Stereo is useful for getting distance to nearby obstacles with edges. It's not
that useful on uniform pavement (no edges to register), gravel (detail too fine), wet surfaces, etc. What you really get from stereo is depth information of varying quality across the image.
Stereo from motion is in theory more useful, since the baseline is much larger, but it's much harder to do. Humans don't r seally do stereo vision beyond a few meters; our eyes are too close together. Depth from motion is how we really resolve the world.
One of the neater ideas in stereo vision is the Triclops, three cameras arranged in a triangle. With three cameras, most of the ambiguities that occur with two cameras go away.
Yes, actives are just too visible. It doesn't matter much in urban environments, though; it's not like you can hide an Army truck driving through a town.
They know he's in Portland. Once they know that, he has to be on either cable or DSL, or mooching off someone else's nearby connection.
The FBI could ask the cable company to reboot, in sequence, the router for each cable segment. When the right cable segment went momentarily offline, the streaming video would stop for a moment. Similarly, each DSLAM could be restarted.
That would narrow it down to a hundred houses or so.
I'd have to question your "returned power decreases as the fourth power of the distance" claim....
With lidar, you have a coherent focused beam on the way there.
Not with a flash LIDAR. There's one big broad flash spread by a beam spreader, not a mechanically scanned narrow beam. I'm amazed that it works over substantial distances. There aren't that many photons coming back per pixel. They don't spread the beam very wide; 1 to 9 degrees is typical for the longer ranged units. At shorter ranges, a wider beam can be used. The detectors have to be really good. They use GaInAs, not PIN diodes. The semiconductor device physics behind this thing is impressive.
I know about the SwissRanger thing; it doesn't have enough range for outdoor operation, but it's a nice little package. That thing modulates an array of LEDs at 20MHz or so, then measures the phase shift of the returned signal in a detector array of about 10K pixels. That thing ought to sell for webcam prices, but it's made by hand in Switzerland and costs too much.
The original article posted to Slashdot doesn't say much; you have to check the ASC web site and read the papers to see how it works.
This is a big step forward. I know this technology. Back in 2004, when we were putting our DARPA Grand Challenge vehicle together, I went down to Advanced Scientific Concepts in Santa Barbara to see the thing. Back then, they had a prototype that worked, but it was on an optical bench (one of those big plates with screw holes to which you attach optical components), nowhere near ready to go on a vehicle. It was just too early.
We had to go with SICK rotating-mirror line scanners, like everybody else. But I was convinced it was the right direction to go, and I dragged a venture capitalist who had some underperforming photonics companies down there to see the thing. He didn't want to fund them, because they were too far from a consumer product; the near term market was DoD-only.
ASC kept working, and by 2006 they had working portable prototypes. By 2007, you could buy a LIDAR about the size of a large-format camera for about $100,000. Now they've downsized it further.
Unlike the laser scanners with spinning mirrors or sensors, which is what everyone else uses, this technology has no moving parts. The system has two main components - a pulse laser with diffusing optics, and a detection and timing IC with one LIDAR receiver per pixel. Neither of these is inherently expensive in quantity. It may take a while to get this down to webcam prices, but $1000 is a reasonable near-term target.
It's amazing that this can be done in an eye-safe way, since this approach is subject to the radar equation - returned power decreases as the fourth power of the distance. But the detectors can be made good enough. Some of their more sensitive detectors use a photomultiplier tube technology, like a night vision system. Night vision systems use a photoelectric detector plate - when a photon hits it, an electron pops out. Electric fields are used to accelerate the electron, which then hits one of the electron detectors on a specially designed IC. Photomultipliers have been around for decades, and can detect single photons. The neat thing about the photoelectric effect is that it's at the atomic level, and happens in picoseconds. So it can be used as a light amplifier for a time-of-flight LIDAR.
The current generation of compact sensor is 128x128 pixels at 30Hz. The sensors are currently smaller than the lasers, but for smaller robots where you need only 10m of range or so, a smaller laser can be used.
This is the sensor that will make automatic driving commercially feasible.
This is basically about internal U.S. Government computer security.
The problem is that the last three agencies assigned this task blew it. Early on, computer security was under NIST, which is really the old National Bureau of Standards.
