It's an interesting machine. It's a shared-memory multiprocessor without cache coherency. So one way to use it is to allocate disjoint memory to each CPU and run it as a cluster. As the article points out, that is "uninteresting", but at least it's something that's known to work.
Doing something fancier requires a new OS, one that manages clusters, not individual machines. One of the major hypervisors, like Xen, might be a good base for that. Xen already knows how to manage a large number of virtual machines. Managing a large number of real machines with semi-shared memory isn't that big a leap. But that just manages the thing as a cluster. It doesn't exploit the intercommunication.
Intel calls this "A Platform for Software Innovation". What that means is "we have no clue how to program this thing effectively. Maybe academia can figure it out". The last time they tried that, the result was the Itanium.
Historically, there have been far too many supercomputer architectures roughly like this, and they've all been duds. The NCube Hypercube, the Transputer, and the BBN Butterfly come to mind. The Cell machines almost fall into this category. There's no problem building the hardware. It's just not very useful, really tough to program, and the software is too closely tied to a very specific hardware architecture.
Shared-memory multiprocessors with with cache coherency have already reached 256 CPUs. You can even run Windows Server or Linux on them. The headaches of dealing with non-cache-coherent memory may not be worth it.
"The cyber security professionals that we are creating today have to make security invisible to the end user. "They have to make it inherent in the out-of-the-box product that you buy and the only way to do that is for us all to work together, industry, government and academia. We need to be partnering on this."
All this crap about "user awareness" is a dead end. It takes too much attention. The mess underneath needs to be fixed. It has to be automatic. (And don't claim that's impossible unless you've read up on SE Linux and NSA's work on secure systems._
The last high-level US Government professional to publicly point this out was Amit Yoran at Homeland Security. He named Microsoft as the problem. He was canned and replaced with a lobbyist.
This is a huge issue for Google. But not because of Google's operations in Ireland. Google's whole tax-avoidance strategy, which gets Google's tax rate down to 2.4% (!), is based on
a tax strategy which exploits Irish law:
Google Inc. cut its taxes by $3.1 billion in the last three years using a technique that moves most of its foreign profits through Ireland and the Netherlands to Bermuda.
Google's income shifting -- involving strategies known to lawyers as the "Double Irish" and the "Dutch Sandwich" -- helped reduce its overseas tax rate to 2.4 percent, the lowest of the top five U.S. technology companies by market capitalization, according to regulatory filings in six countries.
"It's remarkable that Google's effective rate is that low," said Martin A. Sullivan, a tax economist who formerly worked for the U.S. Treasury Department. "We know this company operates throughout the world mostly in high-tax countries where the average corporate rate is well over 20 percent."
The Bloomberg article describes how this works. Google "licenses its advertising technology" to "Google Ireland Holdings", which owns "Google Ireland Limited". That unit sells 88% of Google's $12.5 billion in non-US advertising. Google Ireland Limited then pays royalties to Google Netherlands Holdings B.V. in Amsterdam (which, according to Bloomberg, is a dummy company with no employees), to get the benefit of a tax break for royalties paid between European Union countries. Then Google Netherlands Holdings B.V. pays royalties to Google Ireland Holdings (headquartered in Bermuda) $5.4 billion in "royalties". "You accumulate profits within Ireland, but then you get them out of the country relatively easily. And you do it by using Bermuda." After all that, the tax liability has been laundered out of existence.
That's why Google is concerned about changes in Ireland's tax laws.
I think when robots are better at my job than I could be, I would start to question the meaning of my life.
In the 1980s, there was an article in Chess Life: "Computer Chess - It's Getting Serious". This was when computers started playing chess well enough that grandmasters had to take them seriously. One strong player wrote "I'm starting to feel like John Henry against the steam hammer". Now it's happened. Any good desktop machine can be loaded up with software that plays at world champion level for about $125. (If you haven't been on the cover of Chess Life, a laptop will be enough to trounce you.) People are still playing chess.
Work, though, is another matter. What's happening is the hollowing out of the middle class. There are more crap jobs that pay minimum wage, but fewer ones that pay more. Manufacturing used to pay well; now it pays slightly above minimum wage, if that. That's because the machines are doing the thinking. The workers are just robot hands with minimal skills.
Here's a very clear example of that - The Kiva robotic order fulfillment system. Watch that video. Hundreds of cooperating mobile robots. All the thinking and planning is done by the computers. The workers just take things out of one tray and put them in a box. The computers even control a laser pointer to point to the object they're supposed to pick. Then a bar-code scanner checks that they did it right. "Requires little or no operator training". Zero opportunity for advancement.
First, the "comma" thing strongly affects Google Suggest, which drives Google Instant. It also affects Google Web Search, but not as strongly. Google Suggest, which comes up with those alternatives for Instant, isn't driven by Google PageRank; it's driven by Google Trends. Or rather, it used to be; it's not as strongly trend-driven as it was a few months ago. That's really a side issue.
