Apple also runs iTunes, which isn't very computer-like. In fact, you can think of it as an embryotic record label cum-distribution channel. But don't expect Apple to mention that for a few years yet, because that would really cause the record labels to hit the panic button.
The gold standard is still "TCP/IP Illustrated", Stevens, even though it is getting somewhat long in the tooth. The Kozierok book is essentially all prose descriptions of how the TCP/IP stack works, no code. There is simply no replacement for Stevens if you need to figure out how the stack is dealing with multicast packets or dozens of other situations.
There are some questionable organizational choices, such as starting of with SLIP and PPP in the first chapters. And gopher is pretty quaint these days.
The Kozierok book is probably a better choice for undergrads and people new to networking. The Stevens book is for crusty, bearded, rainbow-suspenders wearing Unix guys, or those who want to be.
The IQ tests aren't being used to identify the gene. The process is to administer the test, and then run a medical test to determine what genes are present. And the genetic tests run on the subjects are quite relaible. Once both tasks are done they simply correlate the IQ scores to the genes.
IQ tests are fine predictors of performance in many areas--people who score well on IQ tests tend to have much better performance at many real world tasks.
20 points is a huge difference. That's approximately the IQ gap between college graduates and non-college graduates in the US. It's more than one standard deviation in the IQ distribution.
That's an interestig possibility: Disney buys Pixar, and Jobs becomes CEO of Disney. This gives Apple access to Disney/ABC content for internet video distribution via iTunes, and there is a gigantic mind-meld of technology, content, and business models.
To use the biz school term, the industry is in the process of "disintermediation". The value-add for record labels was production and distribution of the physical artifact, talent scouting, and promotion. These tasks are (or can be) either disappearing altogether, or migrating to other locations in the supply chain.
The physical artifact is going away. If artists bypass the labels in favor of iTunes distribution CDs are no longer an issue. That leaves scouting and promotion as value-add functions for the record labels.
The marginal cost to Apple of adding a new act to iTunes is pretty close to zero. What does it take--a few megs of disk space on a server? Talent scouts were needed because it was expensive to sign an act, then produce and distribute the physical artifacts. So the question becomes: why do we need talent scouting if the cost of hosting a crappy act is zero? Put it all up. Hip-hop polka from the Cleveland tavern scene? Sure! If it sells it sells. If it doesn't sell, big deal. Of course, people still need to sort through all this, but this can be done by critics and promoters for the band once the music is up.
That leaves promotion. The record labels are good at this. But there's no particular reason that promotion can't be done on a for-hire basis. The bands, with their larger take from online sales, spend a part of that money to buy promotion services, or just tour a lot. Hell, the record companies have long used payola to get air time for their records, only to keep plausible deniablity they've been contracting this out to independent "promoters" who pay the DJs. There's no reason the bands can't hire the "promoters" themselves, or for that matter pay the DJs with groupies and blow. (Another opportunity for disintermediation!)
From Apple's standpoint they're the low cost provider in music distribution. They don't need to make _any_ money off of the iTMS. Their money comes from the iPods. Competitors will always have a higher cost business model, unless they get a special deal from the record labels, which of course also costs the labels money.
The trump card the record companies hold is their popular music back catalogs that date back over the last 40 years. People still buy the Byrds or Jefferson Airplane or David Lee Roth-era Van Halen. If they take their catalog and go home the value proposition of iTunes is greatly reduced. Of course, this also hurts the record companies, so that's a two-way street.
You'll note that Apple has been very nice to some artists. They had a co-marketing arrangement with U2 on their iPods, and obviously a slot on the home page at iTunes is very valuable. I think Jobs is setting up exactly the model described above, though he probably won't pull the trigger on it for a few years if he can help it, because he wants iTunes to solidify its position in the market and reduce the leverage that the record companies have with their back catalogs. Once iTunes controls the distribution channel the record companies don't have nearly as much leverage. Then U2 or a major band like them goes "label-less". As the first defector from the majors they clean up. They've already got brand name recognition, and a new release from them would be guaranteed to sell. Only now they get 70% of the revenue, not 5%. The bands most likely to go for this would be either aging, established acts like U2, Dylan, and the Stones, who have a built-in fan base, or new acts with nothing to lose but their chains.
Apparently not enough geeks around here have been burned by the suits and as a result are insuffienctly cynical.
The US government runs the root servers with a hands-off policy of benign neglect. This is a Good Thing. If that policy changes there is one-stop shopping for going somewhere to complain about it.