They were just an advisory agency on this.
There was also an NSA effort, about which more later.
There's a National Cyber Security Division of Homeland Security. When it was set up, it was headed by Amit Yoran, who actually knew something about the subject. He was unpopular because he publicly mentioned the vulnerabilities of Microsoft operating systems as the biggest single problem. So he was replaced by Gregory Garcia, a lawyer and 3COM's lobbyist in Washington, who has accomplished little, if anything.
The General Services Administration, which handles public buildings and purchasing for most of the U.S. Government, has a role in computer security, but they haven't accomplished much. other than some vendor evaluation.
NSA first got into computer security in the 1980s, when I had some dealings with them. They had an institutional problem. First, it wasn't about the USSR, on which NSA used to be narrowly focused. Second, the computer security effort was located at the "Friendship Annex", which was NSA's lower-security facility near Friendship Airport (now BWI). FANX was where NSA's less important stuff was done - personnel, accounting, etc. Being assigned to FANX was a big career step down within NSA.
NSA went at computer security in the same way they went at safes and locks - you build it, they break it. NSA policy on evaluating the security of computer products was that the vendor got two tries. On try one, NSA told the vendor what was wrong. Try two was pass/fail - if they could break it, it flunked, and went on the rejected list. Vendors hated this.
Under heavy pressure from vendors, security evaluation was outsourced to third party companies, and vendors could retry forever until they wore down the evaluators. The higher levels of security (fully verified everything) were dropped from the evaluation criteria.
NSA Secure Linux was a good idea that didn't really catch on. Most Linux people don't get the point of NSA Secure Linux. It's not about making Linux more secure. It's about getting applications rewritten to work under a tight security model. Unless applications are rewritten to have only very small and heavily verified trusted parts, NSA Secure Linux doesn't help much.
By a "good SSL certificate", I mean one that identifies the business. SiteTruth ignores those "domain control only validated" "Instant SSL" certs. They don't certify much of anything, other than that an e-mail sent to the domain gets to the party that requested the certificate.
It's interesting to see the Web without the bottom-feeders. Most of the spam blogs, phishing sites, link farms, landing pages, directory pages, "affiliates", and related junk go away. What's left is either has a real business behind it, or is noncommercial and ad-free.
The basic criterion is "can we associate a real-world name and address with this business"? We have about five ways of finding the business behind the web site, including scanning the web site text for mailing addresses, and if all of them come up empty, they're toast.
Yes, some legitimate business sites are down-rated because we can't identify the business behind them. It's the responsibility of the web site operator to clearly identify the business behind the web site; in some jurisdictions, it's a criminal offense not to. If they don't, that's their problem. We provide some webmaster tools to help with technical issues in identification; there's no mystery about what's going on.
There's a problem with fake names and addresses, but that's a felony (wire fraud or identity theft), so it's rare. We check phishing databases for those.
The conventional wisdom is that these problems can't be solved. The conventional wisdom is wrong.
I hope Google really does this. They need to, to restore their "don't be evil" reputation. Arguably, Google went over to the dark side when they started offering domain parking.. "Maximize revenue on your parked pages with Google AdSense for domains", they advertise. (Insert Darth Vader quote here.)
"Domain tasting" is a drain on the anti-fraud systems of the Internet. All those domain changes help conceal phishing attacks, many of which involve buying domains with stolen credit cards and exploiting them before the credit card transaction is reversed. Blacklist systems like McAfee SiteAdvisor and PhishTank are always running behind the domain changes.
We rate sites at SiteTruth, and all those domain changes are a headache for us. I'm considering taking the position that all domains less than 30 days old are junk, unless they have a good SSL certificate.
Is that too severe, or a good idea? Comments?
There are
1351 iPhones on eBay right now. Most of them are unlocked. Some of the sellers have hundreds of iPhones. So a sizable chunk of that unsold inventory is over on eBay.
It's like the early days of the PS3, when eBay sellers overbought and there was soon a glut of the things on eBay for months, selling below retail.