Then there are the special-purpose subengines - stocks, health, celebrities, weather, sports, travel, etc. That was actually a Yahoo innovation. Yahoo introduced that in early 2008, with about fifty subengines, and for six months, their search was more on topic than Google's. Few noticed. (I found out about it at a talk by a Yahoo VP.) Then Google copied that idea, and now every major search engine has it. Some of the subengines won't fire with a trailing comma present. The subengines are what the article author is talking about as "hard-coded bias".
Subengines have been around since 2008. What's changing is that some of them now actually sell something. The "weather" and "stocks" subengines don't try to sell anything. The "travel" subengine is different. Try "flight from london to new york". Google has partners ready to sell you tickets. There's a "products" subengine. "dvd player" gets Google results for brands, stores, and types, directing you to Google partners. For neither travel nor products are these entries identified as advertisements.
Here's the actual benchmark used for Top500: "HPL - A Portable Implementation of the High-Performance Linpack Benchmark for Distributed-Memory Computers". It solves linear equations spread across a cluster. The clustered machines have to communicate at a high rate, using MPI 1.1 message passing, to run this program. See this discussion of how the algorithm is parallelized. You can't run this on a set of machines that don't talk much, like "Folding@home" or like cryptanalysis problems.
Linpack is a reasonable approximation of computational fluid dynamics and structural analysis performance. Those are problems that are broken up into cells, with communication between machines about what's happening at the cell boundaries. Those are also the problems for which governments spend money on supercomputers. (The private market for supercomputers is very small.)
So, quit whining. China built the biggest one. Why not? They have more cash right now.
Tesla is doing the battery pack (Li metal-oxide, 30KWh or so), power electronics, and motor. Range will be about 100 "real world" miles, maybe more if they can squeeze in more batteries.
It's not a very exciting use of non-volatile memory. It makes sense, though, to package non-volatile devices for vertical slots like DRAM, and have motherboards that have slots for them. But not DIMM slots - something that actually carries the drive data. The thing announced in the article still needs a drive cable; all it gets from the DIMM slot is power. This looks like an interim product until server motherboards go to that form factor and eliminate drive bays.
The near future for server farms probably looks like that - onboard non-volatile storage in some kind of vertical slot, with rotating disks elsewhere in a storage array.
One of the challenges in computing is to figure out what to do with non-volatile memory besides pretending it is a "disk". Today, CPUs and operating systems understand two kinds of storage - "RAM" and "disk". Design has been locked into that model for decades. Nobody really knows what to do with something that has a 35us access time and no variable latency. Going through the operating system's file system drivers runs up the latency, but making big devices accessible as memory makes them too vulnerable. Some kind of intermediate form of access is needed. A tuple store? A database implemented in an FPGA? Something like that might make sense.
The problem with "boosters" is that they're just amplifiers. They're not players in the cell phone RF protocol system.
Everything that talks in the cell phone bands is supposed to be part of a system that has RF power level control and talks to the
cell phone control station. That's what keeps the transmitters from jamming each other. Adding a dumb transmitter isn't helpful. The right answer would be a "femtocell" unit which connects to an external antenna and connected to the cellular network, and is itself a proper player in the RF protocol.
It would be OK to have a booster if the problem was that you're in a remote location and just need some antenna height to get out. (I'm in such a situation; I'm in a semi-rural area and there's a hill between my house and the nearest cell tower.) What's not OK is installing a booster in Manhattan, where you can't get through because the bands are cluttered, not empty. More RF signal strength just raises the noise floor and cuts system bandwidth. In a crowded area, what's needed is another wired path into the network, not more RF power.
A cell phone that could seamlessly transition from a cell phone network to VoIP over WiFi would be consistent with the system design. There ought to be an Android app for that.
It's hard to think of young ones. Mark Zuckerberg, maybe.
Dean Kamen (the Segway) would be cool to 3rd graders, but he's too old. Same for Mark Raibert (BigDog). Burt Rutan (Scaled Composites, aircraft and spaceplane designer) is way too old. I know some young people (one is under 20) doing cool stuff in the electric car area, but they're not well known yet.
Historically, there have been many inventors who became famous in their 20s, from Edison onward. But I'm having a hard time finding modern examples.
The NSF used to have "Presidential Young Investigator Awards", but after five revisions of that program it's smaller and watered down.
That's not a reaction to CGI, it's a reaction to MTV. Music videos pioneered the quick-cut style of filmmaking. MTV had a big chunk of content at about one cut per second, which was an innovation at the time. That moved into TV production, partly as a way to pick up the pace, and partly as a way to get show length down and commercial time up. Then films started following that trend. By the last James Bond film, "Quantum of Solace", the cut rate had reached the point that action scenes were a bunch of blurry clips. There's a database of average shot length in films; "Quantum of Solace" comes in at an average shot length of 1.5 seconds. This is close to the record for big-budget films.
Long tracking shots are usually a gimmick. "The Player" has an 8-minute long take, but it's a visual joke, and even references long takes. Very few directors use long takes well. "The West Wing" was famous for long tracking shots which advanced the plot effectively. That's rare.
