SDMI has nothing to do with musicians. Probably 8 out of ten of them oppose it. SDMI exists solely for the major record companies. MP3 has major record companies scared stiff-- not only because of piracy, but more importantly because it allows bands & independent labels to cheaply & easily distribute their music worldwide without the record companies getting their cut.
Isn't it possible to take the SDMI technology and use it against the recording industry? Technology is supposed to be double-edge sword. If the purpose of SDMI is to screw the consumer base why could it not be turned against the RIAA?After all SDMI has one advantage over MP3: it sets up formal procedures for making payments. Consider the following scenario:
1. Plans for SDMI-compliant hardware are ditched, the specifcation becomes a wire-protocol
2. Restrictions on copying are removed
3. The protocol is used by independent artists to sell their work over the web
The RIAA is out of the loop, the useful parts of SDMI techology (payment for content) is salvaged and everyone is happy.
The transaction between the client and server includes not only the standard CDDB foo (number of tracks, total length, byte offsets for the beginning of each track) but also has the first 8k or so of raw data for each track.
Strange. At least in the interets of saving bandwidth they could have used a cryptographic hash instead of sending the actual data. This would suggest a challenge-response protocol when the server randomly picks short frames from the CD and the client responds with their SHA-1 hashes. Only problem would be the impractical storage requirements because the server would need access to the uncompressed tracks. This is probably why they will only use a fixed portion of the CD instead of generating random challenges.
How come VA linx's share price is in the hundreds and compaq at 20 something?
This is a non-issue. The absolute price of the stock has nothing to do with valuation. For one thing stocks split, so you could have had half as many shares of Compaq available at twice the price for example.
When people refer to IPO frenzy, "bubbles" about to burst etc. they are referring to market capitalization which is roughly number of shares multiplied by the price of each. New companies with no quarterly earnings (and in some cases such as Amazon, plans to operate in the red for the foreseeable future) have capitalization in the billions of dollars. This is what the orthodox Wall Street logic considers ridiculous.
From this perspective VA Linux and RedHat are clearly overpriced stocks. On the other hand the high-tech stock market has been operating in a very different model. Natural sciences dictate that you cannot create energy or mass-- only exchange between systems. In the stock market so far everyone has been collectively getting richer as if money is being pumped into the system out of nowhere. This is another good reason why people suspect that the bubble will deflate eventually-- it will not burst, we will not have another "Black Monday" but it will all come down.
The packed-up nightly builds which include all of the.sos for browser (with debugging symbols!), mail, news, etc, plus a lot of images, scripts, and demo pages, are about a 5MB download.
People need some clues here. Embedded systems have resource constraints and the highest demands are placed on memory use. The memory footprint (code+data) is going to become the bottleneck.
The original poster is wrong because it is the compiled binary that executes. The source footprint is meaningless and typically grossly overestimates the generated code size. The poster quoted above is also wrong because download size underestimates the actual memory footprint; the distribution is compressed to around 50% of the executable format.
Finally all of these are only static measures of the code size and they make no indication of how much memory will be used once the program executes: eg size of data structures and allocation patterns.
Keiter and Rubin did some interesting work in this direction. Their idea is that instead of using specific remailer, users join a network called "crowd" and each client in effect becomes a mailer. When a message is received the user tosses a coin to decide between forwarding to another link or mailing out to the ultimate destination.
Based on the number of hosts colluding, the system offers varying degrees of anonymity. (Its all worked out mathematically, eg the probability that so-and-so is the originator of the message given the observed traffic.) You can find the paper here:
The first link points to Anonymizer. That site requires signup for decent service but if you are patient enough the site allows anyone to load pages anonymously, except that a frustrating delay is imposed deliberately to "encourage" users to pay the $50/year.
Unfortunately there are several problems with the Anonymizer approach. To hide the origin of the HTTP request, active content including Javascript and applets must be disabled. (Otherwise bad things happen: for example using the object-model script could force another HTTP request to server revealing the original IP) This means that pages which have dependency on script will not work. It is entirely another story whether designing one's site to require Javascript for proper functioning was such brilliant idea and surely Jakob Nielsen will have plenty to comment on this. But the reality is that especially among the recent breed of ecommerce sites have all the latest DHTML incorporated and script is crucial there.
More importantly cookies are not exchanged meaning that shopping carts will not work. Technically you could browse but you could not buy-- moot point since there are other privacy problems in electronic transactions. (Shipping address, credit cards, etc.) Neat feature would have been to create a temporary cookie jar used per session and discarded.