Actually, Microsoft isn't doing that badly. They're finally profitable in games. The original XBox was a huge financial loss, and there was a real question whether investor pressure would make them drop out of the game market. But they've finally been able to make money in that area with the XBox 360.
Operating systems are profitable, but they're not the cash cow they used to be. Microsoft Office remains the big moneymaker.
Microsoft is not doing well in the ad-supported online space, though.
The Martian one can use rockets and parachutes as well as glide.
It turns out that we don't actually know how to land a big craft on Mars. The atmosphere is thin for aerobraking and parachutes, but the gravity is high enough that a powered descent takes considerable fuel. So more mass is required for the descent stage than previously thought. This cascades back into a much larger launch vehicle.
At Adult Entertainment Expo two weeks ago, the Fyre set-top box was launched. This downloads porn from the Internet and outputs to standard TVs. Pay per view, Fyre handles the billing. They launch in April, with 20,000 movies.
Finally, TV-based video on demand with a big catalog. Another first for the adult industry.
The real problem isn't on the browser side. It's in Dreamweaver, the most popular web page design tool. Dreamweaver does not create valid HTML or XHTML. Not even close. Create a page in Dreamweaver for anything later than HTML 3.2 and run it through the W3C validator. It will fail.
The basic problem is that the Dreamweaver people never really figured out what to do about CSS. In theory, CSS is supposed to have some abstract model of the format of some block of text, like TeX does. In practice, there's usually a big block of CSS with machine-generated names at the beginning of the web page. There's a fundamental disconnect between the CSS model and the Dreamweaver "Properties" box. So Dreamweaver is still inserting I, B and FONT tags.
In layout, Dreamweaver does table-based layout quite well, but DIV/FLOAT/CLEAR layout isn't handled well. Much of this is due to the limitations of the DIV/FLOAT/CLEAR approach. Tables are a 2D grid system, and map well to the drag-the-lines editor in Dreamweaver. DIV/FLOAT/CLEAR doesn't map well to a visual layout editor.
The end result is a mess. And HTML 5 doesn't help.
What we need is not used-car salesmen with delusions of grandeur. What we need is better truth-in-advertising regulation. Like this:
The use of the phrase "up to" or synonym thereof in connection with any service quantity is per se deceptive, unless an "at least" guaranteed value is also provided and given equal or greater prominence.
The advertisement of an introductory rate is per se deceptive, unless the highest rate after the introductory period is also provided and given equal or greater prominence.
Advertised rates must include all charges and taxes except for state and local sales taxes.
Advertisements mentioning "rebates" must mention the non-rebate price more prominently than the price after rebate, unless absolutely no conditions are attached to the rebate offer and the rebate offer does not require the consumer to pay, at the time of sale, a price higher than the after-rebate price.
Any customer contract which allows the carrier to change the terms of the contract during the period of the agreement is void as against public policy.
Advertisements must use generic terms for features, rather than proprietary terms, to allow comparisons between vendors. (For example, "World Wide Web access" rather than "Sprint PCS Vision", and "Push to talk intercom" instead of "ReadyLink")
It's discouraging reading this. Especially since I knew some of the Cyc people back in the 1980s, when they were pursuing the same idea. They're still at it. You can even train their system if you like. But after twenty years of their claiming "Strong AI, Real Soon Now", it's probably not happening.
I went through Stanford CS back when it was just becoming clear that "expert systems" were really rather dumb and weren't going to get smarter. Most of the AI faculty was in denial about that. Very discouraging. The "AI Winter" followed; all the startups went bust, most of the research projects ended, and there was a big empty room of cubicles labeled "Knowledge Systems Laboratory" on the second floor of the Gates Building. I still wonder what happened to the people who got degrees in "Knowledge Engineering". "Do you want fries with that?"
MIT went into a phase where Rod Brooks took over the AI Lab and put everybody on little dumb robots, at roughly the Lego Mindstorms level. Minsky bitched that all the students were soldering instead of learning theory. After a decade or so, it became clear that reactive robot AI could get you to insect level, but no further. Brooks went into the floor-cleaning business (Roomba, Scooba, Dirt Dog, etc.) with the technology, with some success.