To the extent that CGI has anything to do with this, it's the fact that action-heavy movies are assembled like cartoons. Traditionally, film directors came from the theater. Production started with a script and a group of actors, sitting around a table and doing a reading. Cartoons, on the other hand, started with a storyboard, a real board filled with rows of cards with sketches. Dialogue was made to fit the action.
Effects-heavy movies require major preplanning. (A Star Wars movie is "three years of pre-production, six months of principal photography, three years of post-production", says one of the participants.) Bringing all the pieces together is a huge logistic job, and improvisation runs the costs through the roof. So directors who get it right on the storyboard, check it out with pre-visualization, and build the movie as designed are favored in Hollywood. I know one successful live-action director who came from stop-motion animation, the most pre-planned of all forms.
This style of production favors short shots, which are assembled in post-production. Action scenes are assembled one bit at a time, pacing can be adjusted in post, and dialogue is re-added using automated dialogue replacement. But that only drives shot lengths down to the 3-5 second range. Below that, it's forced pacing.
The free online "white pages" services have usually obtained their data by scanning phone books. Where will they get their data now?
Since Feist vs. Rural Telephone, it's been settled law in the US that the listings in telephone directories are not copyrightable. There's no originality. This created the third-party directory industry. But for online directories, there are EULAs and rate limiting on queries. There's no way to do a bulk download. "Whitepages.com" has these terms: "Among other limitations, you may not:... compile the Results Data in a database and store such data for any future use... publish, transit, distribute, or resell any Results Data." AnyWho (run by AT&T) has the terms: "You agree that you will not use the Service or the information obtained through the Service... for incorporation into a commercial product or service... to download directory listings or other information by using any type of automated means...".
So another data source that used to be open is now closed.
Vista and 7's driver signing requirement is mainly for DRM purposes.
No, the driver signing requirement is for quality control purposes. 60% of Windows crashes used to be driver-related. Now, Microsoft actually requires a proof of correctness, using their Static Driver Verifier, before a driver is signed. The prover tries to determine that the driver can't call a driver API wrong and is free of pointer errors. The goal is to eliminate damage to the rest of the kernel, not check whether the device itself will work. The prover is quite good - 97% of the time it either reaches a successful termination or reports a counterexample - a test case that will break the driver. 3% of the time, the analysis takes too long, the prover gives up, and the code has to be simplified or cleaned up.
From the pictures, it looks like we're seeing a set of images where the automated part of the scanner picked up something interesting. Big wristwatches and belt buckles are marked for attention. These images were probably captures for that reason.
With that unit, the image resolution is so low that it shouldn't upset anyone.
I don't see why people are so upset about this. You know you're being scanned. What's the big deal?
To survive on Mars for an extended period, far more stuff would have to be sent than was required for a round trip.
It's worth remembering the Biosphere II debacle. They couldn't make a closed environment work even with huge domes, no mass limit, extensive preparation, and a sizable team inside.
Colonizing Mars is a fantasy. The atmosphere has well under 1% of Earth's pressure, and it's mostly CO2. The worst places on the surface of the earth are more livable than any part of Mars. Face it - all the off-Earth real estate in the solar system is awful. Some kind of base is possible, but it would be heavily supported from Earth, much like arctic bases now.
Rumors are just that Apple will offer streaming music. Yawn. This is being hyped as "music in the cloud", whatever that's supposed to mean.
Now a really big announcement would be Apple buying Live Nation. Then Apple would control concerts, venues, ticket sales, band promotion, and many top performers - the parts of the music industry still worth something. The "record labels" and radio, which are in decline, would be cut out.
Just how many (great information emperors) have there been?
Several. William Randoph Hearst (newspapers) and David Sarnoff (RCA, NBC) definitely qualify. Not only did they dominate their respective industries for years, they had the arrogance to go with it. Hearst, of course, actually built a castle. Sarnoff made his people call him "The General". Thomas J. Watson Jr. (IBM) was certainly a "great information emperor", although he wasn't as personally arrogant. He moved IBM into electronic computers and ruled computing for three decades. Today, Rupert Murdoch qualifies.
I'm not too impressed. The Wikipedia admins working on this are named "The Thing That Should Not Be" and "Access Denied". I've never heard of either of them in five years on Wikipedia.
It's amazing how far ahead NSA was technologically in the 1950s and 1960s. Magnetic digital recording was first developed for NSA. They had their own custom supercomputers, mostly built by IBM.
A big chunk of IBM's R&D effort went into machines for NSA. NSA was grinding through Western Union traffic with computers when Western Union itself was still running on paper tape. They had huge tape drives (the "Tractor" system) with an robotic tape library, years before anybody else had technology. They put a lot of effort into cryogenic computing. (Eventually, that worked, but it lost out to ordinary ICs, That technology could be made very fast, and gigahertz clocks were achieved in the early 1960s. But it didn't scale down, because it was partly magnetic, like core memory. Moore's Law didn't help.) NSA did lots of work on RF reception of things nobody thought could be received at long range. They used big dishes and moonbounce to listen in on the USSR, and enormous ground-based antennas for HF.