What will be the policy of an international group? China would certainly like to use any means at their disposal to further their censorship program. Ditto for many other nations. UN organizations are usually set up on a one-nation-one-vote basis. So the question is: would a majority of the governments in the world like a free speech basis for the DNS system? Nope. This is a chance for the non-democratic nations to vote about whether free speech is a good idea or not.
Even if a majority of the nations are in favor of maintaining the current system, there is no guarantee that the committees wouldn't be hijacked by government suits with an agenda. Most of the decsions happen out of sight of the public, and all sorts of shady deals could be cut. China pays a few million to the bank accounts of other committee members in order to get its way? Why not? Exactly that happened with the UN Oil for Food program in Iraq.
Basically, the proposal boils down into one to take power away from the t-shirts, give it to suits with an agenda contrary to that of the t-shirts, and move the field of battle from technical merits to bureaucratic infighting, where the t-shirts are at a comparative disadvantage. No thanks.
Yeah, they were doomed by he PC graphics card industry. NVidia was cranking out new graphics chips every six months, and SGI couldn't keep up with the product cycles. Buyers looked at SGI and saw slower, more expensive graphics when compared to cheap, ubiquitous PCs. The PC graphics industry had far higher volume, and all the talent migrated there. SGI's bread and butter went away, and they never found another meal ticket.
Yeah, I think that's the long-term plan. But Jobs, since he is not an idiot, won't announce that Apple plans to sell OSX/beige a year before they've got a product, thus allowing microsoft to crush them.
But suppose Apple has OSX/Intel on Apple-branded hardware going a year from now, and all the software vendors have their apps ported over. What would it take at that point to start selling OSX for beige? Apple could get several big vendors to sell it pre-loaded. Apple could arguably retain their hardware as a branded, upscale market niche. They've got about 2% of the market now. How much would they gain if they sold for beige? It's not at all unreasonable to think that they could triple their market share, particularly if MS fumbles Longhorn. This is pretty fudamentally different from the Power computing clones--since OSX could run on almost any existing hardware Apple's potential to grab a lot of market share fast is very real.
Once Apple has the transition to Intel complete and OSX running on Apple-branded intel hardware the move to beige boxes can happen very quickly--in the space of a few months. MS could easily be Pearl Harbored.
Actually, it had to do with the floppy drives. Apple had variable speed drives--they spun slower when writing to a sector at the edge, and faster when writing to a sector near the spindle, and they bought special floppy drives to pull this off. As I recall this was a hangover from the Apple ][ days, in fact a Woz hack. It let them get slightly more capacity on the floppy--you could save several extra K on a Mac-formatted floppy compared to a FAT-formatted floppy on an IBM.
Compared to the commerical world the management is highly decentralized. The rule is the one who pays the piper calls the tune, and to a large extent profs are independent operators who pull in research money. If they get research money they have a very big say in how it is spent. This can and often does result in multiple similar projects on campus. If someone is setting up a cluster, odds are there will be five other people doing the same thing.
Typically research money is taxed by the university and the money used to support campus-wide initiatives. So the campus networking people could be independent of any one prof. But they are just one more power center of many, including deans, profs, the admin, schools, colleges, etc.
If you provide free work the university will happily accept it. They're used to having people do things for them for free. If you want to get paid for something, insist on it.
You may often get 'free' labor in the form of students. Often this is a dead loss--it takes time to get them up to speed, but they're usually gone in a few months, and a lot of them have priorities other than working for you. But sometimes you get very, very good students, and they're a joy.
Don't expect a clear line of authority. The watchword is to be collegial.
Gould's book is a series of misdirections. It's popular, but in the academic world its critique has pretty much sunk without a trace.
I recomend Arthur Jensen's _The g Factor_, which discusses many of the things you've mentioned here, including test anxiety. In fact g/IQ tends to be pretty stable across tests. The math in Jensen's book can be daunting, but anyone with a good stats background should be able to follow the general outline.
Gigabit Ethernet gets dissed excessively as a cluster interconnect. It works fine in most small-to-mid installations, though the latency is higher than some other options.
It's not that the academic approach didn't pan out. It's just that the game industry swiped the commercially useful ideas and implemented them on commodity hardware.
Most of the FPS games swiped stuff from the academics and military. These two groups had been doing research on what were effectively FPS in the late 80's and early 90's, only they called them "simulators".
The game industry started wagging the research dog pretty quickly when they became a multi-billion dollar industry. The gamers like EA have historically not been very interested in research. They want to ship by xmas. That's starting to change. EA has been funding a game programming program at USC.