Then came the DARPA Grand Challenge. Dr. Tony Tether, the head of DARPA, decided that AI robotics needed a serious kick in the butt. That's what the DARPA Grand Challenge was really all about. It was made clear to the universities receiving DARPA money that if they didn't do well in that game, the money supply would be turned off. It worked. Levels of effort not before seen on a single AI project produced some good results. Stanford had to replace many of the old faculty, but that worked out well in the end.
This is, at last, encouraging. The top-down strong AI problem was just too hard.
Insect-level AI, with no world model, was too dumb. But robot vehicle AI, with world models updated by sensors, is now real. So there's progress. The robot vehicle problem is nice because it's so unforgiving. The thing actually has to work; you can't hand-wave around the problems.
The classic bit of hubris in AI, by the way, is to have a good idea and then think it's generally applicable. AI has been through this too many times - the General Problem Solver, inference by theorem proving, neural nets, expert systems, neural nets again, and behavior-based AI. Each of those ideas has a ceiling which has been reached.
It's possible to get too deep into some of these ideas. The people there are brilliant, but narrow, and the culture supports this. MIT has "Nerd Pride" buttons. As someone recruiting me for the Media Lab once said, "There are fewer distractions out here" (It was sleeting.) It sounds like that's what happened to these two young people.
"Monster appears in deserted area" has been done many times. My favorite is "The Navy vs. The Night Monsters". Monsters appear on tropical island. After various adventures, people on island get on radio and call in an air strike. Jet fighters are dispatched, blow up monsters. Final scene shows black smoke rising out of jungle. No more monsters.
The 1950s were a very confident period in the US. After winning WWII, no conceivable monster or monsters looked like a serious threat. '50s SF reflects that.
I understand what they're getting at. What makes modern SQL-driven databases so useful is that they optimize queries. If you're asking for every entry in A that's also in B, any modern database will check whether it's faster to look up every A in B, every B in A, or do a match where both databases are read through sequentially by the same key. The best choice depends on the database record counts, available indices, and key types and lengths. The database system figures that out; it's not in the SQL query.
So the user says what they want, and the system figures out how to do it. It's "do what I mean" that really works. We don't see enough of that in programming.
Google search itself works much more like a database than a map/reduce system. Think about what has to happen when you search for multiple keywords. That's a join, and joins on big data sets take forever if you don't have the right data structures and an optimizer.
It's amusing to read this. This guy apparently works for Intel's "find ways to use more CPU time" department. Back when I was working on physics engines, I encountered that group.
Actually, the Holy Grail isn't real time ray tracing. It's real time radiosity. Ray-tracing works backwards from the viewpoint; radiosity works outward from the light sources. All the high-end 3D packages have radiosity renderers now. Here's a typical radiosity image. of a kitchen. Radiosity images are great for interiors, and architects now routinely use them for rendering buildings. Lighting effects work like they do in the real world. In a radiosity renderer, you don't have to add phony light sources to make up for the lack of diffuse lighting.
There's a subtle effect that appears in radiosity images but not ray-traced images. Look at the kitchen image and look for an inside corner. Notice the dark band at the inside corner. Look at an inside corner in the real world and you'll see that, too. Neither ray-tracing nor traditional rendering produces that effect, and it's a cue the human vision system uses to resolve corners. The dark band appears as the light bounces back and forth between the two corners, with more light absorbed on each bounce. Radiosity rendering is iterative; you render the image with the starting light sources, then re-render with each illuminated surface as a light source. Each rendering cycle improves the image, until, somewhere around 5 to 50 cycles, the bounced light has mostly been absorbed.
There are ways to precompute light maps from radiosity, then render in real time with an ordinary renderer, and those yield better-looking images of diffuse surfaces than ray-tracing would. Some games already do this. There's a demo of true real-time radiosity, but it doesn't have the "dark band in corners" effect, so it's not doing very many light bounces. Geometrics has a commercial real-time game rendering system.
Ray-tracing can get you "ooh, shiny thing", but radiosity can get to "is that real?"
The thing NASA does best is PR. Of which they do too much. Their PR budget should be cut back, and NASA's funding for non-flight projects should be shifted to NSF.
This isn't good. Sun has a tendency to bury a basically good technology under a ton of half-debugged crap. Watch for "you have to have Java xxx installed to run MySQL".
