Amusingly, since this is based on Atom, the client still has to poll. It just has to poll fewer sources. The connection between the original source and the "pushsubhub" server really is a "push" connection, but the hub to client connection is not.
Also, the "pushsubhub" caches and redistributes the feeds, which means the feed operator no longer sees their own clients.
They don't seem to have addressed the general RSS problem of "server timestamp/ID changed, but content did not". Some RSS feeds get this right; some don't. Reuters is good, but not perfect. Other sites vary; there's a common problem where the RSS feed is provided from multiple servers on a load balancer, and the servers don't coordinate on timestamps and IDs.
Twitter is awful. An RSS feed from Twitter appears to change on each poll even when the content has not changed.
Actually determining that RSS content really hasn't changed currently requires computing a message digest on the content. If you're going to aggregate RSS feeds, it's probably necessary to do that.
Shockwave already has full 3D capabilities. Here's s reasonably good 3D scene, a haunted house.
Shockwave in current use; both Disney and Dreamworks have 3D promotional games for upcoming movies, and Porsche has a car configurator. It's possible to do a reasonably decent game in Shockwave.
Unfortunately, Shockwave seems to be associated with crap sites full of ad-heavy low-end games that keep trying to download additional plugins.
A big downside of Shockwave is that, unlike Flash, the whole file has to load before it starts. It lacks instant gratification. Did the XML 3D crowd deal with that issue?
Incidentally, this is not the only XML-based 3D system. There's Web3D, which is simply VRML 97 with XML delimiters. VRML itself works quite well today. When it came out in 1997, few people had enough graphics power to run it, and it got a bad reputation. Now, everybody does.
OTOH, black computers are really cool nowadays, where they used to be mostly beige.
It's a cycle. Beige, then black, then silver, then back to beige. This used to be managed by the Color Association of the United States, which issued color "forecasts". The color selection process was once quite organized, because textile companies had long lead times, and needed to know what the "in" colors were going to be months in advance. That was back when the NYC garment district mattered. The CAUS also issued color "forecasts" for products, and the injection-moulded plastic people paid attention to them.
But what seems like a bad deal to me is the concept of extending copyright to 'style'.
The recording (and most of the performing) industry has profited hugely from the provision in US copyright law that allows anyone to make a "cover" of an existing song, paying only a statutory royalty to the songwriter. That's what this technology is automating.
There are going to be branding issues. Whether one can refer to the name of the performer being emulated is a big question. We may end up with something like the markings seen on generic non-prescription drugs - "Compare to the ingredients in Tylenol". It's really about branding, not music.
Some years ago, I was backstage during the preparations for a show by a Big Name Rock Band. I was chatting with some of the promoter's people, who remarked that they had two sets of the stage set and equipment, and there was a crew at the next venue on the tour already beginning setup. I remarked "There are two road companies of Cats. There are two road companies of the Ringling Brothers, Barnum and Bailey Circus. There have been rock groups where every member of the band was replaced over time, and the band kept the same name. This is all about branding and marketing. Why not have two sets of musicians and double your revenue?" The junior people laughed. The senior promoter in the suit looked very thoughtful.
That's a good idea. We do something like that at SiteTruth, where we down-rate commercial sites that don't have a real-world contact address on the site. We're looking at user-visible pages, though, not WHOIS. WHOIS data quality is too low.
I'm all in favor of this sort of thing. But don't drop the messages silently; reject them during the SMTP session if you can, or send a mail bounce if you can't. There's much to be said for having a hard-ass attitude about this, but you have to handle the false positives properly.
Anything that sends mail bounces needs to check SPF records. This makes it possible to stop joe-job mail bounce problems.
(EXIM mailer people: please finish the implementation of SPF checking and advance it from "experimental", so large ISPs can use it.)
Also, quit whining that putting your real name on your WHOIS registration will get you annoying phone calls, threats, or whatever. I've had my real name and contact info on all my web sites and WHOIS information for a decade, and that's just not happening.
Because Nanosolar has supposedly been shipping for three years and there are no visible installations.
Just because they have pictures of a factory interior doesn't mean they have a factory that delivers working product at the claimed cost. It's all too common to see some minor advance in materials science hyped into something becoming real cheap, real soon now. Most of the roll-to-roll processes for making solar cells don't work.
Not all of them, though. Uni-Solar Ovonic really does have a working roll to roll process for making flexible solar cells. You can order their products, they have a dealer network, and they have many installations. Nanosolar has none of that. Just hype.