Also, back then the underlying theory of modern cryptanalysis wasn't publicly known. Friedman's work wasn't known. Before Friedman, cryptanalysis was mostly about counting and guessing. After Friedman, cryptanalysis was about statistical number-crunching. NSA's early years were based mostly on Friedman's work, and he was chief cryptanalyst. The real secret of WWII cryptanalysis was that, with the right theory, you could attack the problem with hardware. The Germans and Japanese were still in the "clever guessing" era of cryptanalysis, while the US was filling up buildings with hardware built by IBM, National Cash Register, and Western Electric. This continued into the NSA era.
By the mid-1980s, though, NSA was falling behind. Too much traffic, somewhat antiquated technology, and no interest from the big computer companies in doing custom one-offs. The 1980s were a frustrating period for military R&D. Up until then, military hardware had been well ahead of civilian technology. When the civilian market became far bigger than the military market, that all changed. Not just in electronics, either. One USAF general complained "My golf clubs have more advanced materials than my airplanes." Today, the military struggles to get the attention of the electronics industry, which doesn't want to make tiny quantities of specialized components.
Then the USSR went down, and the world changed. No need to struggle to find out how many subs the Russians had; you could go and look. On the other hand, all the little wars the superpowers had been keeping under control started to flare up. The Balkans and the Middle East became intelligence targets. The targets were now much smaller. Trying to figure out what a small tribe is up to requires completely different approaches than monitoring a huge country. There are some new books out on how NSA is trying to deal with that.
"Domaining" may be on the way out.
on
The Ascendancy of .co
·
· Score: 2, Interesting
With the October 27th change to Google web search, "domaining" may be on the way out.
Google made huge changes when they merged "Google Places" (which is really Google business search") results into their main web search results. Search for DVD player. There are almost no "organic search results" shown. At the top, there's "Related searches for dvd player - Brands, Stores, Types". There are two "organic" results from Amazon and Best Buy, both Google advertisers. Then a big block of "shopping results" A right side column of ads.
And that's a non-local search. On searches which imply some location ("london hotels" is a good test case), Google displays a map. For a few days, they displayed a big map in the main search area; today it's on the right, above the ads. Between the big ad block at the top, the map at the right, the ads below the the map, and the links in the main search area to the map, only a few organic results are squeezed in.
Google's organic search isn't any better than it used to be at filtering out the bottom-feeders. Down below the fold on "dvd player" search, there's still a result from "bestsoftware4download" (which tries a drive-by install of some.exe). In the "london hotels" search, there are a few junk entries. Most of the stuff visible on the first screen isn't organic search results, though. This makes "domaining" futile.
Google is still fooling around with their layout after their big change, and it hasn't settled yet. (Also, Google's layout changes if you're logged into Google and allow "personalization". The results mentioned above are not "personalized".) The trend, though, is clear. The primary results for a search with commercial intent now come from Google advertisers. Google is pushing advertisers to buy ads directly from Google, not from the "bottom feeders".
So buying up large numbers of ".co" domains may be futile. I expect we'll see many junk domains in ".com" expiring, with nobody picking them up.
That's kind of lame. It's just a one-paragraph summary of the game.
A more promising approach would be to start with a play by play summary. Football play-by-plays look like this:
1st and 10 at ATL 20 (Shotgun) M.Ryan pass short right to T.Gonzalez to ATL 23 for 3 yards (J.Johnson).
2nd and 7 at ATL 23 M.Ryan sacked at ATL 13 for -10 yards (T.Suggs).
3rd and 17 at ATL 13 (Shotgun) M.Ryan pass deep right to R.White pushed ob at ATL 41 for 28 yards (E.Reed).
1st and 10 at ATL 41 (Shotgun) M.Ryan pass short right to R.White pushed ob at ATL 46 for 5 yards (D.Landry).
It's clearly possible to turn that into a sports announcer yelling at you. After all, an engine for that is built into EA Madden NFL. With a sports statistics database, you can throw in stuff like "This is Reeds's biggest gain so far this season".
The same would have happened in Indonesia, Saudi Arabia, Syria, Turkey etc.
Turkey, no; Turkey's government is officially secular. Indonesia, quite possibly, depending on the province. Syria, probably, although it matters whether the person saying it is Islamic, and there are multiple court systems. Saudi Arabia, definitely yes.
I agree with SORBS on this. If you run email through a provider which allows any form of "bulk email", "opt-in" or otherwise, once in a while, some spam will come out of their system. That apparently happened here, and SORBS, correctly, blocked them. That's the risk you take if you sign up with an email provider that isn't sufficiently aggressive about spam.
Notice how fast MessageLabs cut off the spam source, and how much effort they put into fixing the problem. Without punishment from an external checker on their behavior, the spam probably would have continued for weeks, if not indefinitely. Because of SORBS' action, the problem is already fixed.
SORBS has, at various times, blocked Google outgoing mail and Postini outgoing mail. As a result, Google and Postini (now owned by Google) had to become much more effective at knocking off spammers. That's what you want to happen.
If security is to be effective, incidents like this have to occur. There will be some collateral damage.