Paradoxically this service creates a problem of different type. Because the number of registered users is small, web servers getting requests from Anonymizer.com will assume that it is coming from one of the handful of registered users:) In other words previously you could be one of the millions on the web-- but thanks to your proxy you are now identified within a few thousand people. One could only hope that the company can manage to keep its user database private.
The difference in credibility between your examples [#1-#3] is only in the perception. Assuming that you have no background information on the person, how does the presence of "real" name--whatever that means-- increase your confidence?
You argue that a pseudonym only establishes a persistent notion of identity; eg that all postings made by "Foobar" are correlated to the same author. In what way is this different from the use of an actual name? If you have never heard of a person named "Aaron Kostrawitzki" what is the rational basis for attributing more credibility to a posting under that name?
The fact that pseudonyms could be co-opted unintentionally or fraudulently? Sure but in the absence of authentication protocols you could not verify whether a given was indeed made by the person whose name is purportedly associated with that comment.
What about Visual Basic, Visual C++ and Visual J++, to name a few?
Among these Visual Basic is the only example. Visual C++ and Visual J++ are products they are not languages.
Among other things VC contains a fairly standards-compliant compiler. (At least for its C language subset; since C++ was still in draft-standard at the time of release it is harder to judge) Granted it does have proprietary extensions but so does every other compiler including g++. Same goes for Visual J++ except that Sun has been very aggresive about prosecuting-out any extensions there. As for the case of Visual Basic: who cares?
Java is unique in the sense that its specification is closed-- revisions, derivatives or descendant languages are not allowed. For reference it is not the only language developed in commercial environment. C was designed at Bell Labs, and so was C++. SmallTalk came out of Xerox Parc. How many people did AT&T and Xerox go after for implementing compilers with different language semantics or proprietary extensions? I am sorry but criticizing Microsoft and then being apologetic about Sun's behavior is sheer hypochrisy.
They are also not trying to protect artists. How do record labels make their money? By exploiting artists, not protecting them. Now all of a sudden there is a way for budding artists to distribute their music on a large scale WITHOUT GOING THROUGH THE ESTABLISHED SYSTEM. This is the threat that the RIAA is attempting to fight off. The problem with MP3 is not piracy, it is the fact that it allows artists to distribute music and deny the record labels their cut.
This raises an interesting question though. RIAA is fighting the battle on two fronts. First one is suppressing the creation and distribution of technology threatening their revenue stream. Second one is developing alternative technologies and specifications to maintain its monopoly over content distribution. (This is RIAA on the offensive)
The question is, could SDMI lead to the downfall of RIAA?
Since distribution costs are much cheaper for electronic media, why couldnt the Independent labels utilize SDMI for selling their own music? In the long run neither MP3 or any other format will help artists if there is no way of collecting revenue. At some point it is necessary for the publisher to make money. The solution? SDMI is squarely aimed at solving this problem! As long as questions of policy are separated, the use of SDMI need not imply fascist restrictions. (Even though this is what the designers were going to push for.) For example songs could be given away free with unlimited distribution and copying rights. Others may require one-time download fee or pay-per-access charges enforced by the SDMI spec.
Here then is the other side of the coin: on Slashdot the point has been made that tools built for perfectly legitimate "innocent" purposes (eg Napster) could be adapted for questionable ends.
This is the other side of the coin then. Perhaps in SDMI we have the overlooked case of a technology originally developed by an oligopoly for highly questionable purposes-- namely carrying the oligopoly into digital distribution. It would be great irony if the same tool could be coopted by Indies to publish music on the net bypassing the RIAA channel completely.
If their marketing and communications were as good as some of their technology then we'd have MCPs sitting on street corners with cardboard signs offering to reboot computers for food.
Really?
Sanity check: how much does Sun hardware cost? For the price of highend PC you could probably buy about as much as one of the disk drives in your beloved Sun workstation. Face it folks, the joke is over. Once Merced starts pushing the envelope of highend servers/workstation at small fraction of the price, Sun will look like the emperor without clothes.
Among other things Sun created Java, not the dancing paperclip or the ten minute uptime. Take your pick but I know which camp I'm in.
Sanity check #2: Java is the language which makes 4-processor Digital Alphas act like memory-starved 286s. The language which enables tasteless web designers to embed sophomoric jumping-ball applets on web pages, which have the distinction of being even more stupid than the paperclip assistant. Questions such as "which camp are you on?" sounds eerily reminiscent of the Salem witchhunts.
