on one side of the portal should have the gravity and momentum of that inertial frame, and as the character passes through the portal, the new frame should begin to act on the character. But the sample code in Gamasutra treat the character as a single rigid body.
It's a neat problem to make the physics correct as the character moves though a portal. It could certainly be done, even for ragdoll characters. From a gameplay perspective, it would drive players nuts. To make the gameplay tolerable, the designers of this game added a pseudo-force that tends to align the character with the local vertical. Otherwise, characters would have execute proper parachute landing falls when moving through a gravity vector change.
Character physics almost has to be fake. Trying to drive a real car via a game pad is very difficult, and trying to drive a human body via a game pad is worse.
If you want made-in-USA tools for this, try searching Google for "craigslist auto posting tool". Google offers seven paid ads for spamming tools and crackers. ("The worlds Best Selling Craigslist
software. Works with new CAPTCHA!") Three of them (including one that advertises "Only Automated Solution for the new
captcha. Nobody else is automated.") are available through Google Checkout.
This has been going on for months, despite press coverage. I'm beginning to wonder if Google is deliberately promoting tools to kill Craigslist.
The FBI ran into a similar problem with their case automation system. Investigative databases contain items like "informant 345 reports white male, 20s, tan windbreaker, called "Harry" was seen in a bar on 4th street talking about a robbery at 10th and Main last week". How do you utilize a few million items like that? The usual approach is to start with fully-identified people and work outward, but this leads to the traditional cop vice of finding info that reinforces preconceptions. True correlation of loosely identified items is tough, although there are similarity metrics which can help.
Worse, the terrorism people have to deal with names from cultures that have low name uniqueness.
If you have a central trusted key server, there's no problem, and you don't need this. The whole point of public-key encryption is to eliminate the need for a central key server.
How vulnerable is this new thing in a world with a large number of phony "notary" sites?
People used to talk about voting-based "web of trust" approaches, but that stopped working when the bad guys got zombie farms.
I actually read the report, which is quite interesting. This is the first time ever that a major high-rise building has totally collapsed from fire alone.
First, the biggest problem was that the fire sprinklers for the lower floors lost their water supply after the WTC1 collapse, which took out the city water main. The upper floor sprinklers were fed from tanks high in the building, and they were able to contain the fires on upper floors. The firefighters had to abandon the building because they didn't have a water supply.
The building had surprisingly little structural redundancy. The loss of one key interior column was enough to trigger a progressive collapse. The beams between columns were too long and not heavy enough to provide structural redundancy.
Fuel storage wasn't a factor. Paper storage and the use of open-plan bullpens was a much bigger issue. With more interior walls, the fire would have spread more slowly.
The building was built on top of a sizable power substation, but that didn't affect the fire. Nor did the basement-level fuel tanks for emergency generators. Only the small "day tanks" on higher floors fed the fire, and they weren't that big.
The recommendations in the report are surprisingly weak. They're not recommending any building code changes that would have prevented the WTC7 collapse. They're not recommending backup water supplies for sprinklers, for example. If there'd been some emergency arrangement through which the fire pumps could have drawn water from the nearby Hudson River, the building would have survived.
Emergency water supply in fires is a huge issue. In the 1989 San Francisco earthquake, the fire-fighting water supply for parts of the city was lost, despite a system that was supposed to be triply redundant. SF has a main water supply, an emergency system with a mountaintop tank, and cisterns under some key intersections. The main system was knocked out by the quake. The emergency system failed because some mains broke, and there was nobody on duty at the valve house at the tank to turn the valves to cut off the broken mains, so the tank drained out. (Due to a budget cut, that job was unstaffed. For seventy years, there'd been people at the tank on standby, with nothing to do until the next quake.) The cistern system had never been extended to the Marina district, so that backup was unavailable. However, there was yet another backup - SF fire trucks have suction hoses and pumps for use with those cisterns, and a few trucks were able to get to the edge of the Bay and suck in salt water. By stringing enough hoses together, they were able to deliver water to the fires.
It's hard to make this work in a way that doesn't allow the world to deteriorate. How does all the stuff that players destroy get repaired? Probably by a huge number of NPCs working very hard
around the clock.
The NPC AI's need persistent state, too. They need to learn from experience, so they will rebuild better defenses. Walls are built stronger. Weak points are plugged. Overlapping fields of fire are set up. Obstacles to slow up assaults go in place. Towers are built to be mutually supporting.