The earth is naturally slowing down at a rate that makes this sort of thing hardly worth mentioning.
Actually, the slowdown is only about 2ms/century, or about 0.054 microseconds/day. So a 1-microsecond jump in a day should be noticeable. This information is tracked. Here's the raw data from the Earth Rotation Service.
With GPS systems working down to 15cm, changes like this get noticed.
Hey! The Cone of Silence was a required deliverable insisted on by management, okay?
Some of those pictures are renders, and some are real. It turns out that the suspended transparent bubble was deleted before actual construction, along with the oval lighting rings and the curved-glass viewing galleries. Those are renders. The ones with conventional lighting on the ceiling are real, including the fancy podium. In this picture, the PCs, which look like conventional mini-towers, and their cables are in place.
So it's really a big, square mostly-empty room, with a bunch of Barco rear-projection monitors in front, and a fancy podium that cost US$90,000. The architect writes "this is main room in large building of company, which controls all electricity in Moscow. In this room 6 people working 24/7. Working group will not expand, but space is needed for groups of high ranked visitors."
That might actually be a reasonable design, given that problem. I can see the chief operator saying to the architect "Put enough empty space behind us so that when the oligarchs come to visit, they don't get in the way. Don't give them any chairs, or they'll hang around and muck things up. Put a VIP lounge somewhere else in the building, with the booze and the girls. Stick a few screens in there to repeat the big board, so they think they're running things."
At some point, the GPU goes on the CPU chip, and gets faster as a result.
Maybe.
GPUs need enormous bandwidth to memory, and can usefully use several different types of memory with separate data paths. The frame buffer, texture memory, geometry memory, and program memory are all being accessed by different parts of the GPU. Making all that traffic go through the CPU's path to memory, which is already the bottleneck with current CPUs, doesn't help performance.
A single chip solution improves CPU to GPU bandwidth, but that's not usually the worst bottleneck.
What actually determines the solution turns out to be issues like how many pins you can effectively have on a chip.
From Nanosolar's website, it sounds like they've been shipping panels commercially for the last two years, and that they have panel assemblies in both the US and Germany...
Yes, from Nanosolar's web site, it sounds that way. But as of 2009, "not one Solarply cell has been held yet in the hands of a consumer". There are no reports of actual Nanosolar installations. Supposedly they're building big solar panel installations for utility companies. So where are the announcements from those utility companies? Where are the regulatory filings, the planning documents, and the Google Earth pictures? Installing a big solar farm leaves a public record.
Here's an example of just that - the new Moesk control center for Moscow's electric network.
Take a look at the pictures. This looks like a movie set for a Bond movie. The architects got completely out of control here.
Notice the suspended transparent bubble for top management. It looks like it retracts into the ceiling. The lower operator's platform has steeply slanted sides, no railings, and chairs with wheels. The huge room only has eight operator positions.
I'll bet that, within a year or two, the people who actually have to run the grid set up a "field control center" with about twenty people with PCs, cork boards on the walls, 2-way radios for talking to field crews, a conference/map table, and some printers. The real work will be done there. A few people will sit in the big room and answer questions for management.
Here's the actual scientific paper, "Predicted Efficiency of Si Wire Array Solar Cells". That's by the same authors mentioned in the press release. While the thing does trap most of the light hitting it, only a fraction of the energy in that light is converted to electricity. In fact, this thing is currently less efficient than the better commercial solar cells.
From the paper:... simulated photovoltaic efficency of 14.5%.... Conclusion:... "Si wire array solar cells have the potential to reach efficiencies competitive with traditional Si crystalline solar cells."
So, an interesting development, but no big breakthrough. There's a claim that it might be a cheaper way to make solar cells, but everybody who comes up with a new design makes that claim. (Nanosolar comes to mind; their technology is supposed to be cheaper, but so far they've spent half a billion dollars and apparently have only produced sample panels.)
There's a reason relatively little manufacturing is done in the US anymore,
The US, with a third of China's population, still does more manufacturing than China. US manufacturing output is still over twice that of China. US manufacturing employment, though, continues to drop. Manufacturing automation works very well today.
The giant urban slum is a phenomenon of the last 50 years. Fifty years ago, New York City was the biggest city in the world. It had poor people, but most of them had real jobs. New York is now in 13th place (one can argue over how to count), and the top 5 are all third world cities surrounded by slums. This is called "over-urbanization".