Slashdot Mirror
It's an interesting machine. It's a shared-memory multiprocessor without cache coherency. So one way to use it is to allocate disjoint memory to each CPU and run it as a cluster. As the article points out, that is "uninteresting", but at least it's something that's known to work.
Doing something fancier requires a new OS, one that manages clusters, not individual machines. One of the major hypervisors, like Xen, might be a good base for that. Xen already knows how to manage a large number of virtual machines. Managing a large number of real machines with semi-shared memory isn't that big a leap. But that just manages the thing as a cluster. It doesn't exploit the intercommunication.
Intel calls this "A Platform for Software Innovation". What that means is "we have no clue how to program this thing effectively. Maybe academia can figure it out". The last time they tried that, the result was the Itanium.
Historically, there have been far too many supercomputer architectures roughly like this, and they've all been duds. The NCube Hypercube, the Transputer, and the BBN Butterfly come to mind. The Cell machines almost fall into this category. There's no problem building the hardware. It's just not very useful, really tough to program, and the software is too closely tied to a very specific hardware architecture.
Shared-memory multiprocessors with with cache coherency have already reached 256 CPUs. You can even run Windows Server or Linux on them. The headaches of dealing with non-cache-coherent memory may not be worth it.
This guy gets it:
"The cyber security professionals that we are creating today have to make security invisible to the end user. "They have to make it inherent in the out-of-the-box product that you buy and the only way to do that is for us all to work together, industry, government and academia. We need to be partnering on this."
All this crap about "user awareness" is a dead end. It takes too much attention. The mess underneath needs to be fixed. It has to be automatic. (And don't claim that's impossible unless you've read up on SE Linux and NSA's work on secure systems._
The last high-level US Government professional to publicly point this out was Amit Yoran at Homeland Security. He named Microsoft as the problem. He was canned and replaced with a lobbyist.
This is a huge issue for Google. But not because of Google's operations in Ireland. Google's whole tax-avoidance strategy, which gets Google's tax rate down to 2.4% (!), is based on a tax strategy which exploits Irish law:
Google Inc. cut its taxes by $3.1 billion in the last three years using a technique that moves most of its foreign profits through Ireland and the Netherlands to Bermuda.
Google's income shifting -- involving strategies known to lawyers as the "Double Irish" and the "Dutch Sandwich" -- helped reduce its overseas tax rate to 2.4 percent, the lowest of the top five U.S. technology companies by market capitalization, according to regulatory filings in six countries.
"It's remarkable that Google's effective rate is that low," said Martin A. Sullivan, a tax economist who formerly worked for the U.S. Treasury Department. "We know this company operates throughout the world mostly in high-tax countries where the average corporate rate is well over 20 percent."
The Bloomberg article describes how this works. Google "licenses its advertising technology" to "Google Ireland Holdings", which owns "Google Ireland Limited". That unit sells 88% of Google's $12.5 billion in non-US advertising. Google Ireland Limited then pays royalties to Google Netherlands Holdings B.V. in Amsterdam (which, according to Bloomberg, is a dummy company with no employees), to get the benefit of a tax break for royalties paid between European Union countries. Then Google Netherlands Holdings B.V. pays royalties to Google Ireland Holdings (headquartered in Bermuda) $5.4 billion in "royalties". "You accumulate profits within Ireland, but then you get them out of the country relatively easily. And you do it by using Bermuda." After all that, the tax liability has been laundered out of existence.
That's why Google is concerned about changes in Ireland's tax laws.
I think when robots are better at my job than I could be, I would start to question the meaning of my life.
In the 1980s, there was an article in Chess Life: "Computer Chess - It's Getting Serious". This was when computers started playing chess well enough that grandmasters had to take them seriously. One strong player wrote "I'm starting to feel like John Henry against the steam hammer". Now it's happened. Any good desktop machine can be loaded up with software that plays at world champion level for about $125. (If you haven't been on the cover of Chess Life, a laptop will be enough to trounce you.) People are still playing chess.
Work, though, is another matter. What's happening is the hollowing out of the middle class. There are more crap jobs that pay minimum wage, but fewer ones that pay more. Manufacturing used to pay well; now it pays slightly above minimum wage, if that. That's because the machines are doing the thinking. The workers are just robot hands with minimal skills.
Here's a very clear example of that - The Kiva robotic order fulfillment system. Watch that video. Hundreds of cooperating mobile robots. All the thinking and planning is done by the computers. The workers just take things out of one tray and put them in a box. The computers even control a laser pointer to point to the object they're supposed to pick. Then a bar-code scanner checks that they did it right. "Requires little or no operator training". Zero opportunity for advancement.
There's a lot going on here.
First, the "comma" thing strongly affects Google Suggest, which drives Google Instant. It also affects Google Web Search, but not as strongly. Google Suggest, which comes up with those alternatives for Instant, isn't driven by Google PageRank; it's driven by Google Trends. Or rather, it used to be; it's not as strongly trend-driven as it was a few months ago. That's really a side issue.