Checkpoints where players must disarm are put in place. NPC guards discover flanking tactics.
The day will come when the NPC AIs get smart enough to realize that the players are ruining their world and band together to exterminate the players.
I have one of those out front at the main road. It's next to the
County's sewerage lift pump, which is in a bigger and noisier rectangular box. It's not so bad if you're in an area that's rural enough that houses aren't right on top of the things.
What puzzles me is the growing size of traffic signal control boxes. Why does it take something
with about three vertical feet of 19' rack space for one traffic light? Of course, there's a vision system watching the cameras, a network node, and maybe a UPS in addition to the basic signal controller. But there are enough signal controllers you'd think those components would be more integrated.
IBM once made a desktop mainframe, the PC/370. You could run VM on it. But that was in 1985.
Since then, they've avoided offering low-end mainframe compatible machines. There's no reason IBM couldn't offer a 1U server that runs zOS for $2000 or so, but they don't. Remember, most of the software was designed to run on machines well under 100 MIPS.
As other people have pointed out, IBM-type mainframes do virtualization right. Virtualization on x86 is a hack and an afterthought, even with the newer hardware support. x86 virtualization with VT hardware creates a virtual machine that doesn't look like a bare machine with VT hardware; the virtual machine has no "ring -1". VMware actually patches code on the fly to work on older x86 hardware, which makes VMware very complex and vulnerable to bugs. The mainframe people don't have do that. On IBM-compatible mainframes, the virtual machine can look just like a real machine; you can run VM under VM under VM, and it works fine. About ten deep, it's too slow to be useful, but it works. This is good for stability.
For historical reasons, PCs have a primitive I/O architecture. The "bus" concept came from the days when the peripherals and the memory were really on the same bus. That hasn't been true for decades now, but the architecture is still set up as if it was, with peripherals seeing physical, rather than logical addresses. In mainframes, there's an I/O MMU, memory protection between the peripherals and memory, and there's a channel architecture which standardizes how peripherals talk to the computer. PCs are still stuck in the "each peripheral has its own device register layout" era, which is why we have so much trouble with drivers. The "device register" and "bus" concepts are so deeply embedded in PC thinking that FireWire, which is really a local area network, was designed to emulate a bus with device registers.
This update used to be called "C++9x", but the committee got lost in template la-la land for a decade.
Thread-local storage is a good idea. GCC and Microsoft have had it for a decade. There's still no language guarantee that a pointer to a thread-local variable can't be exported outside the thread.
But that was already true of stack variables. (Can't say "auto" any more; that keyword has been "repurposed".)
Threading is basically POSIX threads with new paint, plus some atomic primitives that have been around in incompatible forms since the 1990s. There's no improvement in thread safety.
Unordered maps were supposed to go into the original STL years ago, but somebody didn't get the paperwork in on time. Really.
It's no safer than the previous revision of C++. We're doomed to another decade of crashes and buffer overflows. Amit Yoran pointed this out when he was the director of the National Cyber Security Division at Homeland Security. (He resigned under political pressure for pointing out that Microsoft's poor security was the biggest part of the problem.)
let a web page have a Java component (for example) that doesn't run in a little box... but instead runs in the background and updates the page
I've never tried it, but there's "com.sun.java.browser.dom", which is supposed to let your
applet access the browser's Document Object Model. In keeping with the applet security model,
there are limits on what can be done to the DOM; I think access is read-only, although
the documentation isn't clear.
In typical Sun fashion, rather than having a basic API that works, this is tied in with "Project Metro" and "GlassFish", and is supposed to work with a Java applications server, so Sun can make some money on the server side. Try these JMaki examples, which correspond to simple AJAX applications but are implemented in Java.
Java itself would be a good alternative to JavaScript. At least it scales up better. But Sun insists on burying the language under a mountain of mediocre and ever-changing libraries.
I happen to know the guy who headed the RF software group for the original iPhone. He's a low-key sort, from the industrial high-reliability real time world. He did not like being yelled at by Steve Jobs. So, shortly after the first iPhones were out and working, he quit.
Apple found someone else to do the 3G version. Probably not someone from the industrial high-reliability real time world.
Image Metrics calls this "performance transfer technology". It's not really animation; it's more of a scheme for pasting face A onto actor B. Quite a bit of this already goes on; often, when
you see a stunt performer's face on screen, the face of the principal has been transferred to the image of the stunt performer. With this new technology, that can be done without matching camera angles or going through the whole "dots on the face" makeup ordeal.
