You need to understand the code fragment you are coding to do the must useful optimizations which are algorithmic ones. A badly coded better variant of an algorithm will beat well coded, but inherenetly slower version.
Absolutely. But one need not be programming in assembly to do the sort of algorithmic optimizations you speak of. In fact, in higher-level languages, where source code content is determined largely by the algorithm being implemented (rather than by the low-level details of the implementation, as in assembly language), you have a better chance of 'seeing' and successfully implementing a faster algorithm. This sort of thing is one of the big reasons why assembler loses in the long run.
It seems to me that there are two basic types of optimization task:
Choosing the fastest algorithm for a particular task.
Making a given algorithm run as fast as possible on a given piece of hardware.
Of course, this is a bit of an oversimplification, since it assumes that the fastest algorithm for a given task is going to be the same on all platforms. Usually, this is true, but probably there are a few minor cases where it isn't. But it works most of the time.
At any rate, humans are much better than machines at the first type of optimization. But machines are much better than humans at the second (even if only because normal humans don't have the patience required to do the enormous amount of repetitive analysis that it requires).
I read that the 700 MHz Crusoe chip actually only attains the performance of a 500MHz Pentium
So there you have it - code morphing, or just in time compiliing or whatever gives you about 70% of the performance precompiled code.
Comparing interpreted to compiled code on a MHz-for-MHz basis makes no sense, because MHz is not an intrinsically interesting number. Most people don't care about what their computer's clock speed is. They care about how fast it runs applications (i.e. MIPS), and how much it costs. So don't compare Crusoe to an Intel chip on the basis of a MIPS/MHz ratio. Compare it on the basis of a MIPS/$$$ ratio. How well does a Crusoe perform compared to an Intel chip *of the same price*?
In our reality, the media is all-too-often apt to side with the large corporation. Hackers/Crackers are bad, right?!? Why do we have reason to believe that he'll be cast in something other than negative light?
You're kidding, right? Television news loves the underdog. And they're happy to trash a large corporation, because it's good for ratings. They'll even compromise journalistic integrity to do so - think of Dateline NBC's coverage of those exploding pickup trucks. A David-Goliath conflict like this one is great for ratings - but it requires that you make David look like the good guy.
Any coverage of this that puts a human face on the people involved is good for those of us who consider reverse engineering a legitimate activity. It's far better for us if they interview him, give him a name and face, than if they just stick to referring to him as a nameless "hacker". A real 16-year-old boy can't be anywhere near as menacing on camera as the mainstream media's vague concept of a "hacker" is.
Of course, they can still try to trash him, make him look like a computer criminal, a kindred spirit to Mitnick. But I doubt they will, because Johansen-as-criminal is not likely to sell with the public. He never broke in to anyone's computer, never stole any individual private information (such as credit card numbers). The idea that what he did was wrong will be jarring to most people's intuitive idea of property, which is that you can do whatever the hell you want with the things that are your property.
This is no longer about Jon Johansen, or the cracking of DeCSS, this is about Abuse of privilege.
Indeed. Here, as I see it, is why they're after Johansen:
The CSS licensing is not about money, it's about control. (In fact, as I understand it, the licenses are free.) As long as one has to sign a contract with the DVD CCA to get the information needed to build a DVD drive that handles CSS-protected discs, the DVD CCA has some degree of control over DVD drive manufacturers. I'm sure that the licensing contract prevents manufacturers from selling consumer-priced unrestricted hardware (i.e. drives that happily ignore region coding and can do bit-for-bit writes). Currently, all such drives are well out of the price range of the casual consumer (though not the serious, professional pirate).
But now, thanks to DeCSS, all the information needed to build a DVD drive is out in the open. Anyone who wants to can put together an unrestricted drive and sell it for $200, if that price point is profitable for them.
The DVD CCA faced a tough choice when it came to CSS. They had to choose whether to patent the system (assuming this was possible - probably, giving the current state of patents), or keep it a trade secret. Both choices have advantages and disadvantages:
Patent:In this case, they can prevent anyone from building cheap unrestricted players, because the patent gives them control over the system's use. However, the information needed to decrypt CSS is now out in the open, and anyone who wants to write a DeCSS-like software utility in order to pirate DVDs (to some other media) can do so quite easily - they just can't release it publicly, or must do so anonymously.
Trade Secret:At first, they still have control over who builds drives. And writing DeCSS is harder, because it requires reverse engineering. However, if/when the system is reverse engineered and the information becomes public, they lose all control - now anyone who wants to can legally build an unrestricted player and sell it for $200.
The CCA chose the second option, and it has backfired on them. Now they're trying to save themselves via the court system. This is why the reverse engineering issue is probably far more important to them than the DMCA issue. If they succeed in nailing Johansen, they'll probably have frightened off anyone who was thinking of reverse-engineering the system themselves (or using the DeCSS-derived information now on the web) in order to build a player.
Note that if this happens, the CCA will have effectively aquired the same rights as if they'd patented the system - but without ever formally disclosing how the system works, as a patent would normally require. So this case could set a rather dangerous precedent: The whole point of patents is to enhance technological development, by encouraging disclosure. And the carrot used to encourage disclosure is a government-guaranteed monopoly for a limited period of time. Putting reverse engineering on legally shaky ground means that companies have a better chance of keeping a monopoly on a technology that has trade secret status.