Hint: Sun has no vested interest in making cross-platform software work. Their major revenue stream comes from milking wealthy corporate clients with their overpriced bloated Enterprise servers. Anytime Sun succeeded in having anything work cross-platform they realized that it would be shooting their own product line in the foot. Despite all "corporate spin" to the contrary Sun remains primarily a hardware company. The recent switch in the emphasis on Java from client technology to enterprise (eg from silly applets to EJB) is proof of this: Sun is banking on PC workstations to run thin Java clients so they can supply the backend of Enterprise servers with 6-digit pricetags.
Give their PR some credit. On second looks they are not so naive after all. Well aware that there is no public "glory" in producing hardware-- yet there is ample money-- the SUN marketing line is liberally sprinkled with references to the network computer, Java and open standards. (Some suckers are taken in no doubt.)
Finally going back to the subject of this discussion, SCSL remains a sad compromise between proprietary ownership and open source. The fact that Sun has appropriated Blackdown code ought to convince anyone who had been apologetic about the shortcoming of SCSL.
Slowly closed source software is shooting itself in the foot with all these "trojans" that they add, but that they "don't use for any purpose". They'll soon not be able to use the security through obscurity catch phrase.
This is missing the point completely.
Sad truth is that nobody has ever made money trying to protect privacy. There is no economic incentive for developing safeguards against the abuse of information. Not even open-source software could change that. Any software with privacy enhancing features would be at a disadvantage. In fact the response from websites would be undisguised hostility. This is a zero-sum game where your loss of privacy is the sites gain in focused and targetted advertising. (solving the "dont-sell- diapers-to-bachelors" problem) A web server could conceivably discriminate against clients using privacy-protecting software: assigning lower priority, degrade the level of service or completely ignore requests. You are costing them in lost revenues; how dare you expect getting decent web pages?
There is some short-lived favorable publicity in taking the higher ground or paying tribute to the sacred cow of respecting the consumer privacy. There are press releases written in religious overtones, conferences where the CEO criticizes "other" companies who have not pledged loyalty to the cause of privacy and proud declarations on the web site stamped with the eTrust logo. After the dust settles everything is back to usual: consumer privacy is subordinates to the overriding need for advertising revenues and effective marketing strategies. (Anyone holding their breath for the *next* Real Networks scandal?)
Given that there is so much to gain by collecting and distributing user information (and very little to lose unless you are caught red handed) it is not surprising that applicaitons will evolve in this direction. Consumers are uninformed. The law is extremely liberal, unlike Europe where privacy is taken very seriously and violations are prosecuted aggresively. Isolating the breach of privacy into one spot and pointing the finger at one company alone is very difficult when there are so many companies competing for that purpose.
Companies are not even trying to be apologetic about their behavior. Betting on consumer ignorance and the elusive promise of "customized content" (eg the more we know about you the more enjoyable your web experience will be) they forge ahead completely ignoring privacy concerns unless these have substantial impact to the bottom line-- eg loss of money.
Given the odds open-source is no silver bullet. By virtue of being open-source it is even easier to fork and build an "improved" version with even more flagrant privacy violations. The realistic viewpoint is that the future will see more blatant privacy violations happening more frequently.
Katz is beating a dead horse as usual. First there were the two articles written in response to the Columbine school shootings. Series of Slashdot postings followed endless repeating the refrain about unfairly treated geeks. Finally all this polemic and demagogue is culminating in the "Geeks" book that he is supposedly writing, at least according to an Utne Reader article. (Admittedly it is a major milestone for publishing an article about the misconceptions of Goth culture in the so-called "alternative" media.)
This whining and brow-beating has assumed larger-than-life proportions. As the poster points out the FBI does not screen for geeks but psychologically disturbed individuals. And finally there are acceptable purposes for profiling. These procedures have very strong statistical validity, even if they have disturbingly high rates of false positives-- this only calls for improvements in the method not rejection of the idea altogether.
It is possible rephrase Katz's objection in terms of the testing criteria: according to JonKatz a person categorized as "geek" will be falsely identified as potential trouble-maker according to the criteria. Yet nowhere does he look into the question of whether the individiuals that would be considered "pyschologically distrubed" are correctly labelled as such. The condemening criticism of profiling would be that some dangerous people slip through the cracks in the decision criteria-- in other words false negatives. Clearly one way of avoiding this is to label anyone as potentially dangerous based on the presence of very few "signature" personality traits. Logically this will also increase the problem of false positives and perhaps geeks are disproportionately affected by that. Still this is a far stretch from arguing that profiling is conceptually flawed.