The VRML people made a terrible mistake. They 1) went XML, and 2) got taken over by advertising people. The VRML effort was shut down in favor of something called "X3D", which is, exactly, VRML syntax with XML delimiters. "Now you can have spinning 3D logos with 60 characters of X3D!". This positioned X3D as an ad-delivery system, for which it's terrible.
If you bring up an old VRML viewer on a modern machine with a good broadband connection, it works great. It's still not very useful, but it does work. Most computers of 1997 didn't have enough graphics power to run VRML properly, so it was hopeless back then. (I had a machine that did, because I was using a high-end animation system. But it cost $6000 and sat in a 19 inch rack.)
You can be too early. I was interviewed by "There" when they were starting up. They were determined to make a 3D shared virtual world that would work over a dial-up modem. I told them this was going to produce a terrible user experience, drive them nuts trying to cram that data through a tiny pipe, and that by the time they got the thing going, enough users would have broadband to make a broadband-only product feasible. They stayed with dial-up, launched There just as broadband was starting to get serious market share, never really made it, and downsized when the funding ran out. There is now owned by something called "Makena Technologies", still running, and still designed for dial-up modems.
Cravath and IBM should get more credit
on
Grokking SCO's Demise
·
· Score: 4, Interesting
The real turning point in the case was when IBM decided to fight SCO's claims and put Cravath, Swaine, and Moore LLP on the job. Cravath is very good; they say of themselves "Cravath, Swaine & Moore LLP is known as the premier American law firm.", and nobody laughs. They're very organized and thorough. Cravath was the first firm to use litigation support systems (developed by IBM for an IBM case). They can't be snowed with documents; they'll put enough people and hardware on the job to deal with truckloads of materials when necessary. At times, the staff for a single case has filled a sizable office building. This is expensive, but it works.
It works especially well when the other side has voluminous but bogus claims. That's what happened with SCO. All SCO's claims were analyzed by that huge staff, checked, and countered. In the end, SCO had nothing left.
Groklaw reported on all this, but Cravath really did the work.
San Francisco has similar toilets, from JCDecaux. They're ad-supported, plus most of them charge money. JCDecaux, not the city, services them, and they do a relatively good job, which they have to do to keep the advertising contract. The San Francisco experience is that they work fine in the tourist areas and need too much maintenance in the homeless areas. SF gives homeless people a free token; it opens the toilet like coins, but the token comes back out the coin return.
Part of the problem was the insistence that they be wheelchair accessible. JCDeaux installs a smaller version in Paris, which takes up less space on the street (it will fit on most sidewalks), and isn't big enough for prostitution, drug dealing, or sleeping. But in the US, they're forced to install the big model, which is about the size of a parking space.
Palo Alto has two units. Theirs take credit cards. Really.
These things are far more expensive than they should be, costing about $1,000,000 each over 5 years. There's no good reason these things should cost far more than an SUV, but they do. I've seen the mechanism being serviced. It's put together from stock Telemechanique industrial automation components, which is reliable but is designed for one-off applications. If you built a washing machine that way, it would cost about $20,000. These things are engineered like prototypes. They need to be re-engineered for volume production and the cost brought down to under $50,000.
So science uncovers yet another way in which our world and universe are mediocre instead of special. Is this surprising?
It used to be thought that planetary systems were very rare. But that was based on speculation about how planets form. Now we know that quite a few nearby stars have Jupiter-type gas giant planets. We can't yet detect Earth-sized extrasolar planets; the smallest one sensed so far is 5x the mass of Earth.. As techniques improve, we may find smaller planets of other stars. Right now, though, it looks like our solar system isn't atypical; stars like our sun seem to have planetary systems something like ours.
Americans can't afford to waste their time doing things with their hands. That's what low wage countries are for. Americans have to concentrate on the profitable things, like banking, hedge funds, and real estate speculation. You can't get rich with machine shop skills. Or even with the skills to set up an production line. You don't get any respect for that.
A few years ago, I ran a DARPA Grand Challenge team. We had some bright young people with an interest in robotics and the ability to make complex hardware work. Where are they now? One is running a hedge fund in Santa Fe. One went to Bermuda to work for an offshore financial operation. One went to a search engine company. One headed a group developing software inside the iPhone. They're all making lots of money, but they're not doing robotics. They can't afford to.
Yes, it's sad, Yes, it's leading to the decline of the United States. But if you're young and have college loans to pay off, what can you do?