There's a step up from the "favelas out to the horizon". It's the "housing project". The better-organized big cities have vast numbers of cheap concrete high-rise apartment blocks,
resembling US housing projects of the 1960s. That's what Tokyo, Beijing, Hong Kong, and Shanghai residential areas look like. It's an efficient way of warehousing a large number of humans. Whether it's a good life depends more on the people than the architecture. (Luxury apartment buildings and housing projects cost about the same to build per room for the basic structure. The luxury building may cost less to run, because vandalism and crime will be lower and the number of children per household will be lower.)
The US is unique in having most of its population in suburbs. Only cheap oil made that possible.
The industrial world is transitioning to twisted-pair Ethernet. From an industrial perspective,
twisted-pair Ethernet is a nice interface. It's balanced drive and twisted pair, so it has very good common-mode noise rejection. (Better than RS-232 and 5V encoder signals, in fact.)
There's full error checking and retransmission, unlike serial lines.
This is important when it attaches to a welding robot, or runs past a circuit breaker that's switching 5KV.
Long cable runs of 10baseT and 100baseT work fine. Bridges and routers are easily available and cheap, even in industrial form. And, of course, you can put many devices on one cable.
From a security perspective, though, this creates the problem that it's too easy to get devices on an Ethernet cable connected to the external Internet. Especially since the support programs for the devices tend to run on Windows. Most of the industrial Ethernet devices have little or no security.
It's surprising that nobody has replicated the V-1 "flying bomb" of WWII. That was the first cruise missile, and wasn't expensive. Carrying 850Kg of explosive, it could take out medium-sized buildings. Its main limitation was poor guidance. The Nazis fired about 8,000 at London, but couldn't hit a target smaller than a big city. The guidance system was a gyro/pendulum/magnetic compass system, and just flew the thing in a straight line until a small propeller/odometer had counted enough turns.
Since it flew straight and level, it was easy to shoot down. Still, "easy to shoot down" meant hundreds of interceptors and hundreds of anti-aircraft guns, and the success rate at shoot-down only reached 75% or so.
With a modern guidance system, one that could hit a target and didn't fly straight and level, these things could be formidable weapons today. If someone launched twenty of them, each programmed to take a different path to the same target, some of them would get through. Nobody has enough interceptors any more to take out an attack like that.
In addition to sending human-controlled robots to the moon, lets send along refineries and factories to produce solar panels.
Yeah, right. Back around 1985, I went to a conference where some AI professors were mouthing off about putting self-replicating factories on the Moon within 20 years. I asked "How soon can you do it in Arizona?" They didn't like that.
The article is bogus, but the problem is real. Computer support systems for investigators are hard to build. The FBI has struggled with this, taking about a decade to deploy their "Field Office Automation" system. They're hard for many of the same reasons medical systems are hard - much of the incoming data is unstructured, and many people enter data relevant to the same case. It's even harder than in the medical world, because links between various individuals and events are important, but unreliable. The "customers" aren't cooperative, they usually don't have unique identifiers, and a sizable fraction of the information is bogus. The security problems are tough to even define - exactly who's allowed to see what is a big issue.
The older law enforcement systems didn't offer much searchability. Unless you had a hard search key, like a driver's license number or a full name, you couldn't retrieve much. Now, everybody expects Google-like searchability, and the older systems just didn't have the machinery for that.
Many of you don't get it. Apple "iPad" isn't a general-purpose computer. It's an entertainment device.
It doesn't need a camera, and it doesn't need a keyboard, because it's a mostly-output device. It's an improved "e-reader", not a "netbook". It's not a "convergence" device, it's a "divergence" device.
It's overpriced. The price will drop, of course. Look at the price history of Apple's music players. The entry level device there is now $125.
It will succeed or fail based on what content it can access. If Apple and News Corp. work out a deal, and you can view all News Corp. content on the thing, it will be a big success.
I expect that a John Williams simulator can't be that far off. Williams is the composer who did the symphony orchestra scores for Jaws, Star Wars, most of Speilberg's works, and other industrial-strength dramatic productions.
As music, his music sucks. Listen to his music without a movie, and it brings to mind the tank commander's motto, "When in doubt, use the main gun." But it carries the production along. With a Williams score, a good production designer, and a big budget, a film can be a success even with a dumb plot and bad acting.
Slashdot Mirror
Amusingly, since this is based on Atom, the client still has to poll. It just has to poll fewer sources. The connection between the original source and the "pushsubhub" server really is a "push" connection, but the hub to client connection is not.
Also, the "pushsubhub" caches and redistributes the feeds, which means the feed operator no longer sees their own clients.