It is sad that this type of article causes people to respond with emotion instead of reason.
While we're on the topic of open mathematical conjectures, my favorite still has to be Goldbach's Conjecture. It's tantalizingly simple; it states that any even integer greater than 4 can be expressed as the sum of two prime numbers. It seems
One (trivial) correction: greater than or equal to 4, since 4 = 2+2 can be written as the sum of two primes as well.
A lot of famous mathematicians have tried their hand at this problem, with no success to date. The first one was Euler: in fact this problem was stated by Christian Golbach in a letter to Euler, who apparently believed the conjecture to be correct.
There have been some "close" results. A Russian mathematician circas 1930s proved that every even number can be written as the sum of not more than 300000 (three-hundred-thousand) primes. As William Dunham points in "Journey Through Genius" this proof falls short of the goal slightly, namely by 299998 primes.
Back to the subject of open problems and the STW, it is a much welcome development that the STW has been proved finally. This is because a lot of time has been spent developing algebraic results of the form "If STW then..." FLT is certainly more interesting from a philosophical standpoint but very few results depend on it.
STW on the other hand is very similar to that other great open question, the Riemann hypothesis which factors into many important results. Starting with the 19th century, people used the Riemann hypothesis (and various generalizations) to "prove" results including the density of primes and even efficient algorithms for checking primality. STW being proved false would have some major repercussions, just as the Riemann hypothesis refutation will cause serious trouble.
The point the first poster was making, I think, is that by using a 4-way box, and crowing about their advantage on ludicrously high server loads, MS is aiming at that market. That is, once you step into the realm of 4 processor machines, testing NT vs. Linux is just silly, because who in their right mind would use either one for such hardware? It's like saying that my Cessna is a
And how much would those Sun boxes cost do you think? Today you can get a multiprocessor PC with multiple SCSI drives for under $3K. With that money you have a fighting chance to afford the disk drive for Sun servers. Intel has been pursuing the highend market aggressively-- SMP on Pentiums is most cases much cheaper than buying an ungainly Solaris box. Increasingly the highend machines are being squeezed out of the market into niche applications like scientific computing. To wit: look at the fate of the Digital ALPHA architecture, very solid-engineering that would blow the doors off of anything Sun ever built on benchmarks. During the second half of the 90s it continued to lose share.
In the future expect to see multiprocessor Pentium systems in the enterprise. The CPU technology is up and coming: multiple Pentiums are much more economical and also advantageous from the point of view of scaling. RAID controllers, cheaper SCSI devices etc. are already here. Intel is already trying to put this spin on their PIII chips with the tag line about "enterprise computing"
There is no question that in the future we will see more and more SMP Intel machines (especially with Merced) taking over the work of ungainly and expensive Sun, HP and Alphas. The question is are they going to run NT or Linux? I will hedge my bets on NT.
Well now I am curious to know how you think that NT removes the command prompt from the users actions or how it's really visually different from 95 (Besides the little Name to the left of the start button. I am writing this message on an
Clue #1: NT has an identical GUI with Windows 95. This is completely intentional-- by design to ease the transition from 9X to NT.
NT box). The same icon and the same commands are still in use with good ol' dos. Basically NT is 95 + (a whole lot of extra expensive network functions). Dos may be bad but damn was it
Clue #2: NT is an entirely new operating system. It was developed by David Cutler & friends, former architect of VMS at Digital. The design is influenced by VMS but the underpinnings are very modern: highly portable micro-kernel communicating with user processes by IPC, sitting on top of a hardware abstraction layer Compare this with the archaic 1970s technology of monolithic kernels found in Linux.
On top of the kernel, there is an additional layer implementing the Win32 API. (Called the "Windows Subsystem") But there were also OS/2 and Posix subsystems in NT4 which implemented the corresponding APIs. This additional layer is responsible for the high degree of compatibility with Win9X-- but obviously it is not perfect.
Please try calling 1-800-GET-CLUE before you bash NT next time.
Not true. I implemented a (prototype) secure web email system which simply locked the account for 10 minutes if you got the password wrong 3 times. You could brute force it, but it would take a real long time!