So, basically, VR research still goes on. There's a bunch of esoteric and not-so-esoteric stuff, as usual, scattered around various universities. The military still funds some of it. The "practical" VR research changed its name to game research or interactive graphics. That's nvidia doing faster graphics cards that can do photorealistic rendering, or people writing game engines.
You're looking at the default receive buffer size. You should be able to change that in the Java API with setSoReceiveBufferSize() or something like that. That just calls setSockOpt() on the underlying C API.
Typically you need to specify the receive buffer size at socket creation time, ie with the ServerSocket in Java; you can't dynamically change the receive buffer size.
Everyone was all interested in Taligent until they got a look at it. It was awful. I was working on NeXTs at the time, and it was painful to see the demo at ObjectWorld in the mid-90's. It sucked on a monumental scale. They had 512 MB machines (in 1995!) and it crawled. Plus flashing X-Windows colormaps! And over 10,000 C++ classes in the development framework!
God, that was one of the worst large development projects ever.
Both Buffy and Angel were cult shows. They didn't have huge mass market penetration, but had a respectable and loyal following. For example, Buffy got cancelled on the WB and moved to UPN around season 5. That doesn't happen to hit shows.
I used NeXTSTEP-92, and it was better than XP. The double-buffered graphics in the workspace were way slicker, and display postscript looked nice.
I think Apple has a significant advantage in the Cocoa framework. They can roll out fairly significant GUI apps in less time, and the whole infrastructure of the system is more amenable to development. Windows looks like a magnificent hack that requires more and more work to maintain. Of course, MS does have a lot of resources....
Back in the day there was actually a NeXT display postscript "virus". When it was processed by NeXTMail, the predecessor to the current Mac Mail app, it would make the display appear to melt. Glenn Reid of RightBrain wrote it, as I recall. In that more innocent age everyone thought it was pretty cool.
NeXT figured out that this could potentially be a gigantic security hole and switched off file access from display postscript.
The transaction costs are quite high for small purchases. The credit card companies are taking a big chunk of the revenue in microtransactions, which is why Apple likes to sell baskets of more songs.
Piper Jaffery estimates the operating margin for iTunes could hit 5-10%. "Operating margin" is the pre-tax, pre-interest profit divided by the sales revenue. They may be advertising a bunch, but that is part of Apple's plot to sell more iPods and, indirectly, more Macs, since the iPods are in turn encouraging people to buy Macs.
They're running the iTunes store as more or less a clever promotion that helps sell hardware. The music purchased further helps lock the user into Apple hardware. Suppose you upgrade your iPod in a couple years. Are you going to switch to a new music format on another brand's music player? Not bloody likely if you've already got several hundred dollars worth of music in ACC format.
Slashdot Mirror
Apple also runs iTunes, which isn't very computer-like. In fact, you can think of it as an embryotic record label cum-distribution channel. But don't expect Apple to mention that for a few years yet, because that would really cause the record labels to hit the panic button.
Well, I scanned it in depth.
The gold standard is still "TCP/IP Illustrated", Stevens, even though it is getting somewhat long in the tooth. The Kozierok book is essentially all prose descriptions of how the TCP/IP stack works, no code. There is simply no replacement for Stevens if you need to figure out how the stack is dealing with multicast packets or dozens of other situations.
There are some questionable organizational choices, such as starting of with SLIP and PPP in the first chapters. And gopher is pretty quaint these days.
The Kozierok book is probably a better choice for undergrads and people new to networking. The Stevens book is for crusty, bearded, rainbow-suspenders wearing Unix guys, or those who want to be.
The IQ tests aren't being used to identify the gene. The process is to administer the test, and then run a medical test to determine what genes are present. And the genetic tests run on the subjects are quite relaible. Once both tasks are done they simply correlate the IQ scores to the genes.
IQ tests are fine predictors of performance in many areas--people who score well on IQ tests tend to have much better performance at many real world tasks.
20 points is a huge difference. That's approximately the IQ gap between college graduates and non-college graduates in the US. It's more than one standard deviation in the IQ distribution.
Pfft. The Halliburton Hurricane Machine and Ernst Blofeld's Tsunami Machine are much cooler.
That's an interestig possibility: Disney buys Pixar, and Jobs becomes CEO of Disney. This gives Apple access to Disney/ABC content for internet video distribution via iTunes, and there is a gigantic mind-meld of technology, content, and business models.
The XSAN manual says 1 petabyte per volume is the max under 10.4.