They don't seem to have addressed the general RSS problem of "server timestamp/ID changed, but content did not". Some RSS feeds get this right; some don't. Reuters is good, but not perfect. Other sites vary; there's a common problem where the RSS feed is provided from multiple servers on a load balancer, and the servers don't coordinate on timestamps and IDs. Twitter is awful. An RSS feed from Twitter appears to change on each poll even when the content has not changed.
Actually determining that RSS content really hasn't changed currently requires computing a message digest on the content. If you're going to aggregate RSS feeds, it's probably necessary to do that.
Shockwave already has full 3D capabilities. Here's s reasonably good 3D scene, a haunted house. Shockwave in current use; both Disney and Dreamworks have 3D promotional games for upcoming movies, and Porsche has a car configurator. It's possible to do a reasonably decent game in Shockwave. Unfortunately, Shockwave seems to be associated with crap sites full of ad-heavy low-end games that keep trying to download additional plugins.
A big downside of Shockwave is that, unlike Flash, the whole file has to load before it starts. It lacks instant gratification. Did the XML 3D crowd deal with that issue?
Incidentally, this is not the only XML-based 3D system. There's Web3D, which is simply VRML 97 with XML delimiters. VRML itself works quite well today. When it came out in 1997, few people had enough graphics power to run it, and it got a bad reputation. Now, everybody does.
OTOH, black computers are really cool nowadays, where they used to be mostly beige.
It's a cycle. Beige, then black, then silver, then back to beige. This used to be managed by the Color Association of the United States, which issued color "forecasts". The color selection process was once quite organized, because textile companies had long lead times, and needed to know what the "in" colors were going to be months in advance. That was back when the NYC garment district mattered. The CAUS also issued color "forecasts" for products, and the injection-moulded plastic people paid attention to them.
They changed the color scheme from brown to beige. How exciting.
The small icons are still too cluttered. They're simply smaller versions of the large icons, which never works very well.
But what seems like a bad deal to me is the concept of extending copyright to 'style'.
The recording (and most of the performing) industry has profited hugely from the provision in US copyright law that allows anyone to make a "cover" of an existing song, paying only a statutory royalty to the songwriter. That's what this technology is automating.
There are going to be branding issues. Whether one can refer to the name of the performer being emulated is a big question. We may end up with something like the markings seen on generic non-prescription drugs - "Compare to the ingredients in Tylenol". It's really about branding, not music.
Some years ago, I was backstage during the preparations for a show by a Big Name Rock Band. I was chatting with some of the promoter's people, who remarked that they had two sets of the stage set and equipment, and there was a crew at the next venue on the tour already beginning setup. I remarked "There are two road companies of Cats. There are two road companies of the Ringling Brothers, Barnum and Bailey Circus. There have been rock groups where every member of the band was replaced over time, and the band kept the same name. This is all about branding and marketing. Why not have two sets of musicians and double your revenue?" The junior people laughed. The senior promoter in the suit looked very thoughtful.
A special "insecure software" levy on software responsible for more than 10% of "owned" machines on the net would be more appropriate.
That's a good idea. We do something like that at SiteTruth, where we down-rate commercial sites that don't have a real-world contact address on the site. We're looking at user-visible pages, though, not WHOIS. WHOIS data quality is too low.
I'm all in favor of this sort of thing. But don't drop the messages silently; reject them during the SMTP session if you can, or send a mail bounce if you can't. There's much to be said for having a hard-ass attitude about this, but you have to handle the false positives properly.
Anything that sends mail bounces needs to check SPF records. This makes it possible to stop joe-job mail bounce problems. (EXIM mailer people: please finish the implementation of SPF checking and advance it from "experimental", so large ISPs can use it.)
Also, quit whining that putting your real name on your WHOIS registration will get you annoying phone calls, threats, or whatever. I've had my real name and contact info on all my web sites and WHOIS information for a decade, and that's just not happening.
Why so seriously skeptic?
Because Nanosolar has supposedly been shipping for three years and there are no visible installations.
Just because they have pictures of a factory interior doesn't mean they have a factory that delivers working product at the claimed cost. It's all too common to see some minor advance in materials science hyped into something becoming real cheap, real soon now. Most of the roll-to-roll processes for making solar cells don't work.
Not all of them, though. Uni-Solar Ovonic really does have a working roll to roll process for making flexible solar cells. You can order their products, they have a dealer network, and they have many installations. Nanosolar has none of that. Just hype.
The earth is naturally slowing down at a rate that makes this sort of thing hardly worth mentioning.