That is no protection against brute-force at all. Any password scheme operating in the clear (eg without shared secrets and without public-key cryptography) is vulnerable to offline attacks. The adversary eavesdrops on the connection and then mounts a dictionary attack offline-- since there are no secrets beside the password its always possible to guess the password and compare against the actual observed traffic. Perhaps some 14-year old amateur would try to brute-force the account by repeated login attempts but its doubtful the serious attacker might do something that silly. As others pointed out, locking out the account because of incorrect login can easily lead to denial-of-service attacks. BP
The better colleges with solid academic-reputation have "need-blind" admission policies. In other words whether you need financial aid (or how much you need it) *can not* affect your chances of admission. In fact for government-funded schools (eg state universities) I believe this is legally required-- they are not allowed to discriminate applicants on the basis of income.
Private schools on the other hand can do this; in other words they are under no obligation to have need-blind policies in place. All good schools however state they are need-blind, claiming their admission does not take economic factors into account.
This is particulary true about the "cream of the crop", the schols being bad-mouthed here as elitist: the Ivy league, MIT, Stanford, etc. These schools have huge endowments and constant flood of money from their wealthy alumni to use for their students. (Did you know that for most schools tution covers only half the total expenses? If it werent for the funds from elsewhere the costs would be exorbitantly high in order to break even.)
One could argue ofcourse that admissions without the aid is useless. But in the majority of cases students who are admitted to these schools also receive the full scholarship they ask for.
Consider my case: I am not even a US citizen yet I attended an Ivy with near full scholarship. That, despite the fact that admission for international students is decidedly *NOT* need-blind. I have seen complete idiots (foreign nationals) pay their way into some of the highly-regarded schools. If anything one has to be grateful for the generosity here. You would be hard pressed to find schools in Europe which will accommodate such a diverse student body from all economic backgrounds with the necessary financial aid.
Going back to the point, yes rich people might have an early start in the admissions race. But personal accomplishments (not necessrily academic) will dwarf that difference, making it insignificant and irrelevant.
The Alpha architecture is partially under the control of Intel, because they acquired some DEC facilities when Compaq was taking over. Last summer around July a friend had a meeting with the friendly folks from Intel. They put it in so many words: "At Intel we build Intel chips."
So you might need to tone down the pipe-dreams about Alpha competing with Intel.
Secondly, MS dropped the ball on Alpha only because *Compaq* announced it would stop supporting NT on the Alpha. First it was announced that 32b support would go, then the brilliant management decided 64b NT would not be supported either. The engineers at DEC-West (the Pacific Northwest facility which used to do all the NT development and credited with the excellent emulation layer FX!32) dont have jobs anymore.
Wonder of wonders Compaq still supports OpenVMS. Apparently they are trying to position the Alpha platform for Tru64, OpenVMS and Linux.
Except for very high-performance applications, Alpha is just not price-competitive. Don't get me wrong: I am very impresed with the Alpha family. At work I have a dual-proc 533mhz Ultimate WKS. Alpha shines at scientific computing and highend. But for things like web servers, you can get 4 or 8-way PIIIs that support clustering, much cheaper than the Alpha system with comparable performance. Once Merced comes out the gap will be closed even more.
The idea of "Alpha at home" will not fly with consumers.
Slashdot Mirror
Isn't it possible to take the SDMI technology and use it against the recording industry? Technology is supposed to be double-edge sword. If the purpose of SDMI is to screw the consumer base why could it not be turned against the RIAA?After all SDMI has one advantage over MP3: it sets up formal procedures for making payments. Consider the following scenario:
1. Plans for SDMI-compliant hardware are ditched, the specifcation becomes a wire-protocol
2. Restrictions on copying are removed
3. The protocol is used by independent artists to sell their work over the web
The RIAA is out of the loop, the useful parts of SDMI techology (payment for content) is salvaged and everyone is happy.
BluesPower
Strange. At least in the interets of saving bandwidth they could have used a cryptographic hash instead of sending the actual data. This would suggest a challenge-response protocol when the server randomly picks short frames from the CD and the client responds with their SHA-1 hashes. Only problem would be the impractical storage requirements because the server would need access to the uncompressed tracks. This is probably why they will only use a fixed portion of the CD instead of generating random challenges.
BluesPower
This is a non-issue. The absolute price of the stock has nothing to do with valuation. For one thing stocks split, so you could have had half as many shares of Compaq available at twice the price for example.
When people refer to IPO frenzy, "bubbles" about to burst etc. they are referring to market capitalization which is roughly number of shares multiplied by the price of each. New companies with no quarterly earnings (and in some cases such as Amazon, plans to operate in the red for the foreseeable future) have capitalization in the billions of dollars. This is what the orthodox Wall Street logic considers ridiculous.