Right now, when companies are deciding whether or not to patent something, they have to ask themselves the question: "How long before someone will be able to re-create this technology without spying on us?" If the answer is over a certain threshold, it's probably better to keep it a trade secret. If reverse engineering becomes de-facto illegal, then the question becomes "How long before someone will be able to re-create this technology without spying on us or reverse-engineering our product?" Obviously, the answer to the second question will often be greater (and never less) than the answer to the first, and thus is more likely to be over the magic threshold where patenting becomes a bad idea.
And when you consider that the question isn't really whether or not someone resorted to spying (or reverse engineering), but rather whether or not you can convince a court of this, it's even worse - making a case for reverse engineering is probably a good bit easier, if the judge doesn't understand technology well enough to understand what reverse engineering is all about. So we may even see companies trying to convince judges that a competitor illegally reverse-engineered their product when in fact no such thing took place.
Finally, it should also be noted that while this may prevent companies from going for some patents, it won't prevent them from going after most stupid ones, like the Amazon one-click, because the ability to legally reverse engineer doesn't do much to help you figure out most such 'technologies', anyways.
So, if not to inform, what purpose do adds serv? They serv only to create want and need, and unless you're the mental age of a 10-year-old, they won't work on you..
Advertisements no longer teach or inform, they merely reinforce consumerism.
We seem to be having a basic misunderstanding. As I see it, there are 3 possible explanations for the huge amount of advertising that is done:
Because it informs consumers of products' existence.
Because it enforces consumerism - i.e. convinces (via appeal to reason or gut reaction) people that they need to purchase your product.
Because advertisers have been convinced that it is necessary, when in fact it isn't.
I think we can agree that the answer isn't really #1 - only a small amount of advertising is justified for the purpose of informing consumers. In the comment I originally responded to, you seemed to be advocating #3 - you said that creating wants and needs only worked on mental-10-year-olds, and that advertising didn't really work. This is where my disagreement lies - I do think that advertising works to create wants and needs on the general public, and not because most members of the general public have a mental age of 10. Or at least, television and radio advertising works this way - mainly because it's so intrusive, compared to a banner or billboard, which someone can easily look away from.
I agree. However, back in the early 20th, advertisments were less instrusive (compared to now).
I do agree that there has been a shift in the common view of marketing in the last 50 years or so. There was a recent article in Newsweek whose existence serves as rather nice evidence of such a shift: The article was all about how Toyota was working to improve its image in the eyes of younger drivers - apparently the median age of Toyota drivers is fairly high. The article talked about how they'd hired a squad of hip young ad designers, about new designs they were coming out with, and so on - various people from Toyota commented, in an official capacity, on this new campaign. What's interesting here is that they are utterly shameless about the fact that they are attempting to manipulate people - they're happy to talk to Newsweek about their strategies for doing this. 50 years ago, I'm guessing that most companies would want to keep quiet about such strategies - letting it be known how consciously they were working to manipulate the consumer would lead to a backlash from consumers who didn't like being manipulated. But today, you can find articles like the Toyota one in any weekly newsmagazine.
(Speaking of commercials, have you noticed how US TV averages between 2.4 and 2.9 minutes vs 1.7 to 2.2 minutes of advertisement time on Canadian TV?)
Well, noticing this this would require that I own a television.;) I got rid of mine a few months ago. And I never watched Canadian TV even when I had one. However, when in Brussels a year ago, I found it really interesting to watch US imports (specifically, the Simpsons) with all the commercials cut out. A scene would end with a minor cliffhanger, in a way that was obviously meant to be followed by a commercial. And then it would cut immediately to the following scene, with the cliffhanger's resolution.
So, if not to inform, what purpose do adds serv? They serv only to create want and need, and unless you're the mental age of a 10-year-old, they won't work on you.
I'm afraid you're being incredibly optimistic here. Can you realistically say that most advertising exists to inform consumers of a products' existence? I don't think so. How many people do you think are unaware of the existence of Coca-Cola? If they were to cut their advertising budget in half, would there be a significant drop in their brand recognition? I doubt it.
Only the momentum of the "consumerism" of the 1950s through 1980s keeps people advertising in this day and age.
In his book Democracy In America, Alexis De Toqueville commented on how pervasive he found commercial advertising to be in the US. He was writing in first part of the 19th century. Advertising has been with us since long before the 1950s.
I can't wait for it to die.
I can, but only because I'm afraid I don't have much choice about it.
Students pay for education, not for unlimited mp3-downloading bandwidth.
The idea that universities install dorm-wide networks to improve education is laughable. They do it because they're competing for students who want such a connection - and who mostly want it for reasons that aren't particularly educational in nature.
1) To burn a disk in the amount of time a consumer wishes to wait would require a very very expensive DVD mastering system. I don't think many consumers would hang around for the 30 minutes to an hour it would take to copy a DVD.
That's why you give users an account, just like they have at Blockbuster now. When someone wants to rent something, they go to the store's web page, log in to their account, browse through available titles, order the ones they want. The store checks the stock of pre-burned CDs, and if none of the requested title are available, notifies attendants to burn new copies. Later in the day, on the way home from work, they stop by to pay and pick up the discs.
There's another issue here: You've probably got high school students paid minimum wage operating this thing. They're going to be taking home free copies of every movie in the store, unless you have some good controls and tracking built into the hardware doing the burning.