There's a lot of solar power generation going into Mojave. This project is only one of the ones going in or already running. Right now, there's about half a gigawatt of installed solar capacity at Mojave. Several companies are putting in new plants. Some use solar panels, some use concentrator cells, and some use mirrors to heat oil to make steam.
About 10 GW is needed to cover peak Southern California air conditioning load. That's what to go for, and at peak-hour bulk power prices, it makes money. Solar power and air conditioning load peak at the same time, which works out nicely. (Wind is cheaper but somewhat random. Even averaged over a wide area, you get maybe 80-90% uptime at best, so you need other sources which are "dispatchable", that is, will deliver power when asked. About 15% of capacity can be met from wind wind without a need for extra dispatchable capacity.)
10GW in 10 years is well within reach, and will probably happen from commercial activity.
10GW in 2 years is unlikely, but 10GW in 5 years is probable.
This won't help with base load. California's base load is about 19GW; that's the low level at night. That should be on nuclear power. California has about 4GW of nuclear power now, (two plants, 4 reactors) and that needs to be increased by about a factor of 4 to 5.
Siting nuclear power plants may be a solveable problem. It used to be hard to find sites for prisons, but small towns with declining industry started competing for them, and now Northern California has many prisons, all in rather remote areas. A similar competitive approach might be used to site nuclear plants. All new plants should be in green earthquake zones, toward the eastern edge of the state.
If both of those things get done, most of the rest of California's power will come from existing hydropower sources, with peaking from natural gas. Al Gore's
Reverse-engineer away all you like on a solid PKI system, see if I care. If the system is designed correctly, it should be 100% impervious to such attacks.
A PKI system alone won't help. If you have a valid, unused stored value card, and can store that data and copy it to another card exactly, you can "recharge" cards by rewriting them with the valid card data. It doesn't matter what the recorded contents are. Without validation against a database, if you can duplicate a card exactly, that's sufficient.
Some technologies have been proposed to make cards unduplicatable. The better ideas in that direction involve using some manufactured material with a random component as part of the card.
One such scheme involved mixing some retroreflective glass beads into the material mix for the magnetic stripe on the card. The position of the beads was uncontrolled during manufacture, thus embedding some random information in each card. So when a card was initialized, the position of the beads was read and encoded onto the magnetic stripe with a public key system. Cards could then be validated by reading the stripe and the bead positions, and checking the actual bead positions against the signed information about where they should be.
This only works if the random element can't be duplicated in a controlled way. Since you can load up an ink-jet printer with metallic ink and put shiny dots anywhere you want with considerable precision, this approach was not pursued. So far, nobody has a good random, un-duplicatable element that's suitable for use with a simple card scanning device.
If you pay more money, you should be able to hire more hirelings.
Having done work in both AI and game physics, I have the suspicion that the first true
AI entity will be an NPC. There's ongoing demand for smarter NPCs, they have a world
with which they can interact, they're physical within that world, not abstract
intelligences, and they compete. That's the space in which we can make progress.
Laugh now, but someday we'll be in charge - an NPC.
It would be a cute technology to develop. Active pumping shouldn't be necessary; put the drive in a glass case, like a vacuum tube, bring out the connections through the glass using wires with the same coefficient of expansion of the glass, and pump it down once at the factory. Tubes hold vacuum for decades; so should this. The bearings aren't a major problem; active magnetic bearings are no more complex than a motor, and there's been some success with passive magnetic bearings. that work on the same principle as the Levitron.
It's flying the head in vacuum that's the problem. With no air cushion, the whole job has to be done actively. Seagate has a patent (#6,473,259) on one approach to this, so the problem has been addressed.
So there are several potential approaches to ultra high rotation speed disk drives.
It's probably a technological dead end, though. You're never going to get more than 10x current rotation speeds, and the cost will go up, which makes this a lose vs. flash and other forms of nonvolatile solid state memory.
Back in the late 1970s, Bell Labs put several hundred direct-wired rain sensors around Murray Hill, NJ. (This isn't hard when you're the phone company.) They could then make "movies" of the patterns of rain when storms went through. This resulted in cute pictures, but didn't provide any predictive value, so it was dropped.
Slashdot Mirror
There's a fair amount of fake physics involved. Properly, the parts of the character on one side of the portal should have the gravity and momentum of that inertial frame, and as the character passes through the portal, the new frame should begin to act on the character. But the sample code in Gamasutra treat the character as a single rigid body.