From this perspective VA Linux and RedHat are clearly overpriced stocks. On the other hand the high-tech stock market has been operating in a very different model. Natural sciences dictate that you cannot create energy or mass-- only exchange between systems. In the stock market so far everyone has been collectively getting richer as if money is being pumped into the system out of nowhere. This is another good reason why people suspect that the bubble will deflate eventually-- it will not burst, we will not have another "Black Monday" but it will all come down.
BluesPower
People need some clues here. Embedded systems have resource constraints and the highest demands are placed on memory use. The memory footprint (code+data) is going to become the bottleneck.
The original poster is wrong because it is the compiled binary that executes. The source footprint is meaningless and typically grossly overestimates the generated code size. The poster quoted above is also wrong because download size underestimates the actual memory footprint; the distribution is compressed to around 50% of the executable format.
Finally all of these are only static measures of the code size and they make no indication of how much memory will be used once the program executes: eg size of data structures and allocation patterns.
BluesPower
Keiter and Rubin did some interesting work in this direction. Their idea is that instead of using specific remailer, users join a network called "crowd" and each client in effect becomes a mailer. When a message is received the user tosses a coin to decide between forwarding to another link or mailing out to the ultimate destination.
Based on the number of hosts colluding, the system offers varying degrees of anonymity. (Its all worked out mathematically, eg the probability that so-and-so is the originator of the message given the observed traffic.) You can find the paper here:
Crowds: anonymity for Web transactions
BluesPower
The first link points to Anonymizer.
:)
That site requires signup for decent service but if you are patient enough the site allows anyone to load pages anonymously, except that a frustrating delay is imposed deliberately to "encourage" users to pay the $50/year.
Unfortunately there are several problems with the Anonymizer approach. To hide the origin of the HTTP request, active content including Javascript and applets must be disabled. (Otherwise bad things happen: for example using the object-model script could force another HTTP request to server revealing the original IP)
This means that pages which have dependency on script will not work. It is entirely another story whether designing one's site to require Javascript for proper functioning was such brilliant idea and surely Jakob Nielsen will have plenty to comment on this. But the reality is that especially among the recent breed of ecommerce sites have all the latest DHTML incorporated and script is crucial there.
More importantly cookies are not exchanged meaning that shopping carts will not work. Technically you could browse but you could not buy-- moot point since there are other privacy problems in electronic transactions. (Shipping address, credit cards, etc.) Neat feature would have been to create a temporary cookie jar used per session and discarded.
Paradoxically this service creates a problem of different type. Because the number of registered users is small, web servers getting requests from Anonymizer.com will assume that it is coming from one of the handful of registered users
In other words previously you could be one of the millions on the web-- but thanks to your proxy you are now identified within a few thousand people. One could only hope that the company can manage to keep its user database private.
BluesPower
The difference in credibility between your examples [#1-#3] is only in the perception. Assuming that you have no background information on the person, how does the presence of "real" name--whatever that means-- increase your confidence?
You argue that a pseudonym only establishes a persistent notion of identity; eg that all postings made by "Foobar" are correlated to the same author. In what way is this different from the use of an actual name? If you have never heard of a person named "Aaron Kostrawitzki" what is the rational basis for attributing more credibility to a posting under that name?
The fact that pseudonyms could be co-opted unintentionally or fraudulently? Sure but in the absence of authentication protocols you could not verify whether a given was indeed made by the person whose name is purportedly associated with that comment.
BluesPower
Among these Visual Basic is the only example. Visual C++ and Visual J++ are products they are not languages.
Among other things VC contains a fairly standards-compliant compiler. (At least for its C language subset; since C++ was still in draft-standard at the time of release it is harder to judge) Granted it does have proprietary extensions but so does every other compiler including g++. Same goes for Visual J++ except that Sun has been very aggresive about prosecuting-out any extensions there. As for the case of Visual Basic: who cares?
Java is unique in the sense that its specification is closed-- revisions, derivatives or descendant languages are not allowed. For reference it is not the only language developed in commercial environment. C was designed at Bell Labs, and so was C++. SmallTalk came out of Xerox Parc. How many people did AT&T and Xerox go after for implementing compilers with different language semantics or proprietary extensions? I am sorry but criticizing Microsoft and then being apologetic about Sun's behavior is sheer hypochrisy.
BluesPower
This raises an interesting question though. RIAA is fighting the battle on two fronts. First one is suppressing the creation and distribution of technology threatening their revenue stream. Second one is developing alternative technologies and specifications to maintain its monopoly over content distribution. (This is RIAA on the offensive)
The question is, could SDMI lead to the downfall of RIAA?