2) Unless you charge the customer who has a copy of a DVD for the length of time they have the DVD (i.e., 1day=$2, 2days=$3, 3days=$4 etc.) there would be no incentive for them to ever bring it back. You would be basically selling DVDs at a very low price.
Okay, so giving them the discs indefinitely is a problem. Require them to return the discs within, say, a month. If they don't, you're going to put a hefty charge (say, twice what they'd pay to buy each movie outright) on their credit card - which you will happily refund, once they do actually return the discs.
To satisfy my curiosity, could you cite specific examples of left-wing propaganda on NPR? Perhaps this is different between local stations, but I've been impressed by the balance of opposing viewpoints on NPR in my area (WBEZ in Chicago).
It should be noted that NPR recieves little or no direct government funding. NPR is a content provider - they don't operate any radio stations. Instead, radio stations pay them to broadcast their content. Now, many of these radio stations are public (and therefore recieve government funds), so it's not as if they're completely untainted. But like yourself, I find that they generally give balanced coverage of issues - certainly moreso than most strictly-privately-funded media.
Rubbish. Their focus on X86 is based on their desire to make a buck. Additional code morphing software is simply not required because the market isn't there to justify it.
It's quite possible that they're releasing the current Crusoe and hyping the low power requirements because it's a way for them to go ahead and start making some revenue to fund future development. There's no reason to assume that Crusoe is the ultimate application they've visualized for the technology.
You could only justify the PowerPC emulation if you were extremely generous.
One of the beauties of the software-based approach is that Transmeta need not develop emulation themselves for every IS that gets supported. They could license the required technical info (VLIW layer, etc.) to Apple under NDA, and let Apple develop the PPC morphing layer. So this way they both make money from the licensing, and from selling chips to Apple. Generosity need not enter into it.
MacOS has had "virtual memory" since System 7, but it blows. It basically spills over to disk when you run out of physical RAM, though it's been improving very much with each OS release.
Uh, this is a feature of a lot of VM systems. What do you think a Linux swap partition, or the Windows swapfile, are for?
But yeah, System 7 had a pretty lousy VM implementation, from what I've read. If I understand correctly (and I could be way off), the third-party utility 'RamDoubler' was basically a replacement for the OS's built-in VM system.
You don't know far enough. FX!32, at least, did binary-to-binary translation, and at least some JVMs translate Java bytecodes to native code. (I don't know whether the 68K emulation in PowerPC MacOS ever did binary-to-binary translation.)
I'm aware of how FX!32 works - I went and looked it up after I wrote that. But I don't retract my point: it's still a strictly software solution, and the hardware provides no support for the process. Presumably, a considerable amount of software infrastructure is needed to make it work - x86 binaries have to be recognized and translated, and those translated binaries stored, for example. OTOH, unless you're trying to handle multiple instruction sets at once, there's no reason why Crusoe should require any more 'help' from the OS than an ordinary x86. And even if it does start handling multiple instruction sets, the solution is considerably simpler and more elegant than FX!32.
Also, FX!32's performance target is considerably less ambitious than that of Crusoe: The FX!32 white paper says they're shooting for code with 70% of the performance of a native-compiled Alpha application.
However, in it's own domain - high end systems - the Athlon *does* have quite an edge over the PIII. It has 3 integer execution units compared to the PIII's 2, plus 3 floating point execution units again compared to 2 for the PIII. Just as importantly, unlike the PIII the Athlon actually has the unrestricted ability to simultaneously issue the instructions to keep those execution units busy.
The question is: Do the technical improvements you mention result in observable improvements in areas that are important to the customer? The user cares how fast applications run, not how many integer execution units his CPU has. If the Athlon's technical advantages don't translate into significant speed improvements (say, because the CPU spends so much time waiting on other system components anyways), then they aren't significant in the marketplace.
On the other hand, Crusoe's technical advantages do have the potential to improve system performance in areas visible and important to the user: lower cost, and either greater battery life or lower battery weight. And the low temperature has a potential to lead to user-visible improvements, as well.
I don't think this is really a paradigm shift because similar things have been done before. DEC's FX/32, the Java Virtual Machine, and the 68K emulation in the PowerPC MacOS have been doing this for a while.
AFAIK, all the examples you name are software emulation. Sure, FX/32 allows an Alpha to emulate an x86 at speeds comparable to x86 native hardware. But an Alpha capable of doing this is much more expensive than the x86 to which it compares, and the program still executes at only around 70% of the speed of native-compiled Alpha code.
If Crusoe (or its descendants) allow you to execute an instruction set at 90+% of the speed of a CPU that is hardwired for that set, when the Crusoe and the hardwired CPU in question are the same price, then this could indeed be a paradigm shift.
They did say that the chip itself can run multiple instruction sets without resetting between them, but that's still a far cry from saying they've solved the problem of scheduling multiple OSes: since every OS has its own scheduler, you'd need a meta-OS with a meta-scheduler. Not to mention the problems of locking hardware access, etc.
I don't see too many people wanting to run multiple OSs at once - it's an absurd waste of resources. But I do see people using this technology to build emulators that compare to native hardware. Consider what might happen to SoftWindows, on MacOS, if this chip were able to run PPC code at a good pace. In that case, one could write a SoftWindows that ran x86 apps as fast as if they were on x86 hardware. However, the trick is that you need OS support to really do this right.