Actually, the slowdown is only about 2ms/century, or about 0.054 microseconds/day. So a 1-microsecond jump in a day should be noticeable. This information is tracked. Here's the raw data from the Earth Rotation Service.
With GPS systems working down to 15cm, changes like this get noticed.
Hey! The Cone of Silence was a required deliverable insisted on by management, okay?
Some of those pictures are renders, and some are real. It turns out that the suspended transparent bubble was deleted before actual construction, along with the oval lighting rings and the curved-glass viewing galleries. Those are renders. The ones with conventional lighting on the ceiling are real, including the fancy podium. In this picture, the PCs, which look like conventional mini-towers, and their cables are in place.
So it's really a big, square mostly-empty room, with a bunch of Barco rear-projection monitors in front, and a fancy podium that cost US$90,000. The architect writes "this is main room in large building of company, which controls all electricity in Moscow. In this room 6 people working 24/7. Working group will not expand, but space is needed for groups of high ranked visitors."
That might actually be a reasonable design, given that problem. I can see the chief operator saying to the architect "Put enough empty space behind us so that when the oligarchs come to visit, they don't get in the way. Don't give them any chairs, or they'll hang around and muck things up. Put a VIP lounge somewhere else in the building, with the booze and the girls. Stick a few screens in there to repeat the big board, so they think they're running things."
At some point, the GPU goes on the CPU chip, and gets faster as a result.
Maybe.
GPUs need enormous bandwidth to memory, and can usefully use several different types of memory with separate data paths. The frame buffer, texture memory, geometry memory, and program memory are all being accessed by different parts of the GPU. Making all that traffic go through the CPU's path to memory, which is already the bottleneck with current CPUs, doesn't help performance.
A single chip solution improves CPU to GPU bandwidth, but that's not usually the worst bottleneck.
What actually determines the solution turns out to be issues like how many pins you can effectively have on a chip.
From Nanosolar's website, it sounds like they've been shipping panels commercially for the last two years, and that they have panel assemblies in both the US and Germany...
Yes, from Nanosolar's web site, it sounds that way. But as of 2009, "not one Solarply cell has been held yet in the hands of a consumer". There are no reports of actual Nanosolar installations. Supposedly they're building big solar panel installations for utility companies. So where are the announcements from those utility companies? Where are the regulatory filings, the planning documents, and the Google Earth pictures? Installing a big solar farm leaves a public record.
Here's an example of just that - the new Moesk control center for Moscow's electric network.
Take a look at the pictures. This looks like a movie set for a Bond movie. The architects got completely out of control here.
Notice the suspended transparent bubble for top management. It looks like it retracts into the ceiling. The lower operator's platform has steeply slanted sides, no railings, and chairs with wheels. The huge room only has eight operator positions.
I'll bet that, within a year or two, the people who actually have to run the grid set up a "field control center" with about twenty people with PCs, cork boards on the walls, 2-way radios for talking to field crews, a conference/map table, and some printers. The real work will be done there. A few people will sit in the big room and answer questions for management.
Here's the actual scientific paper, "Predicted Efficiency of Si Wire Array Solar Cells". That's by the same authors mentioned in the press release. While the thing does trap most of the light hitting it, only a fraction of the energy in that light is converted to electricity. In fact, this thing is currently less efficient than the better commercial solar cells.
From the paper: ... simulated photovoltaic efficency of 14.5%. ... Conclusion: ... "Si wire array solar cells have the potential to reach efficiencies competitive with traditional Si crystalline solar cells."
So, an interesting development, but no big breakthrough. There's a claim that it might be a cheaper way to make solar cells, but everybody who comes up with a new design makes that claim. (Nanosolar comes to mind; their technology is supposed to be cheaper, but so far they've spent half a billion dollars and apparently have only produced sample panels.)
There's a reason relatively little manufacturing is done in the US anymore,
The US, with a third of China's population, still does more manufacturing than China. US manufacturing output is still over twice that of China. US manufacturing employment, though, continues to drop. Manufacturing automation works very well today.
The giant urban slum is a phenomenon of the last 50 years. Fifty years ago, New York City was the biggest city in the world. It had poor people, but most of them had real jobs. New York is now in 13th place (one can argue over how to count), and the top 5 are all third world cities surrounded by slums. This is called "over-urbanization".