It's a neat problem to make the physics correct as the character moves though a portal. It could certainly be done, even for ragdoll characters. From a gameplay perspective, it would drive players nuts. To make the gameplay tolerable, the designers of this game added a pseudo-force that tends to align the character with the local vertical. Otherwise, characters would have execute proper parachute landing falls when moving through a gravity vector change.
Character physics almost has to be fake. Trying to drive a real car via a game pad is very difficult, and trying to drive a human body via a game pad is worse.
If you want made-in-USA tools for this, try searching Google for "craigslist auto posting tool". Google offers seven paid ads for spamming tools and crackers. ("The worlds Best Selling Craigslist software. Works with new CAPTCHA!") Three of them (including one that advertises "Only Automated Solution for the new captcha. Nobody else is automated.") are available through Google Checkout.
This has been going on for months, despite press coverage. I'm beginning to wonder if Google is deliberately promoting tools to kill Craigslist.
The FBI ran into a similar problem with their case automation system. Investigative databases contain items like "informant 345 reports white male, 20s, tan windbreaker, called "Harry" was seen in a bar on 4th street talking about a robbery at 10th and Main last week". How do you utilize a few million items like that? The usual approach is to start with fully-identified people and work outward, but this leads to the traditional cop vice of finding info that reinforces preconceptions. True correlation of loosely identified items is tough, although there are similarity metrics which can help.
Worse, the terrorism people have to deal with names from cultures that have low name uniqueness.
If you have a central trusted key server, there's no problem, and you don't need this. The whole point of public-key encryption is to eliminate the need for a central key server. How vulnerable is this new thing in a world with a large number of phony "notary" sites?
People used to talk about voting-based "web of trust" approaches, but that stopped working when the bad guys got zombie farms.
At last, a phone from Linux fanatics! You can dial from the command line. Just type:
echo 0 >
echo 1 >
cu -l
AT+CFUN=1
AT+CPIN="<pin>"
AT+COPS
ATD<number>
You are now connected. See how easy it is!
I actually read the report, which is quite interesting. This is the first time ever that a major high-rise building has totally collapsed from fire alone.
First, the biggest problem was that the fire sprinklers for the lower floors lost their water supply after the WTC1 collapse, which took out the city water main. The upper floor sprinklers were fed from tanks high in the building, and they were able to contain the fires on upper floors. The firefighters had to abandon the building because they didn't have a water supply.
The building had surprisingly little structural redundancy. The loss of one key interior column was enough to trigger a progressive collapse. The beams between columns were too long and not heavy enough to provide structural redundancy.
Fuel storage wasn't a factor. Paper storage and the use of open-plan bullpens was a much bigger issue. With more interior walls, the fire would have spread more slowly.
The building was built on top of a sizable power substation, but that didn't affect the fire. Nor did the basement-level fuel tanks for emergency generators. Only the small "day tanks" on higher floors fed the fire, and they weren't that big.
The recommendations in the report are surprisingly weak. They're not recommending any building code changes that would have prevented the WTC7 collapse. They're not recommending backup water supplies for sprinklers, for example. If there'd been some emergency arrangement through which the fire pumps could have drawn water from the nearby Hudson River, the building would have survived.
Emergency water supply in fires is a huge issue. In the 1989 San Francisco earthquake, the fire-fighting water supply for parts of the city was lost, despite a system that was supposed to be triply redundant. SF has a main water supply, an emergency system with a mountaintop tank, and cisterns under some key intersections. The main system was knocked out by the quake. The emergency system failed because some mains broke, and there was nobody on duty at the valve house at the tank to turn the valves to cut off the broken mains, so the tank drained out. (Due to a budget cut, that job was unstaffed. For seventy years, there'd been people at the tank on standby, with nothing to do until the next quake.) The cistern system had never been extended to the Marina district, so that backup was unavailable. However, there was yet another backup - SF fire trucks have suction hoses and pumps for use with those cisterns, and a few trucks were able to get to the edge of the Bay and suck in salt water. By stringing enough hoses together, they were able to deliver water to the fires.
The records involving elected officials should be unusually valuable.
If some of those are published, that could be interesting.
It's hard to make this work in a way that doesn't allow the world to deteriorate. How does all the stuff that players destroy get repaired? Probably by a huge number of NPCs working very hard around the clock. The NPC AI's need persistent state, too. They need to learn from experience, so they will rebuild better defenses. Walls are built stronger. Weak points are plugged. Overlapping fields of fire are set up. Obstacles to slow up assaults go in place. Towers are built to be mutually supporting. Checkpoints where players must disarm are put in place. NPC guards discover flanking tactics.