What I'm envisioning is that there's a register somewhere in the chip that indicates what instruction set it's currently using. There's also an instruction to atomically set both this register and the program counter register at the same time. Then, the OS stores a value for this register (along with the program counter and various other info that it already maintains) local to each process. When a process is dispatched, that register gets set simultaneously with the program counter, using the instruction I mentioned. Of course, this means your scheduler needs to be written in the VLIW instruction set (since this instruction won't exist in an emulated instruction set). But basically, any time a context switch is required to get from one chunk of code to another, they can have different instruction sets. (Note also that hardware interrupt vectors will have to contain a value for this register as well as a memory address.)
So does the VLIW substrate have an sort of intrinsic endianness? If it is little-endian in nature like the x86 that it's so good at emulating, then this could result in a noticable performance hit when emulating a big-endian CPU like the 68k or PPC.
I mean, Linux exists because there is a NEED for it. Just how much NEEDS there is for Linux to run on N64 or Playstation, or microwave oven or toaster?
Are you sure about this? How much of Linus' motivation for writing Linux was to fill his need for a decent free Unix-ish system on x86 hardware? And how much of it was simply the fact that he derived a sort of playful enjoyment from it?
More 'important' new developments than we often realize arise from play. The guy attempting this port may not end up generating anything useful. But at the very least, he'll learn a good deal in the process, and be able to apply that knowledge on more genuinely useful projects involving Linux or the N64, or other software or hardware entirely. And maybe he'll generate a neat toy that others can pick up, play with, and learn from.
Not really. The speed is plenty acceptable to me, and certainly no worse than any other web-based e-mail I've tried. the accounts are inundated by AOL-level amounts of spam
Evidence? In the year I've been using it, I've only ever gotten one piece of spam in my hotmail account, and I'm pretty sure that this was a result of me providing my e-mail address to a website that didn't deserve it. The only e-mail address where I see significant spam is the one I use when I post to Usenet. But many people sign up for the account when they first begin to use the internet, and don't want the trouble of migrating even after they acquire cluefullness.
I was clueful well before I started using Hotmail - I was using the net and e-mail from a shell account well before 'the Internet' became synonymous with 'the web', and certainly well before Hotmail came into existence. But when I got out of school, I decided I wanted an account for my personal e-mail that I could get at from pretty much anywhere (this means web-accessible: firewalls often block telnet, and the 'batch transfer' nature of HTTP (all my editing happens on the local machine, rather than interactively) makes it far more usable on a high-latency connection in the event that, say, I'm overseas). And I wanted an address that wouldn't go away if I changed jobs or ISPs. Hotmail fit the bill.
In short: Because he doesn't really respect slashdot or the net at all. He is fundamentally authoritarian in outlook, but posting to a forum that is dominated by people who oppose authority - or at least, unearned authority.
The long version:
Most of the people who participate in these discussions have probably read ESR's "The Cathedral and the Bazaar". And most think that Bazaars are pretty cool, and that Cathedrals are kinda lame. But Katz seems like a guy who would be more at home in an old-media Cathedral than a new-media Bazaar. And so he attempts to treat Slashdot like an old-media Cathedral, rather than the new-media Bazaar that it is (mostly) intended to be. He posts his essays as if they were wisdom handed down from upon high, and rarely if ever participates in the discussions that ensue (or if he does, it's as an AC or under a pseudonym). When he addresses opponents, it's not in the comments, on equal terms. Instead he quotes their e-mail in his column, where he has complete control over what is printed, and can simply cut out a point he can't answer. He also writes articles with titles like 'Voices from the Hellmouth', which contain lots of quotes from his e-mail. Seeming to suggest that he is somehow making these voices heard, when they otherwise wouldn't - nevermind that most of the people who e-mailed him could have easily posted their comments directly instead.
On the other hand, in recent months we've seen people like Orson Scott Card and John Carmack posting comments to slashdot. If you'd told me at age 15, right after I read Ender's Game for the first time, that I would one day be able to read and participate in a forum that OSC apparently sometimes read and participated in, I'd've fallen off my chair. Substitute "17", "played Doom", and "John Carmack" in the appropriate places in the above sentence, and it still holds true. These are guys whose authority comes from excellence - Ender's Game and the Doom and Quake engines were extraordinary achievements, far more significant than anything Katz is ever likely to do. But they are quite willing to step out from behind the pulpit and interact with random people in a forum where pretty much anyone with a net connection can come and participate. These are people who seem to understand new media, and aren't afraid of it.
Katz, on the other hand, is afraid of new media. (And new technology in general: Note the Luddite-ism that runs through some of his essays.) Because he apparently wants authority, but he's not really good enough at anything to aquire the authority that comes from excellence. So instead he seeks the authority that comes from possessing a voice on a limited-access medium, the same kind of authority that a talking head on the television has.
And so the resentment that many Slashdotters feel towards Katz is actually quite akin to the resentment they feel towards a mainstream news source that they believe reports poorly on some issue important to him: They feel that he has authority that he does not deserve. Evaluated honestly, Katz's essays are no more informative, insightful or well-written than the best user comments on Slashdot - often, they're less so. They don't deserve to be elevated to a position of greater prominence than those comments.
Interesting. I've never played it, but a friend with a Mac has been raving about it for years.