Slashdot Mirror
Stereo is useful for getting distance to nearby obstacles with edges. It's not that useful on uniform pavement (no edges to register), gravel (detail too fine), wet surfaces, etc. What you really get from stereo is depth information of varying quality across the image.
Stereo from motion is in theory more useful, since the baseline is much larger, but it's much harder to do. Humans don't r seally do stereo vision beyond a few meters; our eyes are too close together. Depth from motion is how we really resolve the world.
One of the neater ideas in stereo vision is the Triclops, three cameras arranged in a triangle. With three cameras, most of the ambiguities that occur with two cameras go away.
The video is encoded with some wierd codec (WMV-9?) that tried to install itself but the firewall wouldn't let it. Bleah.
Yes, actives are just too visible. It doesn't matter much in urban environments, though; it's not like you can hide an Army truck driving through a town.
They know he's in Portland. Once they know that, he has to be on either cable or DSL, or mooching off someone else's nearby connection.
The FBI could ask the cable company to reboot, in sequence, the router for each cable segment. When the right cable segment went momentarily offline, the streaming video would stop for a moment. Similarly, each DSLAM could be restarted. That would narrow it down to a hundred houses or so.
I'd have to question your "returned power decreases as the fourth power of the distance" claim. ...
With lidar, you have a coherent focused beam on the way there.
Not with a flash LIDAR. There's one big broad flash spread by a beam spreader, not a mechanically scanned narrow beam. I'm amazed that it works over substantial distances. There aren't that many photons coming back per pixel. They don't spread the beam very wide; 1 to 9 degrees is typical for the longer ranged units. At shorter ranges, a wider beam can be used. The detectors have to be really good. They use GaInAs, not PIN diodes. The semiconductor device physics behind this thing is impressive.
I know about the SwissRanger thing; it doesn't have enough range for outdoor operation, but it's a nice little package. That thing modulates an array of LEDs at 20MHz or so, then measures the phase shift of the returned signal in a detector array of about 10K pixels. That thing ought to sell for webcam prices, but it's made by hand in Switzerland and costs too much.
The original article posted to Slashdot doesn't say much; you have to check the ASC web site and read the papers to see how it works.
This is a big step forward. I know this technology. Back in 2004, when we were putting our DARPA Grand Challenge vehicle together, I went down to Advanced Scientific Concepts in Santa Barbara to see the thing. Back then, they had a prototype that worked, but it was on an optical bench (one of those big plates with screw holes to which you attach optical components), nowhere near ready to go on a vehicle. It was just too early. We had to go with SICK rotating-mirror line scanners, like everybody else. But I was convinced it was the right direction to go, and I dragged a venture capitalist who had some underperforming photonics companies down there to see the thing. He didn't want to fund them, because they were too far from a consumer product; the near term market was DoD-only.
ASC kept working, and by 2006 they had working portable prototypes. By 2007, you could buy a LIDAR about the size of a large-format camera for about $100,000. Now they've downsized it further.
Unlike the laser scanners with spinning mirrors or sensors, which is what everyone else uses, this technology has no moving parts. The system has two main components - a pulse laser with diffusing optics, and a detection and timing IC with one LIDAR receiver per pixel. Neither of these is inherently expensive in quantity. It may take a while to get this down to webcam prices, but $1000 is a reasonable near-term target.
It's amazing that this can be done in an eye-safe way, since this approach is subject to the radar equation - returned power decreases as the fourth power of the distance. But the detectors can be made good enough. Some of their more sensitive detectors use a photomultiplier tube technology, like a night vision system. Night vision systems use a photoelectric detector plate - when a photon hits it, an electron pops out. Electric fields are used to accelerate the electron, which then hits one of the electron detectors on a specially designed IC. Photomultipliers have been around for decades, and can detect single photons. The neat thing about the photoelectric effect is that it's at the atomic level, and happens in picoseconds. So it can be used as a light amplifier for a time-of-flight LIDAR.
The current generation of compact sensor is 128x128 pixels at 30Hz. The sensors are currently smaller than the lasers, but for smaller robots where you need only 10m of range or so, a smaller laser can be used.