The day will come when the NPC AIs get smart enough to realize that the players are ruining their world and band together to exterminate the players.
I have one of those out front at the main road. It's next to the County's sewerage lift pump, which is in a bigger and noisier rectangular box. It's not so bad if you're in an area that's rural enough that houses aren't right on top of the things.
What puzzles me is the growing size of traffic signal control boxes. Why does it take something with about three vertical feet of 19' rack space for one traffic light? Of course, there's a vision system watching the cameras, a network node, and maybe a UPS in addition to the basic signal controller. But there are enough signal controllers you'd think those components would be more integrated.
IBM once made a desktop mainframe, the PC/370. You could run VM on it. But that was in 1985. Since then, they've avoided offering low-end mainframe compatible machines. There's no reason IBM couldn't offer a 1U server that runs zOS for $2000 or so, but they don't. Remember, most of the software was designed to run on machines well under 100 MIPS.
As other people have pointed out, IBM-type mainframes do virtualization right. Virtualization on x86 is a hack and an afterthought, even with the newer hardware support. x86 virtualization with VT hardware creates a virtual machine that doesn't look like a bare machine with VT hardware; the virtual machine has no "ring -1". VMware actually patches code on the fly to work on older x86 hardware, which makes VMware very complex and vulnerable to bugs. The mainframe people don't have do that. On IBM-compatible mainframes, the virtual machine can look just like a real machine; you can run VM under VM under VM, and it works fine. About ten deep, it's too slow to be useful, but it works. This is good for stability.
For historical reasons, PCs have a primitive I/O architecture. The "bus" concept came from the days when the peripherals and the memory were really on the same bus. That hasn't been true for decades now, but the architecture is still set up as if it was, with peripherals seeing physical, rather than logical addresses. In mainframes, there's an I/O MMU, memory protection between the peripherals and memory, and there's a channel architecture which standardizes how peripherals talk to the computer. PCs are still stuck in the "each peripheral has its own device register layout" era, which is why we have so much trouble with drivers. The "device register" and "bus" concepts are so deeply embedded in PC thinking that FireWire, which is really a local area network, was designed to emulate a bus with device registers.
This update used to be called "C++9x", but the committee got lost in template la-la land for a decade.
Thread-local storage is a good idea. GCC and Microsoft have had it for a decade. There's still no language guarantee that a pointer to a thread-local variable can't be exported outside the thread. But that was already true of stack variables. (Can't say "auto" any more; that keyword has been "repurposed".)
Threading is basically POSIX threads with new paint, plus some atomic primitives that have been around in incompatible forms since the 1990s. There's no improvement in thread safety.
Unordered maps were supposed to go into the original STL years ago, but somebody didn't get the paperwork in on time. Really.
It's no safer than the previous revision of C++. We're doomed to another decade of crashes and buffer overflows. Amit Yoran pointed this out when he was the director of the National Cyber Security Division at Homeland Security. (He resigned under political pressure for pointing out that Microsoft's poor security was the biggest part of the problem.)
At least they didn't put in "concepts".
let a web page have a Java component (for example) that doesn't run in a little box... but instead runs in the background and updates the page
I've never tried it, but there's "com.sun.java.browser.dom", which is supposed to let your applet access the browser's Document Object Model. In keeping with the applet security model, there are limits on what can be done to the DOM; I think access is read-only, although the documentation isn't clear.
In typical Sun fashion, rather than having a basic API that works, this is tied in with "Project Metro" and "GlassFish", and is supposed to work with a Java applications server, so Sun can make some money on the server side. Try these JMaki examples, which correspond to simple AJAX applications but are implemented in Java.
Java itself would be a good alternative to JavaScript. At least it scales up better. But Sun insists on burying the language under a mountain of mediocre and ever-changing libraries.
I happen to know the guy who headed the RF software group for the original iPhone. He's a low-key sort, from the industrial high-reliability real time world. He did not like being yelled at by Steve Jobs. So, shortly after the first iPhones were out and working, he quit.
Apple found someone else to do the 3G version. Probably not someone from the industrial high-reliability real time world.
Image Metrics calls this "performance transfer technology". It's not really animation; it's more of a scheme for pasting face A onto actor B. Quite a bit of this already goes on; often, when you see a stunt performer's face on screen, the face of the principal has been transferred to the image of the stunt performer. With this new technology, that can be done without matching camera angles or going through the whole "dots on the face" makeup ordeal.