I remember watching a friend play it briefly. At one point, he opened a door and saw this weird flying creature he'd never seen before. He yelped in surprise and terror. This was not a person who frequently yelped, or felt surprise and terror. Like Half-Life, and unlike Quake, Marathon demonstrates that what really makes a game succeed is good play, design, and plotting. Great graphics technology won't wrap your mind up in the world of the game the way that a good treatment of these other elements will.
I believe it creates the illusion of looking up + down through a technique called "Y-shearing" to move the viewpoint.
You are correct. To be precise, what happens is that the view plane moves vertically up or down, rather than rotating around the center of projection. IIRC, Duke Nukem 3D used a similar approach.
Why should the brain implant be necessary? Why couldn't you deliver the "visual" information using a device close to a Braille converter? I imagine a little device that you can strap onto the back of your hand or wherever the skin has enough nervous endings to discriminate separate tactile impulses.
This has been done. I forget all the details, but Daniel Dennett writes about it in Consciousness Explained. Essentially the system involved cameras mounted on glasses, and a grid of "tinglers" strapped to the subjects' abdomen or back. With some practice, they could learn to interpret the impulses from the grid as 'sight'. A few interesting notes:
The response time is fairly slow - on the order of several seconds to identify/recognize anything.
Subjects were able to distinguish between tingler input and normal tactile sensations, so that they didn't (for example) interpret an itch in the tingler area as 'visual' input.
Supposedly, one subject (a man blind from birth) was given cameras with a zoom lens. When the researcher hit the zoom button without the subject's knowledge (causing the image to 'loom'), the subject jumped back and shielded himself with his arms. This one sounds almost too neat to be true.
Long term, this approach is unlikely to produce anything useful, partly because there's no place on the skin where nerves are packed densely enough that you can use a tingler grid with sufficient resolution to be useful, and partly because of the slow response time (OTOH, babies must 'learn to see' as well, and this doesn't happen overnight - maybe with time they'd adapt well enough that the response time was reasonable).
Slashdot Mirror
Absolutely. But one need not be programming in assembly to do the sort of algorithmic optimizations you speak of. In fact, in higher-level languages, where source code content is determined largely by the algorithm being implemented (rather than by the low-level details of the implementation, as in assembly language), you have a better chance of 'seeing' and successfully implementing a faster algorithm. This sort of thing is one of the big reasons why assembler loses in the long run.
It seems to me that there are two basic types of optimization task:
Of course, this is a bit of an oversimplification, since it assumes that the fastest algorithm for a given task is going to be the same on all platforms. Usually, this is true, but probably there are a few minor cases where it isn't. But it works most of the time.
At any rate, humans are much better than machines at the first type of optimization. But machines are much better than humans at the second (even if only because normal humans don't have the patience required to do the enormous amount of repetitive analysis that it requires).
Complete brain fart. That should read "Comparing Crusoe to Intel..."
So there you have it - code morphing, or just in time compiliing or whatever gives you about 70% of the performance precompiled code.
Comparing interpreted to compiled code on a MHz-for-MHz basis makes no sense, because MHz is not an intrinsically interesting number. Most people don't care about what their computer's clock speed is. They care about how fast it runs applications (i.e. MIPS), and how much it costs. So don't compare Crusoe to an Intel chip on the basis of a MIPS/MHz ratio. Compare it on the basis of a MIPS/$$$ ratio. How well does a Crusoe perform compared to an Intel chip *of the same price*?
You're kidding, right? Television news loves the underdog. And they're happy to trash a large corporation, because it's good for ratings. They'll even compromise journalistic integrity to do so - think of Dateline NBC's coverage of those exploding pickup trucks. A David-Goliath conflict like this one is great for ratings - but it requires that you make David look like the good guy.
Any coverage of this that puts a human face on the people involved is good for those of us who consider reverse engineering a legitimate activity. It's far better for us if they interview him, give him a name and face, than if they just stick to referring to him as a nameless "hacker". A real 16-year-old boy can't be anywhere near as menacing on camera as the mainstream media's vague concept of a "hacker" is.
Of course, they can still try to trash him, make him look like a computer criminal, a kindred spirit to Mitnick. But I doubt they will, because Johansen-as-criminal is not likely to sell with the public. He never broke in to anyone's computer, never stole any individual private information (such as credit card numbers). The idea that what he did was wrong will be jarring to most people's intuitive idea of property, which is that you can do whatever the hell you want with the things that are your property.
Indeed. Here, as I see it, is why they're after Johansen:
The CSS licensing is not about money, it's about control. (In fact, as I understand it, the licenses are free.) As long as one has to sign a contract with the DVD CCA to get the information needed to build a DVD drive that handles CSS-protected discs, the DVD CCA has some degree of control over DVD drive manufacturers. I'm sure that the licensing contract prevents manufacturers from selling consumer-priced unrestricted hardware (i.e. drives that happily ignore region coding and can do bit-for-bit writes). Currently, all such drives are well out of the price range of the casual consumer (though not the serious, professional pirate).
But now, thanks to DeCSS, all the information needed to build a DVD drive is out in the open. Anyone who wants to can put together an unrestricted drive and sell it for $200, if that price point is profitable for them.