This is the sensor that will make automatic driving commercially feasible.
This is basically about internal U.S. Government computer security. The problem is that the last three agencies assigned this task blew it. Early on, computer security was under NIST, which is really the old National Bureau of Standards. They were just an advisory agency on this. There was also an NSA effort, about which more later.
There's a National Cyber Security Division of Homeland Security. When it was set up, it was headed by Amit Yoran, who actually knew something about the subject. He was unpopular because he publicly mentioned the vulnerabilities of Microsoft operating systems as the biggest single problem. So he was replaced by Gregory Garcia, a lawyer and 3COM's lobbyist in Washington, who has accomplished little, if anything.
The General Services Administration, which handles public buildings and purchasing for most of the U.S. Government, has a role in computer security, but they haven't accomplished much. other than some vendor evaluation.
NSA first got into computer security in the 1980s, when I had some dealings with them. They had an institutional problem. First, it wasn't about the USSR, on which NSA used to be narrowly focused. Second, the computer security effort was located at the "Friendship Annex", which was NSA's lower-security facility near Friendship Airport (now BWI). FANX was where NSA's less important stuff was done - personnel, accounting, etc. Being assigned to FANX was a big career step down within NSA.
NSA went at computer security in the same way they went at safes and locks - you build it, they break it. NSA policy on evaluating the security of computer products was that the vendor got two tries. On try one, NSA told the vendor what was wrong. Try two was pass/fail - if they could break it, it flunked, and went on the rejected list. Vendors hated this.
Under heavy pressure from vendors, security evaluation was outsourced to third party companies, and vendors could retry forever until they wore down the evaluators. The higher levels of security (fully verified everything) were dropped from the evaluation criteria.
NSA Secure Linux was a good idea that didn't really catch on. Most Linux people don't get the point of NSA Secure Linux. It's not about making Linux more secure. It's about getting applications rewritten to work under a tight security model. Unless applications are rewritten to have only very small and heavily verified trusted parts, NSA Secure Linux doesn't help much.
By a "good SSL certificate", I mean one that identifies the business. SiteTruth ignores those "domain control only validated" "Instant SSL" certs. They don't certify much of anything, other than that an e-mail sent to the domain gets to the party that requested the certificate.
It's interesting to see the Web without the bottom-feeders. Most of the spam blogs, phishing sites, link farms, landing pages, directory pages, "affiliates", and related junk go away. What's left is either has a real business behind it, or is noncommercial and ad-free.
The basic criterion is "can we associate a real-world name and address with this business"? We have about five ways of finding the business behind the web site, including scanning the web site text for mailing addresses, and if all of them come up empty, they're toast.
Yes, some legitimate business sites are down-rated because we can't identify the business behind them. It's the responsibility of the web site operator to clearly identify the business behind the web site; in some jurisdictions, it's a criminal offense not to. If they don't, that's their problem. We provide some webmaster tools to help with technical issues in identification; there's no mystery about what's going on.
There's a problem with fake names and addresses, but that's a felony (wire fraud or identity theft), so it's rare. We check phishing databases for those.
The conventional wisdom is that these problems can't be solved. The conventional wisdom is wrong.
I hope Google really does this. They need to, to restore their "don't be evil" reputation. Arguably, Google went over to the dark side when they started offering domain parking.. "Maximize revenue on your parked pages with Google AdSense for domains", they advertise. (Insert Darth Vader quote here.)
"Domain tasting" is a drain on the anti-fraud systems of the Internet. All those domain changes help conceal phishing attacks, many of which involve buying domains with stolen credit cards and exploiting them before the credit card transaction is reversed. Blacklist systems like McAfee SiteAdvisor and PhishTank are always running behind the domain changes.
We rate sites at SiteTruth, and all those domain changes are a headache for us. I'm considering taking the position that all domains less than 30 days old are junk, unless they have a good SSL certificate. Is that too severe, or a good idea? Comments?
There are 1351 iPhones on eBay right now. Most of them are unlocked. Some of the sellers have hundreds of iPhones. So a sizable chunk of that unsold inventory is over on eBay.