The VRML people made a terrible mistake. They 1) went XML, and 2) got taken over by advertising people. The VRML effort was shut down in favor of something called "X3D", which is, exactly, VRML syntax with XML delimiters. "Now you can have spinning 3D logos with 60 characters of X3D!". This positioned X3D as an ad-delivery system, for which it's terrible.
If you bring up an old VRML viewer on a modern machine with a good broadband connection, it works great. It's still not very useful, but it does work. Most computers of 1997 didn't have enough graphics power to run VRML properly, so it was hopeless back then. (I had a machine that did, because I was using a high-end animation system. But it cost $6000 and sat in a 19 inch rack.)
You can be too early. I was interviewed by "There" when they were starting up. They were determined to make a 3D shared virtual world that would work over a dial-up modem. I told them this was going to produce a terrible user experience, drive them nuts trying to cram that data through a tiny pipe, and that by the time they got the thing going, enough users would have broadband to make a broadband-only product feasible. They stayed with dial-up, launched There just as broadband was starting to get serious market share, never really made it, and downsized when the funding ran out. There is now owned by something called "Makena Technologies", still running, and still designed for dial-up modems.
The real turning point in the case was when IBM decided to fight SCO's claims and put Cravath, Swaine, and Moore LLP on the job. Cravath is very good; they say of themselves "Cravath, Swaine & Moore LLP is known as the premier American law firm.", and nobody laughs. They're very organized and thorough. Cravath was the first firm to use litigation support systems (developed by IBM for an IBM case). They can't be snowed with documents; they'll put enough people and hardware on the job to deal with truckloads of materials when necessary. At times, the staff for a single case has filled a sizable office building. This is expensive, but it works.
It works especially well when the other side has voluminous but bogus claims. That's what happened with SCO. All SCO's claims were analyzed by that huge staff, checked, and countered. In the end, SCO had nothing left.
Groklaw reported on all this, but Cravath really did the work.
San Francisco has similar toilets, from JCDecaux. They're ad-supported, plus most of them charge money. JCDecaux, not the city, services them, and they do a relatively good job, which they have to do to keep the advertising contract. The San Francisco experience is that they work fine in the tourist areas and need too much maintenance in the homeless areas. SF gives homeless people a free token; it opens the toilet like coins, but the token comes back out the coin return.
Part of the problem was the insistence that they be wheelchair accessible. JCDeaux installs a smaller version in Paris, which takes up less space on the street (it will fit on most sidewalks), and isn't big enough for prostitution, drug dealing, or sleeping. But in the US, they're forced to install the big model, which is about the size of a parking space.
Palo Alto has two units. Theirs take credit cards. Really.
These things are far more expensive than they should be, costing about $1,000,000 each over 5 years. There's no good reason these things should cost far more than an SUV, but they do. I've seen the mechanism being serviced. It's put together from stock Telemechanique industrial automation components, which is reliable but is designed for one-off applications. If you built a washing machine that way, it would cost about $20,000. These things are engineered like prototypes. They need to be re-engineered for volume production and the cost brought down to under $50,000.
So science uncovers yet another way in which our world and universe are mediocre instead of special. Is this surprising?
It used to be thought that planetary systems were very rare. But that was based on speculation about how planets form. Now we know that quite a few nearby stars have Jupiter-type gas giant planets. We can't yet detect Earth-sized extrasolar planets; the smallest one sensed so far is 5x the mass of Earth.. As techniques improve, we may find smaller planets of other stars. Right now, though, it looks like our solar system isn't atypical; stars like our sun seem to have planetary systems something like ours.
Americans can't afford to waste their time doing things with their hands. That's what low wage countries are for. Americans have to concentrate on the profitable things, like banking, hedge funds, and real estate speculation. You can't get rich with machine shop skills. Or even with the skills to set up an production line. You don't get any respect for that.
A few years ago, I ran a DARPA Grand Challenge team. We had some bright young people with an interest in robotics and the ability to make complex hardware work. Where are they now? One is running a hedge fund in Santa Fe. One went to Bermuda to work for an offshore financial operation. One went to a search engine company. One headed a group developing software inside the iPhone. They're all making lots of money, but they're not doing robotics. They can't afford to.
Yes, it's sad, Yes, it's leading to the decline of the United States. But if you're young and have college loans to pay off, what can you do?