The DVD CCA faced a tough choice when it came to CSS. They had to choose whether to patent the system (assuming this was possible - probably, giving the current state of patents), or keep it a trade secret. Both choices have advantages and disadvantages:
The CCA chose the second option, and it has backfired on them. Now they're trying to save themselves via the court system. This is why the reverse engineering issue is probably far more important to them than the DMCA issue. If they succeed in nailing Johansen, they'll probably have frightened off anyone who was thinking of reverse-engineering the system themselves (or using the DeCSS-derived information now on the web) in order to build a player.
Note that if this happens, the CCA will have effectively aquired the same rights as if they'd patented the system - but without ever formally disclosing how the system works, as a patent would normally require. So this case could set a rather dangerous precedent: The whole point of patents is to enhance technological development, by encouraging disclosure. And the carrot used to encourage disclosure is a government-guaranteed monopoly for a limited period of time. Putting reverse engineering on legally shaky ground means that companies have a better chance of keeping a monopoly on a technology that has trade secret status.
Right now, when companies are deciding whether or not to patent something, they have to ask themselves the question: "How long before someone will be able to re-create this technology without spying on us?" If the answer is over a certain threshold, it's probably better to keep it a trade secret. If reverse engineering becomes de-facto illegal, then the question becomes "How long before someone will be able to re-create this technology without spying on us or reverse-engineering our product?" Obviously, the answer to the second question will often be greater (and never less) than the answer to the first, and thus is more likely to be over the magic threshold where patenting becomes a bad idea.
And when you consider that the question isn't really whether or not someone resorted to spying (or reverse engineering), but rather whether or not you can convince a court of this, it's even worse - making a case for reverse engineering is probably a good bit easier, if the judge doesn't understand technology well enough to understand what reverse engineering is all about. So we may even see companies trying to convince judges that a competitor illegally reverse-engineered their product when in fact no such thing took place.
Finally, it should also be noted that while this may prevent companies from going for some patents, it won't prevent them from going after most stupid ones, like the Amazon one-click, because the ability to legally reverse engineer doesn't do much to help you figure out most such 'technologies', anyways.
So, if not to inform, what purpose do adds serv? They serv only to create want and need, and unless you're the mental age of a 10-year-old, they won't work on you..
Advertisements no longer teach or inform, they merely reinforce consumerism.
We seem to be having a basic misunderstanding. As I see it, there are 3 possible explanations for the huge amount of advertising that is done:
I think we can agree that the answer isn't really #1 - only a small amount of advertising is justified for the purpose of informing consumers. In the comment I originally responded to, you seemed to be advocating #3 - you said that creating wants and needs only worked on mental-10-year-olds, and that advertising didn't really work. This is where my disagreement lies - I do think that advertising works to create wants and needs on the general public, and not because most members of the general public have a mental age of 10. Or at least, television and radio advertising works this way - mainly because it's so intrusive, compared to a banner or billboard, which someone can easily look away from.
I agree. However, back in the early 20th, advertisments were less instrusive (compared to now).
I do agree that there has been a shift in the common view of marketing in the last 50 years or so. There was a recent article in Newsweek whose existence serves as rather nice evidence of such a shift: The article was all about how Toyota was working to improve its image in the eyes of younger drivers - apparently the median age of Toyota drivers is fairly high. The article talked about how they'd hired a squad of hip young ad designers, about new designs they were coming out with, and so on - various people from Toyota commented, in an official capacity, on this new campaign. What's interesting here is that they are utterly shameless about the fact that they are attempting to manipulate people - they're happy to talk to Newsweek about their strategies for doing this. 50 years ago, I'm guessing that most companies would want to keep quiet about such strategies - letting it be known how consciously they were working to manipulate the consumer would lead to a backlash from consumers who didn't like being manipulated. But today, you can find articles like the Toyota one in any weekly newsmagazine.
(Speaking of commercials, have you noticed how US TV averages between 2.4 and 2.9 minutes vs 1.7 to 2.2 minutes of advertisement time on Canadian TV?)
Well, noticing this this would require that I own a television. ;) I got rid of mine a few months ago. And I never watched Canadian TV even when I had one. However, when in Brussels a year ago, I found it really interesting to watch US imports (specifically, the Simpsons) with all the commercials cut out. A scene would end with a minor cliffhanger, in a way that was obviously meant to be followed by a commercial. And then it would cut immediately to the following scene, with the cliffhanger's resolution.
I'm afraid you're being incredibly optimistic here. Can you realistically say that most advertising exists to inform consumers of a products' existence? I don't think so. How many people do you think are unaware of the existence of Coca-Cola? If they were to cut their advertising budget in half, would there be a significant drop in their brand recognition? I doubt it.
Only the momentum of the "consumerism" of the 1950s through 1980s keeps people advertising in this day and age.
In his book Democracy In America, Alexis De Toqueville commented on how pervasive he found commercial advertising to be in the US. He was writing in first part of the 19th century. Advertising has been with us since long before the 1950s.
I can't wait for it to die.
I can, but only because I'm afraid I don't have much choice about it.
This sort of sarcasm can cut both ways: The MPAA can just as easily be made to look like a pack of crybabies.
The idea that universities install dorm-wide networks to improve education is laughable. They do it because they're competing for students who want such a connection - and who mostly want it for reasons that aren't particularly educational in nature.
That's why you give users an account, just like they have at Blockbuster now. When someone wants to rent something, they go to the store's web page, log in to their account, browse through available titles, order the ones they want. The store checks the stock of pre-burned CDs, and if none of the requested title are available, notifies attendants to burn new copies. Later in the day, on the way home from work, they stop by to pay and pick up the discs.