Since distribution costs are much cheaper for electronic media, why couldnt the Independent labels utilize SDMI for selling their own music? In the long run neither MP3 or any other format will help artists if there is no way of collecting revenue. At some point it is necessary for the publisher to make money. The solution? SDMI is squarely aimed at solving this problem! As long as questions of policy are separated, the use of SDMI need not imply fascist restrictions. (Even though this is what the designers were going to push for.) For example songs could be given away free with unlimited distribution and copying rights. Others may require one-time download fee or pay-per-access charges enforced by the SDMI spec.
Here then is the other side of the coin: on Slashdot the point has been made that tools built for perfectly legitimate "innocent" purposes (eg Napster) could be adapted for questionable ends.
This is the other side of the coin then. Perhaps in SDMI we have the overlooked case of a technology originally developed by an oligopoly for highly questionable purposes-- namely carrying the oligopoly into digital distribution. It would be great irony if the same tool could be coopted by Indies to publish music on the net bypassing the RIAA channel completely.
BluesPower
Really?
Sanity check: how much does Sun hardware cost? For the price of highend PC you could probably buy about as much as one of the disk drives in your beloved Sun workstation. Face it folks, the joke is over. Once Merced starts pushing the envelope of highend servers/workstation at small fraction of the price, Sun will look like the emperor without clothes.
Among other things Sun created Java, not the dancing paperclip or the ten minute uptime. Take your pick but I know which camp I'm in.
Sanity check #2: Java is the language which makes 4-processor Digital Alphas act like memory-starved 286s. The language which enables tasteless web designers to embed sophomoric jumping-ball applets on web pages, which have the distinction of being even more stupid than the paperclip assistant. Questions such as "which camp are you on?" sounds eerily reminiscent of the Salem witchhunts.
Hint: Sun has no vested interest in making cross-platform software work. Their major revenue stream comes from milking wealthy corporate clients with their overpriced bloated Enterprise servers. Anytime Sun succeeded in having anything work cross-platform they realized that it would be shooting their own product line in the foot. Despite all "corporate spin" to the contrary Sun remains primarily a hardware company. The recent switch in the emphasis on Java from client technology to enterprise (eg from silly applets to EJB) is proof of this: Sun is banking on PC workstations to run thin Java clients so they can supply the backend of Enterprise servers with 6-digit pricetags.
Give their PR some credit. On second looks they are not so naive after all. Well aware that there is no public "glory" in producing hardware-- yet there is ample money-- the SUN marketing line is liberally sprinkled with references to the network computer, Java and open standards. (Some suckers are taken in no doubt.)
Finally going back to the subject of this discussion, SCSL remains a sad compromise between proprietary ownership and open source. The fact that Sun has appropriated Blackdown code ought to convince anyone who had been apologetic about the shortcoming of SCSL.
BluesPower
This is missing the point completely.
Sad truth is that nobody has ever made money trying to protect privacy. There is no economic incentive for developing safeguards against the abuse of information. Not even open-source software could change that. Any software with privacy enhancing features would be at a disadvantage. In fact the response from websites would be undisguised hostility. This is a zero-sum game where your loss of privacy is the sites gain in focused and targetted advertising. (solving the "dont-sell- diapers-to-bachelors" problem) A web server could conceivably discriminate against clients using privacy-protecting software: assigning lower priority, degrade the level of service or completely ignore requests. You are costing them in lost revenues; how dare you expect getting decent web pages?
There is some short-lived favorable publicity in taking the higher ground or paying tribute to the sacred cow of respecting the consumer privacy. There are press releases written in religious overtones, conferences where the CEO criticizes "other" companies who have not pledged loyalty to the cause of privacy and proud declarations on the web site stamped with the eTrust logo. After the dust settles everything is back to usual: consumer privacy is subordinates to the overriding need for advertising revenues and effective marketing strategies. (Anyone holding their breath for the *next* Real Networks scandal?)
Given that there is so much to gain by collecting and distributing user information (and very little to lose unless you are caught red handed) it is not surprising that applicaitons will evolve in this direction. Consumers are uninformed. The law is extremely liberal, unlike Europe where privacy is taken very seriously and violations are prosecuted aggresively. Isolating the breach of privacy into one spot and pointing the finger at one company alone is very difficult when there are so many companies competing for that purpose.