It's like the early days of the PS3, when eBay sellers overbought and there was soon a glut of the things on eBay for months, selling below retail.
Actually, Microsoft isn't doing that badly. They're finally profitable in games. The original XBox was a huge financial loss, and there was a real question whether investor pressure would make them drop out of the game market. But they've finally been able to make money in that area with the XBox 360. Operating systems are profitable, but they're not the cash cow they used to be. Microsoft Office remains the big moneymaker.
Microsoft is not doing well in the ad-supported online space, though.
The Martian one can use rockets and parachutes as well as glide.
It turns out that we don't actually know how to land a big craft on Mars. The atmosphere is thin for aerobraking and parachutes, but the gravity is high enough that a powered descent takes considerable fuel. So more mass is required for the descent stage than previously thought. This cascades back into a much larger launch vehicle.
At Adult Entertainment Expo two weeks ago, the Fyre set-top box was launched. This downloads porn from the Internet and outputs to standard TVs. Pay per view, Fyre handles the billing. They launch in April, with 20,000 movies.
Finally, TV-based video on demand with a big catalog. Another first for the adult industry.
The real problem isn't on the browser side. It's in Dreamweaver, the most popular web page design tool. Dreamweaver does not create valid HTML or XHTML. Not even close. Create a page in Dreamweaver for anything later than HTML 3.2 and run it through the W3C validator. It will fail.
The basic problem is that the Dreamweaver people never really figured out what to do about CSS. In theory, CSS is supposed to have some abstract model of the format of some block of text, like TeX does. In practice, there's usually a big block of CSS with machine-generated names at the beginning of the web page. There's a fundamental disconnect between the CSS model and the Dreamweaver "Properties" box. So Dreamweaver is still inserting I, B and FONT tags.
In layout, Dreamweaver does table-based layout quite well, but DIV/FLOAT/CLEAR layout isn't handled well. Much of this is due to the limitations of the DIV/FLOAT/CLEAR approach. Tables are a 2D grid system, and map well to the drag-the-lines editor in Dreamweaver. DIV/FLOAT/CLEAR doesn't map well to a visual layout editor.
The end result is a mess. And HTML 5 doesn't help.
This is lame. It's neither insightful nor funny.
What we need is not used-car salesmen with delusions of grandeur. What we need is better truth-in-advertising regulation. Like this:
It's discouraging reading this. Especially since I knew some of the Cyc people back in the 1980s, when they were pursuing the same idea. They're still at it. You can even train their system if you like. But after twenty years of their claiming "Strong AI, Real Soon Now", it's probably not happening.
I went through Stanford CS back when it was just becoming clear that "expert systems" were really rather dumb and weren't going to get smarter. Most of the AI faculty was in denial about that. Very discouraging. The "AI Winter" followed; all the startups went bust, most of the research projects ended, and there was a big empty room of cubicles labeled "Knowledge Systems Laboratory" on the second floor of the Gates Building. I still wonder what happened to the people who got degrees in "Knowledge Engineering". "Do you want fries with that?"
MIT went into a phase where Rod Brooks took over the AI Lab and put everybody on little dumb robots, at roughly the Lego Mindstorms level. Minsky bitched that all the students were soldering instead of learning theory. After a decade or so, it became clear that reactive robot AI could get you to insect level, but no further. Brooks went into the floor-cleaning business (Roomba, Scooba, Dirt Dog, etc.) with the technology, with some success.
Then came the DARPA Grand Challenge. Dr. Tony Tether, the head of DARPA, decided that AI robotics needed a serious kick in the butt. That's what the DARPA Grand Challenge was really all about. It was made clear to the universities receiving DARPA money that if they didn't do well in that game, the money supply would be turned off. It worked. Levels of effort not before seen on a single AI project produced some good results. Stanford had to replace many of the old faculty, but that worked out well in the end.
This is, at last, encouraging. The top-down strong AI problem was just too hard. Insect-level AI, with no world model, was too dumb. But robot vehicle AI, with world models updated by sensors, is now real. So there's progress. The robot vehicle problem is nice because it's so unforgiving. The thing actually has to work; you can't hand-wave around the problems.