There's another issue here: You've probably got high school students paid minimum wage operating this thing. They're going to be taking home free copies of every movie in the store, unless you have some good controls and tracking built into the hardware doing the burning.
2) Unless you charge the customer who has a copy of a DVD for the length of time they have the DVD (i.e., 1day=$2, 2days=$3, 3days=$4 etc.) there would be no incentive for them to ever bring it back. You would be basically selling DVDs at a very low price.
Okay, so giving them the discs indefinitely is a problem. Require them to return the discs within, say, a month. If they don't, you're going to put a hefty charge (say, twice what they'd pay to buy each movie outright) on their credit card - which you will happily refund, once they do actually return the discs.
It should be noted that NPR recieves little or no direct government funding. NPR is a content provider - they don't operate any radio stations. Instead, radio stations pay them to broadcast their content. Now, many of these radio stations are public (and therefore recieve government funds), so it's not as if they're completely untainted. But like yourself, I find that they generally give balanced coverage of issues - certainly moreso than most strictly-privately-funded media.
Rubbish. Their focus on X86 is based on their desire to make a buck. Additional code morphing software is simply not required because the market isn't there to justify it.
It's quite possible that they're releasing the current Crusoe and hyping the low power requirements because it's a way for them to go ahead and start making some revenue to fund future development. There's no reason to assume that Crusoe is the ultimate application they've visualized for the technology.
You could only justify the PowerPC emulation if you were extremely generous.
One of the beauties of the software-based approach is that Transmeta need not develop emulation themselves for every IS that gets supported. They could license the required technical info (VLIW layer, etc.) to Apple under NDA, and let Apple develop the PPC morphing layer. So this way they both make money from the licensing, and from selling chips to Apple. Generosity need not enter into it.
Uh, this is a feature of a lot of VM systems. What do you think a Linux swap partition, or the Windows swapfile, are for?
But yeah, System 7 had a pretty lousy VM implementation, from what I've read. If I understand correctly (and I could be way off), the third-party utility 'RamDoubler' was basically a replacement for the OS's built-in VM system.
I'm aware of how FX!32 works - I went and looked it up after I wrote that. But I don't retract my point: it's still a strictly software solution, and the hardware provides no support for the process. Presumably, a considerable amount of software infrastructure is needed to make it work - x86 binaries have to be recognized and translated, and those translated binaries stored, for example. OTOH, unless you're trying to handle multiple instruction sets at once, there's no reason why Crusoe should require any more 'help' from the OS than an ordinary x86. And even if it does start handling multiple instruction sets, the solution is considerably simpler and more elegant than FX!32.
Also, FX!32's performance target is considerably less ambitious than that of Crusoe: The FX!32 white paper says they're shooting for code with 70% of the performance of a native-compiled Alpha application.
The question is: Do the technical improvements you mention result in observable improvements in areas that are important to the customer? The user cares how fast applications run, not how many integer execution units his CPU has. If the Athlon's technical advantages don't translate into significant speed improvements (say, because the CPU spends so much time waiting on other system components anyways), then they aren't significant in the marketplace.
On the other hand, Crusoe's technical advantages do have the potential to improve system performance in areas visible and important to the user: lower cost, and either greater battery life or lower battery weight. And the low temperature has a potential to lead to user-visible improvements, as well.
This is a far better use for Beam than actually drinking it. Now, had you suggested using Maker's Mark in this way, I'd think you were nuts.
AFAIK, all the examples you name are software emulation. Sure, FX/32 allows an Alpha to emulate an x86 at speeds comparable to x86 native hardware. But an Alpha capable of doing this is much more expensive than the x86 to which it compares, and the program still executes at only around 70% of the speed of native-compiled Alpha code.
If Crusoe (or its descendants) allow you to execute an instruction set at 90+% of the speed of a CPU that is hardwired for that set, when the Crusoe and the hardwired CPU in question are the same price, then this could indeed be a paradigm shift.
I don't see too many people wanting to run multiple OSs at once - it's an absurd waste of resources. But I do see people using this technology to build emulators that compare to native hardware. Consider what might happen to SoftWindows, on MacOS, if this chip were able to run PPC code at a good pace. In that case, one could write a SoftWindows that ran x86 apps as fast as if they were on x86 hardware. However, the trick is that you need OS support to really do this right.
What I'm envisioning is that there's a register somewhere in the chip that indicates what instruction set it's currently using. There's also an instruction to atomically set both this register and the program counter register at the same time. Then, the OS stores a value for this register (along with the program counter and various other info that it already maintains) local to each process. When a process is dispatched, that register gets set simultaneously with the program counter, using the instruction I mentioned. Of course, this means your scheduler needs to be written in the VLIW instruction set (since this instruction won't exist in an emulated instruction set). But basically, any time a context switch is required to get from one chunk of code to another, they can have different instruction sets. (Note also that hardware interrupt vectors will have to contain a value for this register as well as a memory address.)
So does the VLIW substrate have an sort of intrinsic endianness? If it is little-endian in nature like the x86 that it's so good at emulating, then this could result in a noticable performance hit when emulating a big-endian CPU like the 68k or PPC.