The physical artifact is going away. If artists bypass the labels in favor of iTunes distribution CDs are no longer an issue. That leaves scouting and promotion as value-add functions for the record labels.
The marginal cost to Apple of adding a new act to iTunes is pretty close to zero. What does it take--a few megs of disk space on a server? Talent scouts were needed because it was expensive to sign an act, then produce and distribute the physical artifacts. So the question becomes: why do we need talent scouting if the cost of hosting a crappy act is zero? Put it all up. Hip-hop polka from the Cleveland tavern scene? Sure! If it sells it sells. If it doesn't sell, big deal. Of course, people still need to sort through all this, but this can be done by critics and promoters for the band once the music is up.
That leaves promotion. The record labels are good at this. But there's no particular reason that promotion can't be done on a for-hire basis. The bands, with their larger take from online sales, spend a part of that money to buy promotion services, or just tour a lot. Hell, the record companies have long used payola to get air time for their records, only to keep plausible deniablity they've been contracting this out to independent "promoters" who pay the DJs. There's no reason the bands can't hire the "promoters" themselves, or for that matter pay the DJs with groupies and blow. (Another opportunity for disintermediation!)
From Apple's standpoint they're the low cost provider in music distribution. They don't need to make _any_ money off of the iTMS. Their money comes from the iPods. Competitors will always have a higher cost business model, unless they get a special deal from the record labels, which of course also costs the labels money.
The trump card the record companies hold is their popular music back catalogs that date back over the last 40 years. People still buy the Byrds or Jefferson Airplane or David Lee Roth-era Van Halen. If they take their catalog and go home the value proposition of iTunes is greatly reduced. Of course, this also hurts the record companies, so that's a two-way street.
You'll note that Apple has been very nice to some artists. They had a co-marketing arrangement with U2 on their iPods, and obviously a slot on the home page at iTunes is very valuable. I think Jobs is setting up exactly the model described above, though he probably won't pull the trigger on it for a few years if he can help it, because he wants iTunes to solidify its position in the market and reduce the leverage that the record companies have with their back catalogs. Once iTunes controls the distribution channel the record companies don't have nearly as much leverage. Then U2 or a major band like them goes "label-less". As the first defector from the majors they clean up. They've already got brand name recognition, and a new release from them would be guaranteed to sell. Only now they get 70% of the revenue, not 5%. The bands most likely to go for this would be either aging, established acts like U2, Dylan, and the Stones, who have a built-in fan base, or new acts with nothing to lose but their chains.
The US government runs the root servers with a hands-off policy of benign neglect. This is a Good Thing. If that policy changes there is one-stop shopping for going somewhere to complain about it.
What will be the policy of an international group? China would certainly like to use any means at their disposal to further their censorship program. Ditto for many other nations. UN organizations are usually set up on a one-nation-one-vote basis. So the question is: would a majority of the governments in the world like a free speech basis for the DNS system? Nope. This is a chance for the non-democratic nations to vote about whether free speech is a good idea or not.
Even if a majority of the nations are in favor of maintaining the current system, there is no guarantee that the committees wouldn't be hijacked by government suits with an agenda. Most of the decsions happen out of sight of the public, and all sorts of shady deals could be cut. China pays a few million to the bank accounts of other committee members in order to get its way? Why not? Exactly that happened with the UN Oil for Food program in Iraq. Basically, the proposal boils down into one to take power away from the t-shirts, give it to suits with an agenda contrary to that of the t-shirts, and move the field of battle from technical merits to bureaucratic infighting, where the t-shirts are at a comparative disadvantage. No thanks.
I don't see how this can hold up in court. Everyone knows there ain't no Sanity Clause.
No muc?
Yeah, they were doomed by he PC graphics card industry. NVidia was cranking out new graphics chips every six months, and SGI couldn't keep up with the product cycles. Buyers looked at SGI and saw slower, more expensive graphics when compared to cheap, ubiquitous PCs. The PC graphics industry had far higher volume, and all the talent migrated there. SGI's bread and butter went away, and they never found another meal ticket.
But suppose Apple has OSX/Intel on Apple-branded hardware going a year from now, and all the software vendors have their apps ported over. What would it take at that point to start selling OSX for beige? Apple could get several big vendors to sell it pre-loaded. Apple could arguably retain their hardware as a branded, upscale market niche. They've got about 2% of the market now. How much would they gain if they sold for beige? It's not at all unreasonable to think that they could triple their market share, particularly if MS fumbles Longhorn. This is pretty fudamentally different from the Power computing clones--since OSX could run on almost any existing hardware Apple's potential to grab a lot of market share fast is very real.