There's a lot of solar power generation going into Mojave. This project is only one of the ones going in or already running. Right now, there's about half a gigawatt of installed solar capacity at Mojave. Several companies are putting in new plants. Some use solar panels, some use concentrator cells, and some use mirrors to heat oil to make steam.
About 10 GW is needed to cover peak Southern California air conditioning load. That's what to go for, and at peak-hour bulk power prices, it makes money. Solar power and air conditioning load peak at the same time, which works out nicely. (Wind is cheaper but somewhat random. Even averaged over a wide area, you get maybe 80-90% uptime at best, so you need other sources which are "dispatchable", that is, will deliver power when asked. About 15% of capacity can be met from wind wind without a need for extra dispatchable capacity.)
10GW in 10 years is well within reach, and will probably happen from commercial activity. 10GW in 2 years is unlikely, but 10GW in 5 years is probable.
This won't help with base load. California's base load is about 19GW; that's the low level at night. That should be on nuclear power. California has about 4GW of nuclear power now, (two plants, 4 reactors) and that needs to be increased by about a factor of 4 to 5.
Siting nuclear power plants may be a solveable problem. It used to be hard to find sites for prisons, but small towns with declining industry started competing for them, and now Northern California has many prisons, all in rather remote areas. A similar competitive approach might be used to site nuclear plants. All new plants should be in green earthquake zones, toward the eastern edge of the state.
If both of those things get done, most of the rest of California's power will come from existing hydropower sources, with peaking from natural gas. Al Gore's
Reverse-engineer away all you like on a solid PKI system, see if I care. If the system is designed correctly, it should be 100% impervious to such attacks.
A PKI system alone won't help. If you have a valid, unused stored value card, and can store that data and copy it to another card exactly, you can "recharge" cards by rewriting them with the valid card data. It doesn't matter what the recorded contents are. Without validation against a database, if you can duplicate a card exactly, that's sufficient.
Some technologies have been proposed to make cards unduplicatable. The better ideas in that direction involve using some manufactured material with a random component as part of the card. One such scheme involved mixing some retroreflective glass beads into the material mix for the magnetic stripe on the card. The position of the beads was uncontrolled during manufacture, thus embedding some random information in each card. So when a card was initialized, the position of the beads was read and encoded onto the magnetic stripe with a public key system. Cards could then be validated by reading the stripe and the bead positions, and checking the actual bead positions against the signed information about where they should be.
This only works if the random element can't be duplicated in a controlled way. Since you can load up an ink-jet printer with metallic ink and put shiny dots anywhere you want with considerable precision, this approach was not pursued. So far, nobody has a good random, un-duplicatable element that's suitable for use with a simple card scanning device.
If you pay more money, you should be able to hire more hirelings.
Having done work in both AI and game physics, I have the suspicion that the first true AI entity will be an NPC. There's ongoing demand for smarter NPCs, they have a world with which they can interact, they're physical within that world, not abstract intelligences, and they compete. That's the space in which we can make progress.
Laugh now, but someday we'll be in charge - an NPC.
It would be a cute technology to develop. Active pumping shouldn't be necessary; put the drive in a glass case, like a vacuum tube, bring out the connections through the glass using wires with the same coefficient of expansion of the glass, and pump it down once at the factory. Tubes hold vacuum for decades; so should this. The bearings aren't a major problem; active magnetic bearings are no more complex than a motor, and there's been some success with passive magnetic bearings. that work on the same principle as the Levitron.
It's flying the head in vacuum that's the problem. With no air cushion, the whole job has to be done actively. Seagate has a patent (#6,473,259) on one approach to this, so the problem has been addressed.
There's an experimental technology for hard drives which uses an electron beam for reading and writing. That might lead to high-RPM drives.
So there are several potential approaches to ultra high rotation speed disk drives. It's probably a technological dead end, though. You're never going to get more than 10x current rotation speeds, and the cost will go up, which makes this a lose vs. flash and other forms of nonvolatile solid state memory.
Back in the late 1970s, Bell Labs put several hundred direct-wired rain sensors around Murray Hill, NJ. (This isn't hard when you're the phone company.) They could then make "movies" of the patterns of rain when storms went through. This resulted in cute pictures, but didn't provide any predictive value, so it was dropped.
Oh well.
Google has a Service Level Agreement. If they have excessive downtime, you can get up to 15 days of free service. No refunds.
Tell that to your boss. It's not your problem. That's what the company signed up for. Welcome to "cloud computing".