There's a step up from the "favelas out to the horizon". It's the "housing project". The better-organized big cities have vast numbers of cheap concrete high-rise apartment blocks, resembling US housing projects of the 1960s. That's what Tokyo, Beijing, Hong Kong, and Shanghai residential areas look like. It's an efficient way of warehousing a large number of humans. Whether it's a good life depends more on the people than the architecture. (Luxury apartment buildings and housing projects cost about the same to build per room for the basic structure. The luxury building may cost less to run, because vandalism and crime will be lower and the number of children per household will be lower.)
The US is unique in having most of its population in suburbs. Only cheap oil made that possible.
It seems surprising that the tank isn't crushed, but other tanks have come down from space without being crushed flat.
The industrial world is transitioning to twisted-pair Ethernet. From an industrial perspective, twisted-pair Ethernet is a nice interface. It's balanced drive and twisted pair, so it has very good common-mode noise rejection. (Better than RS-232 and 5V encoder signals, in fact.) There's full error checking and retransmission, unlike serial lines. This is important when it attaches to a welding robot, or runs past a circuit breaker that's switching 5KV. Long cable runs of 10baseT and 100baseT work fine. Bridges and routers are easily available and cheap, even in industrial form. And, of course, you can put many devices on one cable.
From a security perspective, though, this creates the problem that it's too easy to get devices on an Ethernet cable connected to the external Internet. Especially since the support programs for the devices tend to run on Windows. Most of the industrial Ethernet devices have little or no security.
Well, I have a current loop device on my desk. I had to design and build a USB to current loop converter for a 60mA 120VDC current loop.
But that, of course, is a retro technology tour de force.
It's surprising that nobody has replicated the V-1 "flying bomb" of WWII. That was the first cruise missile, and wasn't expensive. Carrying 850Kg of explosive, it could take out medium-sized buildings. Its main limitation was poor guidance. The Nazis fired about 8,000 at London, but couldn't hit a target smaller than a big city. The guidance system was a gyro/pendulum/magnetic compass system, and just flew the thing in a straight line until a small propeller/odometer had counted enough turns.
Since it flew straight and level, it was easy to shoot down. Still, "easy to shoot down" meant hundreds of interceptors and hundreds of anti-aircraft guns, and the success rate at shoot-down only reached 75% or so.
With a modern guidance system, one that could hit a target and didn't fly straight and level, these things could be formidable weapons today. If someone launched twenty of them, each programmed to take a different path to the same target, some of them would get through. Nobody has enough interceptors any more to take out an attack like that.
In addition to sending human-controlled robots to the moon, lets send along refineries and factories to produce solar panels.
Yeah, right. Back around 1985, I went to a conference where some AI professors were mouthing off about putting self-replicating factories on the Moon within 20 years. I asked "How soon can you do it in Arizona?" They didn't like that.
The article is bogus, but the problem is real. Computer support systems for investigators are hard to build. The FBI has struggled with this, taking about a decade to deploy their "Field Office Automation" system. They're hard for many of the same reasons medical systems are hard - much of the incoming data is unstructured, and many people enter data relevant to the same case. It's even harder than in the medical world, because links between various individuals and events are important, but unreliable. The "customers" aren't cooperative, they usually don't have unique identifiers, and a sizable fraction of the information is bogus. The security problems are tough to even define - exactly who's allowed to see what is a big issue.
The older law enforcement systems didn't offer much searchability. Unless you had a hard search key, like a driver's license number or a full name, you couldn't retrieve much. Now, everybody expects Google-like searchability, and the older systems just didn't have the machinery for that.
creating new algorithms that will make more tangential recommendations to users, which can help expand their interests,
The advertising industry already has that technology. Their idea of "expand interests" usually involves shopping, of course.
Many of you don't get it. Apple "iPad" isn't a general-purpose computer. It's an entertainment device.
It doesn't need a camera, and it doesn't need a keyboard, because it's a mostly-output device. It's an improved "e-reader", not a "netbook". It's not a "convergence" device, it's a "divergence" device.
It's overpriced. The price will drop, of course. Look at the price history of Apple's music players. The entry level device there is now $125.
It will succeed or fail based on what content it can access. If Apple and News Corp. work out a deal, and you can view all News Corp. content on the thing, it will be a big success.
I expect that a John Williams simulator can't be that far off. Williams is the composer who did the symphony orchestra scores for Jaws, Star Wars, most of Speilberg's works, and other industrial-strength dramatic productions.
As music, his music sucks. Listen to his music without a movie, and it brings to mind the tank commander's motto, "When in doubt, use the main gun." But it carries the production along. With a Williams score, a good production designer, and a big budget, a film can be a success even with a dumb plot and bad acting.