Then there are the special-purpose subengines - stocks, health, celebrities, weather, sports, travel, etc. That was actually a Yahoo innovation. Yahoo introduced that in early 2008, with about fifty subengines, and for six months, their search was more on topic than Google's. Few noticed. (I found out about it at a talk by a Yahoo VP.) Then Google copied that idea, and now every major search engine has it. Some of the subengines won't fire with a trailing comma present. The subengines are what the article author is talking about as "hard-coded bias".
Subengines have been around since 2008. What's changing is that some of them now actually sell something. The "weather" and "stocks" subengines don't try to sell anything. The "travel" subengine is different. Try "flight from london to new york". Google has partners ready to sell you tickets. There's a "products" subengine. "dvd player" gets Google results for brands, stores, and types, directing you to Google partners. For neither travel nor products are these entries identified as advertisements.
This is where Google is pushing the line between search results and paid ads. This previously got them into trouble with the Federal Trade Commission back in 2002. Now it's more subtle, but it's back.
Yes, noticed that.
Here's the actual benchmark used for Top500: "HPL - A Portable Implementation of the High-Performance Linpack Benchmark for Distributed-Memory Computers". It solves linear equations spread across a cluster. The clustered machines have to communicate at a high rate, using MPI 1.1 message passing, to run this program. See this discussion of how the algorithm is parallelized. You can't run this on a set of machines that don't talk much, like "Folding@home" or like cryptanalysis problems.
Linpack is a reasonable approximation of computational fluid dynamics and structural analysis performance. Those are problems that are broken up into cells, with communication between machines about what's happening at the cell boundaries. Those are also the problems for which governments spend money on supercomputers. (The private market for supercomputers is very small.)
So, quit whining. China built the biggest one. Why not? They have more cash right now.
Tesla is doing the battery pack (Li metal-oxide, 30KWh or so), power electronics, and motor. Range will be about 100 "real world" miles, maybe more if they can squeeze in more batteries.
The RAV4 is much bigger than it used to be. Compare the original RAV4 and today's oversized version.
Fifteen years of battery progress later, electrics are almost good enough.
It's not a very exciting use of non-volatile memory. It makes sense, though, to package non-volatile devices for vertical slots like DRAM, and have motherboards that have slots for them. But not DIMM slots - something that actually carries the drive data. The thing announced in the article still needs a drive cable; all it gets from the DIMM slot is power. This looks like an interim product until server motherboards go to that form factor and eliminate drive bays. The near future for server farms probably looks like that - onboard non-volatile storage in some kind of vertical slot, with rotating disks elsewhere in a storage array.
One of the challenges in computing is to figure out what to do with non-volatile memory besides pretending it is a "disk". Today, CPUs and operating systems understand two kinds of storage - "RAM" and "disk". Design has been locked into that model for decades. Nobody really knows what to do with something that has a 35us access time and no variable latency. Going through the operating system's file system drivers runs up the latency, but making big devices accessible as memory makes them too vulnerable. Some kind of intermediate form of access is needed. A tuple store? A database implemented in an FPGA? Something like that might make sense.
The problem with "boosters" is that they're just amplifiers. They're not players in the cell phone RF protocol system.
Everything that talks in the cell phone bands is supposed to be part of a system that has RF power level control and talks to the cell phone control station. That's what keeps the transmitters from jamming each other. Adding a dumb transmitter isn't helpful. The right answer would be a "femtocell" unit which connects to an external antenna and connected to the cellular network, and is itself a proper player in the RF protocol.
It would be OK to have a booster if the problem was that you're in a remote location and just need some antenna height to get out. (I'm in such a situation; I'm in a semi-rural area and there's a hill between my house and the nearest cell tower.) What's not OK is installing a booster in Manhattan, where you can't get through because the bands are cluttered, not empty. More RF signal strength just raises the noise floor and cuts system bandwidth. In a crowded area, what's needed is another wired path into the network, not more RF power.
A cell phone that could seamlessly transition from a cell phone network to VoIP over WiFi would be consistent with the system design. There ought to be an Android app for that.
It's hard to think of young ones. Mark Zuckerberg, maybe.
Dean Kamen (the Segway) would be cool to 3rd graders, but he's too old. Same for Mark Raibert (BigDog). Burt Rutan (Scaled Composites, aircraft and spaceplane designer) is way too old. I know some young people (one is under 20) doing cool stuff in the electric car area, but they're not well known yet.
Historically, there have been many inventors who became famous in their 20s, from Edison onward. But I'm having a hard time finding modern examples.
The NSF used to have "Presidential Young Investigator Awards", but after five revisions of that program it's smaller and watered down.
That's not a reaction to CGI, it's a reaction to MTV. Music videos pioneered the quick-cut style of filmmaking. MTV had a big chunk of content at about one cut per second, which was an innovation at the time. That moved into TV production, partly as a way to pick up the pace, and partly as a way to get show length down and commercial time up. Then films started following that trend. By the last James Bond film, "Quantum of Solace", the cut rate had reached the point that action scenes were a bunch of blurry clips. There's a database of average shot length in films; "Quantum of Solace" comes in at an average shot length of 1.5 seconds. This is close to the record for big-budget films.