Once Apple has the transition to Intel complete and OSX running on Apple-branded intel hardware the move to beige boxes can happen very quickly--in the space of a few months. MS could easily be Pearl Harbored.
Actually, it had to do with the floppy drives. Apple had variable speed drives--they spun slower when writing to a sector at the edge, and faster when writing to a sector near the spindle, and they bought special floppy drives to pull this off. As I recall this was a hangover from the Apple ][ days, in fact a Woz hack. It let them get slightly more capacity on the floppy--you could save several extra K on a Mac-formatted floppy compared to a FAT-formatted floppy on an IBM.
Typically research money is taxed by the university and the money used to support campus-wide initiatives. So the campus networking people could be independent of any one prof. But they are just one more power center of many, including deans, profs, the admin, schools, colleges, etc.
If you provide free work the university will happily accept it. They're used to having people do things for them for free. If you want to get paid for something, insist on it.
You may often get 'free' labor in the form of students. Often this is a dead loss--it takes time to get them up to speed, but they're usually gone in a few months, and a lot of them have priorities other than working for you. But sometimes you get very, very good students, and they're a joy.
Don't expect a clear line of authority. The watchword is to be collegial.
I recomend Arthur Jensen's _The g Factor_, which discusses many of the things you've mentioned here, including test anxiety. In fact g/IQ tends to be pretty stable across tests. The math in Jensen's book can be daunting, but anyone with a good stats background should be able to follow the general outline.
Gigabit Ethernet gets dissed excessively as a cluster interconnect. It works fine in most small-to-mid installations, though the latency is higher than some other options.
Most of the FPS games swiped stuff from the academics and military. These two groups had been doing research on what were effectively FPS in the late 80's and early 90's, only they called them "simulators".
The game industry started wagging the research dog pretty quickly when they became a multi-billion dollar industry. The gamers like EA have historically not been very interested in research. They want to ship by xmas. That's starting to change. EA has been funding a game programming program at USC.
So, basically, VR research still goes on. There's a bunch of esoteric and not-so-esoteric stuff, as usual, scattered around various universities. The military still funds some of it. The "practical" VR research changed its name to game research or interactive graphics. That's nvidia doing faster graphics cards that can do photorealistic rendering, or people writing game engines.
You're looking at the default receive buffer size. You should be able to change that in the Java API with setSoReceiveBufferSize() or something like that. That just calls setSockOpt() on the underlying C API.
Typically you need to specify the receive buffer size at socket creation time, ie with the ServerSocket in Java; you can't dynamically change the receive buffer size.
Everyone was all interested in Taligent until they got a look at it. It was awful. I was working on NeXTs at the time, and it was painful to see the demo at ObjectWorld in the mid-90's. It sucked on a monumental scale. They had 512 MB machines (in 1995!) and it crawled. Plus flashing X-Windows colormaps! And over 10,000 C++ classes in the development framework! God, that was one of the worst large development projects ever.
Both Buffy and Angel were cult shows. They didn't have huge mass market penetration, but had a respectable and loyal following. For example, Buffy got cancelled on the WB and moved to UPN around season 5. That doesn't happen to hit shows.
Longhorn: It will work two years from now, after the first major upgrade to the original release.
Actually, the Sainted Steve Jobs popularized the phrase with NeXT fifteen years ago.
I used NeXTSTEP-92, and it was better than XP. The double-buffered graphics in the workspace were way slicker, and display postscript looked nice. I think Apple has a significant advantage in the Cocoa framework. They can roll out fairly significant GUI apps in less time, and the whole infrastructure of the system is more amenable to development. Windows looks like a magnificent hack that requires more and more work to maintain. Of course, MS does have a lot of resources....
NeXT figured out that this could potentially be a gigantic security hole and switched off file access from display postscript.
Piper Jaffery estimates the operating margin for iTunes could hit 5-10%. "Operating margin" is the pre-tax, pre-interest profit divided by the sales revenue. They may be advertising a bunch, but that is part of Apple's plot to sell more iPods and, indirectly, more Macs, since the iPods are in turn encouraging people to buy Macs.
They're running the iTunes store as more or less a clever promotion that helps sell hardware. The music purchased further helps lock the user into Apple hardware. Suppose you upgrade your iPod in a couple years. Are you going to switch to a new music format on another brand's music player? Not bloody likely if you've already got several hundred dollars worth of music in ACC format.