The classic bit of hubris in AI, by the way, is to have a good idea and then think it's generally applicable. AI has been through this too many times - the General Problem Solver, inference by theorem proving, neural nets, expert systems, neural nets again, and behavior-based AI. Each of those ideas has a ceiling which has been reached.
It's possible to get too deep into some of these ideas. The people there are brilliant, but narrow, and the culture supports this. MIT has "Nerd Pride" buttons. As someone recruiting me for the Media Lab once said, "There are fewer distractions out here" (It was sleeting.) It sounds like that's what happened to these two young people.
"Monster appears in deserted area" has been done many times. My favorite is "The Navy vs. The Night Monsters". Monsters appear on tropical island. After various adventures, people on island get on radio and call in an air strike. Jet fighters are dispatched, blow up monsters. Final scene shows black smoke rising out of jungle. No more monsters.
The 1950s were a very confident period in the US. After winning WWII, no conceivable monster or monsters looked like a serious threat. '50s SF reflects that.
No, crawling the web isn't a map/reduce type problem. It's a large number of long-running processes feeding a database-like engine.
Map/reduce is for batch-like jobs. Long-running systems with intercommunication have to be organized differently.
At least they handled overflow right. I'm impressed. If it wrapped around to zero, or went negative, some small number of users would be screaming.
Back in the 1980s, the number of ticker symbols for stocks and funds passed 32767, and for a few days, no new companies could get on the exchanges.
I understand what they're getting at. What makes modern SQL-driven databases so useful is that they optimize queries. If you're asking for every entry in A that's also in B, any modern database will check whether it's faster to look up every A in B, every B in A, or do a match where both databases are read through sequentially by the same key. The best choice depends on the database record counts, available indices, and key types and lengths. The database system figures that out; it's not in the SQL query.
So the user says what they want, and the system figures out how to do it. It's "do what I mean" that really works. We don't see enough of that in programming.
Google search itself works much more like a database than a map/reduce system. Think about what has to happen when you search for multiple keywords. That's a join, and joins on big data sets take forever if you don't have the right data structures and an optimizer.
It's amusing to read this. This guy apparently works for Intel's "find ways to use more CPU time" department. Back when I was working on physics engines, I encountered that group.
Actually, the Holy Grail isn't real time ray tracing. It's real time radiosity. Ray-tracing works backwards from the viewpoint; radiosity works outward from the light sources. All the high-end 3D packages have radiosity renderers now. Here's a typical radiosity image. of a kitchen. Radiosity images are great for interiors, and architects now routinely use them for rendering buildings. Lighting effects work like they do in the real world. In a radiosity renderer, you don't have to add phony light sources to make up for the lack of diffuse lighting.
There's a subtle effect that appears in radiosity images but not ray-traced images. Look at the kitchen image and look for an inside corner. Notice the dark band at the inside corner. Look at an inside corner in the real world and you'll see that, too. Neither ray-tracing nor traditional rendering produces that effect, and it's a cue the human vision system uses to resolve corners. The dark band appears as the light bounces back and forth between the two corners, with more light absorbed on each bounce. Radiosity rendering is iterative; you render the image with the starting light sources, then re-render with each illuminated surface as a light source. Each rendering cycle improves the image, until, somewhere around 5 to 50 cycles, the bounced light has mostly been absorbed.
There are ways to precompute light maps from radiosity, then render in real time with an ordinary renderer, and those yield better-looking images of diffuse surfaces than ray-tracing would. Some games already do this. There's a demo of true real-time radiosity, but it doesn't have the "dark band in corners" effect, so it's not doing very many light bounces. Geometrics has a commercial real-time game rendering system.
Ray-tracing can get you "ooh, shiny thing", but radiosity can get to "is that real?"
The thing NASA does best is PR. Of which they do too much. Their PR budget should be cut back, and NASA's funding for non-flight projects should be shifted to NSF.
The headline is wrong. Microsoft has not openly released the specs. They've said they are going to do so. Not the same thing.
Remember when they released some specifications in copy-protected Office documents?
This isn't good. Sun has a tendency to bury a basically good technology under a ton of half-debugged crap. Watch for "you have to have Java xxx installed to run MySQL".