Are you sure about this? How much of Linus' motivation for writing Linux was to fill his need for a decent free Unix-ish system on x86 hardware? And how much of it was simply the fact that he derived a sort of playful enjoyment from it?
More 'important' new developments than we often realize arise from play. The guy attempting this port may not end up generating anything useful. But at the very least, he'll learn a good deal in the process, and be able to apply that knowledge on more genuinely useful projects involving Linux or the N64, or other software or hardware entirely. And maybe he'll generate a neat toy that others can pick up, play with, and learn from.
Not really. The speed is plenty acceptable to me, and certainly no worse than any other web-based e-mail I've tried. the accounts are inundated by AOL-level amounts of spam
Evidence? In the year I've been using it, I've only ever gotten one piece of spam in my hotmail account, and I'm pretty sure that this was a result of me providing my e-mail address to a website that didn't deserve it. The only e-mail address where I see significant spam is the one I use when I post to Usenet. But many people sign up for the account when they first begin to use the internet, and don't want the trouble of migrating even after they acquire cluefullness.
I was clueful well before I started using Hotmail - I was using the net and e-mail from a shell account well before 'the Internet' became synonymous with 'the web', and certainly well before Hotmail came into existence. But when I got out of school, I decided I wanted an account for my personal e-mail that I could get at from pretty much anywhere (this means web-accessible: firewalls often block telnet, and the 'batch transfer' nature of HTTP (all my editing happens on the local machine, rather than interactively) makes it far more usable on a high-latency connection in the event that, say, I'm overseas). And I wanted an address that wouldn't go away if I changed jobs or ISPs. Hotmail fit the bill.
In short: Because he doesn't really respect slashdot or the net at all. He is fundamentally authoritarian in outlook, but posting to a forum that is dominated by people who oppose authority - or at least, unearned authority.
The long version:
Most of the people who participate in these discussions have probably read ESR's "The Cathedral and the Bazaar". And most think that Bazaars are pretty cool, and that Cathedrals are kinda lame. But Katz seems like a guy who would be more at home in an old-media Cathedral than a new-media Bazaar. And so he attempts to treat Slashdot like an old-media Cathedral, rather than the new-media Bazaar that it is (mostly) intended to be. He posts his essays as if they were wisdom handed down from upon high, and rarely if ever participates in the discussions that ensue (or if he does, it's as an AC or under a pseudonym). When he addresses opponents, it's not in the comments, on equal terms. Instead he quotes their e-mail in his column, where he has complete control over what is printed, and can simply cut out a point he can't answer. He also writes articles with titles like 'Voices from the Hellmouth', which contain lots of quotes from his e-mail. Seeming to suggest that he is somehow making these voices heard, when they otherwise wouldn't - nevermind that most of the people who e-mailed him could have easily posted their comments directly instead.
On the other hand, in recent months we've seen people like Orson Scott Card and John Carmack posting comments to slashdot. If you'd told me at age 15, right after I read Ender's Game for the first time, that I would one day be able to read and participate in a forum that OSC apparently sometimes read and participated in, I'd've fallen off my chair. Substitute "17", "played Doom", and "John Carmack" in the appropriate places in the above sentence, and it still holds true. These are guys whose authority comes from excellence - Ender's Game and the Doom and Quake engines were extraordinary achievements, far more significant than anything Katz is ever likely to do. But they are quite willing to step out from behind the pulpit and interact with random people in a forum where pretty much anyone with a net connection can come and participate. These are people who seem to understand new media, and aren't afraid of it.
Katz, on the other hand, is afraid of new media. (And new technology in general: Note the Luddite-ism that runs through some of his essays.) Because he apparently wants authority, but he's not really good enough at anything to aquire the authority that comes from excellence. So instead he seeks the authority that comes from possessing a voice on a limited-access medium, the same kind of authority that a talking head on the television has.
And so the resentment that many Slashdotters feel towards Katz is actually quite akin to the resentment they feel towards a mainstream news source that they believe reports poorly on some issue important to him: They feel that he has authority that he does not deserve. Evaluated honestly, Katz's essays are no more informative, insightful or well-written than the best user comments on Slashdot - often, they're less so. They don't deserve to be elevated to a position of greater prominence than those comments.
I remember watching a friend play it briefly. At one point, he opened a door and saw this weird flying creature he'd never seen before. He yelped in surprise and terror. This was not a person who frequently yelped, or felt surprise and terror. Like Half-Life, and unlike Quake, Marathon demonstrates that what really makes a game succeed is good play, design, and plotting. Great graphics technology won't wrap your mind up in the world of the game the way that a good treatment of these other elements will.
You are correct. To be precise, what happens is that the view plane moves vertically up or down, rather than rotating around the center of projection. IIRC, Duke Nukem 3D used a similar approach.
This has been done. I forget all the details, but Daniel Dennett writes about it in Consciousness Explained. Essentially the system involved cameras mounted on glasses, and a grid of "tinglers" strapped to the subjects' abdomen or back. With some practice, they could learn to interpret the impulses from the grid as 'sight'. A few interesting notes:
Long term, this approach is unlikely to produce anything useful, partly because there's no place on the skin where nerves are packed densely enough that you can use a tingler grid with sufficient resolution to be useful, and partly because of the slow response time (OTOH, babies must 'learn to see' as well, and this doesn't happen overnight - maybe with time they'd adapt well enough that the response time was reasonable).