Long tracking shots are usually a gimmick. "The Player" has an 8-minute long take, but it's a visual joke, and even references long takes. Very few directors use long takes well. "The West Wing" was famous for long tracking shots which advanced the plot effectively. That's rare.
To the extent that CGI has anything to do with this, it's the fact that action-heavy movies are assembled like cartoons. Traditionally, film directors came from the theater. Production started with a script and a group of actors, sitting around a table and doing a reading. Cartoons, on the other hand, started with a storyboard, a real board filled with rows of cards with sketches. Dialogue was made to fit the action.
Effects-heavy movies require major preplanning. (A Star Wars movie is "three years of pre-production, six months of principal photography, three years of post-production", says one of the participants.) Bringing all the pieces together is a huge logistic job, and improvisation runs the costs through the roof. So directors who get it right on the storyboard, check it out with pre-visualization, and build the movie as designed are favored in Hollywood. I know one successful live-action director who came from stop-motion animation, the most pre-planned of all forms.
This style of production favors short shots, which are assembled in post-production. Action scenes are assembled one bit at a time, pacing can be adjusted in post, and dialogue is re-added using automated dialogue replacement. But that only drives shot lengths down to the 3-5 second range. Below that, it's forced pacing.
The free online "white pages" services have usually obtained their data by scanning phone books. Where will they get their data now?
Since Feist vs. Rural Telephone, it's been settled law in the US that the listings in telephone directories are not copyrightable. There's no originality. This created the third-party directory industry. But for online directories, there are EULAs and rate limiting on queries. There's no way to do a bulk download. "Whitepages.com" has these terms: "Among other limitations, you may not: ... compile the Results Data in a database and store such data for any future use ... publish, transit, distribute, or resell any Results Data." AnyWho (run by AT&T) has the terms: "You agree that you will not use the Service or the information obtained through the Service ... for incorporation into a commercial product or service ... to download directory listings or other information by using any type of automated means ...".
So another data source that used to be open is now closed.
Vista and 7's driver signing requirement is mainly for DRM purposes.
No, the driver signing requirement is for quality control purposes. 60% of Windows crashes used to be driver-related. Now, Microsoft actually requires a proof of correctness, using their Static Driver Verifier, before a driver is signed. The prover tries to determine that the driver can't call a driver API wrong and is free of pointer errors. The goal is to eliminate damage to the rest of the kernel, not check whether the device itself will work. The prover is quite good - 97% of the time it either reaches a successful termination or reports a counterexample - a test case that will break the driver. 3% of the time, the analysis takes too long, the prover gives up, and the code has to be simplified or cleaned up.
Since the announcement, Apple stock has dropped 2%. The markets were unimpressed.
From the pictures, it looks like we're seeing a set of images where the automated part of the scanner picked up something interesting. Big wristwatches and belt buckles are marked for attention. These images were probably captures for that reason.
With that unit, the image resolution is so low that it shouldn't upset anyone.
I don't see why people are so upset about this. You know you're being scanned. What's the big deal?
To survive on Mars for an extended period, far more stuff would have to be sent than was required for a round trip. It's worth remembering the Biosphere II debacle. They couldn't make a closed environment work even with huge domes, no mass limit, extensive preparation, and a sizable team inside.
Colonizing Mars is a fantasy. The atmosphere has well under 1% of Earth's pressure, and it's mostly CO2. The worst places on the surface of the earth are more livable than any part of Mars. Face it - all the off-Earth real estate in the solar system is awful. Some kind of base is possible, but it would be heavily supported from Earth, much like arctic bases now.
Rumors are just that Apple will offer streaming music. Yawn. This is being hyped as "music in the cloud", whatever that's supposed to mean.
Now a really big announcement would be Apple buying Live Nation. Then Apple would control concerts, venues, ticket sales, band promotion, and many top performers - the parts of the music industry still worth something. The "record labels" and radio, which are in decline, would be cut out.
Just how many (great information emperors) have there been?
Several. William Randoph Hearst (newspapers) and David Sarnoff (RCA, NBC) definitely qualify. Not only did they dominate their respective industries for years, they had the arrogance to go with it. Hearst, of course, actually built a castle. Sarnoff made his people call him "The General". Thomas J. Watson Jr. (IBM) was certainly a "great information emperor", although he wasn't as personally arrogant. He moved IBM into electronic computers and ruled computing for three decades. Today, Rupert Murdoch qualifies.
I'm not too impressed. The Wikipedia admins working on this are named "The Thing That Should Not Be" and "Access Denied". I've never heard of either of them in five years on Wikipedia.
The problem will come if OpenOffice starts to demand Oracle's proprietary version of Java, and then Oracle starts to tighten the screws on Java.
I'm much more worried about MySQL under Oracle's control. Oracle has every reason to make MySQL worse, especially the versions that scale up.
It's amazing how far ahead NSA was technologically in the 1950s and 1960s. Magnetic digital recording was first developed for NSA. They had their own custom supercomputers, mostly built by IBM. A big chunk of IBM's R&D effort went into machines for NSA. NSA was grinding through Western Union traffic with computers when Western Union itself was still running on paper tape. They had huge tape drives (the "Tractor" system) with an robotic tape library, years before anybody else had technology. They put a lot of effort into cryogenic computing. (Eventually, that worked, but it lost out to ordinary ICs, That technology could be made very fast, and gigahertz clocks were achieved in the early 1960s. But it didn't scale down, because it was partly magnetic, like core memory. Moore's Law didn't help.) NSA did lots of work on RF reception of things nobody thought could be received at long range. They used big dishes and moonbounce to listen in on the USSR, and enormous ground-based antennas for HF.
Also, back then the underlying theory of modern cryptanalysis wasn't publicly known. Friedman's work wasn't known. Before Friedman, cryptanalysis was mostly about counting and guessing. After Friedman, cryptanalysis was about statistical number-crunching. NSA's early years were based mostly on Friedman's work, and he was chief cryptanalyst. The real secret of WWII cryptanalysis was that, with the right theory, you could attack the problem with hardware. The Germans and Japanese were still in the "clever guessing" era of cryptanalysis, while the US was filling up buildings with hardware built by IBM, National Cash Register, and Western Electric. This continued into the NSA era.
By the mid-1980s, though, NSA was falling behind. Too much traffic, somewhat antiquated technology, and no interest from the big computer companies in doing custom one-offs. The 1980s were a frustrating period for military R&D. Up until then, military hardware had been well ahead of civilian technology. When the civilian market became far bigger than the military market, that all changed. Not just in electronics, either. One USAF general complained "My golf clubs have more advanced materials than my airplanes." Today, the military struggles to get the attention of the electronics industry, which doesn't want to make tiny quantities of specialized components.
Then the USSR went down, and the world changed. No need to struggle to find out how many subs the Russians had; you could go and look. On the other hand, all the little wars the superpowers had been keeping under control started to flare up. The Balkans and the Middle East became intelligence targets. The targets were now much smaller. Trying to figure out what a small tribe is up to requires completely different approaches than monitoring a huge country. There are some new books out on how NSA is trying to deal with that.
With the October 27th change to Google web search, "domaining" may be on the way out.
Google made huge changes when they merged "Google Places" (which is really Google business search") results into their main web search results. Search for DVD player. There are almost no "organic search results" shown. At the top, there's "Related searches for dvd player - Brands, Stores, Types". There are two "organic" results from Amazon and Best Buy, both Google advertisers. Then a big block of "shopping results" A right side column of ads.
And that's a non-local search. On searches which imply some location ("london hotels" is a good test case), Google displays a map. For a few days, they displayed a big map in the main search area; today it's on the right, above the ads. Between the big ad block at the top, the map at the right, the ads below the the map, and the links in the main search area to the map, only a few organic results are squeezed in.
Google's organic search isn't any better than it used to be at filtering out the bottom-feeders. Down below the fold on "dvd player" search, there's still a result from "bestsoftware4download" (which tries a drive-by install of some .exe). In the "london hotels" search, there are a few junk entries. Most of the stuff visible on the first screen isn't organic search results, though. This makes "domaining" futile.
Google is still fooling around with their layout after their big change, and it hasn't settled yet. (Also, Google's layout changes if you're logged into Google and allow "personalization". The results mentioned above are not "personalized".) The trend, though, is clear. The primary results for a search with commercial intent now come from Google advertisers. Google is pushing advertisers to buy ads directly from Google, not from the "bottom feeders".
So buying up large numbers of ".co" domains may be futile. I expect we'll see many junk domains in ".com" expiring, with nobody picking them up.
That's kind of lame. It's just a one-paragraph summary of the game.
A more promising approach would be to start with a play by play summary. Football play-by-plays look like this:
It's clearly possible to turn that into a sports announcer yelling at you. After all, an engine for that is built into EA Madden NFL. With a sports statistics database, you can throw in stuff like "This is Reeds's biggest gain so far this season".
Somebody has probably already done this.
The same would have happened in Indonesia, Saudi Arabia, Syria, Turkey etc.
Turkey, no; Turkey's government is officially secular. Indonesia, quite possibly, depending on the province. Syria, probably, although it matters whether the person saying it is Islamic, and there are multiple court systems. Saudi Arabia, definitely yes.
I agree with SORBS on this. If you run email through a provider which allows any form of "bulk email", "opt-in" or otherwise, once in a while, some spam will come out of their system. That apparently happened here, and SORBS, correctly, blocked them. That's the risk you take if you sign up with an email provider that isn't sufficiently aggressive about spam.
Notice how fast MessageLabs cut off the spam source, and how much effort they put into fixing the problem. Without punishment from an external checker on their behavior, the spam probably would have continued for weeks, if not indefinitely. Because of SORBS' action, the problem is already fixed.
SORBS has, at various times, blocked Google outgoing mail and Postini outgoing mail. As a result, Google and Postini (now owned by Google) had to become much more effective at knocking off spammers. That's what you want to happen.
If security is to be effective, incidents like this have to occur. There will be some collateral damage.