Companies are not even trying to be apologetic about their behavior. Betting on consumer ignorance and the elusive promise of "customized content" (eg the more we know about you the more enjoyable your web experience will be) they forge ahead completely ignoring privacy concerns unless these have substantial impact to the bottom line-- eg loss of money.
Given the odds open-source is no silver bullet. By virtue of being open-source it is even easier to fork and build an "improved" version with even more flagrant privacy violations. The realistic viewpoint is that the future will see more blatant privacy violations happening more frequently.
BP
Completely agreed.
Katz is beating a dead horse as usual.
First there were the two articles written in response to the Columbine school shootings. Series of Slashdot postings followed endless repeating the refrain about unfairly treated geeks. Finally all this polemic and demagogue is culminating in the "Geeks" book that he is supposedly writing, at least according to an Utne Reader article. (Admittedly it is a major milestone for publishing an article about the misconceptions of Goth culture in the so-called "alternative" media.)
This whining and brow-beating has assumed larger-than-life proportions.
As the poster points out the FBI does not screen for geeks but psychologically disturbed individuals.
And finally there are acceptable purposes for profiling. These procedures have very strong statistical validity, even if they have disturbingly high rates of false positives-- this only calls for improvements in the method not rejection of the idea altogether.
It is possible rephrase Katz's objection in terms of the testing criteria: according to JonKatz a person categorized as "geek" will be falsely identified as potential trouble-maker according to the criteria. Yet nowhere does he look into the question of whether the individiuals that would be considered "pyschologically distrubed" are correctly labelled as such. The condemening criticism of profiling would be that some dangerous people slip through the cracks in the decision criteria-- in other words false negatives.
Clearly one way of avoiding this is to label anyone as potentially dangerous based on the presence of very few "signature" personality traits. Logically this will also increase the problem of false positives and perhaps geeks are disproportionately affected by that.
Still this is a far stretch from arguing that profiling is conceptually flawed.
It is sad that this type of article causes people to respond with emotion instead of reason.
BP
One (trivial) correction: greater than or equal to 4, since 4 = 2+2 can be written as the sum of two primes as well.
A lot of famous mathematicians have tried their hand at this problem, with no success to date. The first one was Euler: in fact this problem was stated by Christian Golbach in a letter to Euler, who apparently believed the conjecture to be correct.
There have been some "close" results. A Russian mathematician circas 1930s proved that every even number can be written as the sum of not more than 300000 (three-hundred-thousand) primes. As William Dunham points in "Journey Through Genius" this proof falls short of the goal slightly, namely by 299998 primes.
Back to the subject of open problems and the STW, it is a much welcome development that the STW has been proved finally. This is because a lot of time has been spent developing algebraic results of the form "If STW then..." FLT is certainly more interesting from a philosophical standpoint but very few results depend on it.
STW on the other hand is very similar to that other great open question, the Riemann hypothesis which factors into many important results. Starting with the 19th century, people used the Riemann hypothesis (and various generalizations) to "prove" results including the density of primes and even efficient algorithms for checking primality. STW being proved false would have some major repercussions, just as the Riemann hypothesis refutation will cause serious trouble.
--bluespower
And how much would those Sun boxes cost do you think? Today you can get a multiprocessor PC with multiple SCSI drives for under $3K. With that money you have a fighting chance to afford the disk drive for Sun servers. Intel has been pursuing the highend market aggressively-- SMP on Pentiums is most cases much cheaper than buying an ungainly Solaris box. Increasingly the highend machines are being squeezed out of the market into niche applications like scientific computing. To wit: look at the fate of the Digital ALPHA architecture, very solid-engineering that would blow the doors off of anything Sun ever built on benchmarks. During the second half of the 90s it continued to lose share.
In the future expect to see multiprocessor Pentium systems in the enterprise. The CPU technology is up and coming: multiple Pentiums are much more economical and also advantageous from the point of view of scaling. RAID controllers, cheaper SCSI devices etc. are already here. Intel is already trying to put this spin on their PIII chips with the tag line about "enterprise computing"
There is no question that in the future we will see more and more SMP Intel machines (especially with Merced) taking over the work of ungainly and expensive Sun, HP and Alphas. The question is are they going to run NT or Linux? I will hedge my bets on NT.
Blues
Clue #1: NT has an identical GUI with Windows 95. This is completely intentional-- by design to ease the transition from 9X to NT.
NT box). The same icon and the same commands are still in use with good ol' dos. Basically NT is 95 + (a whole lot of extra expensive network functions). Dos may be bad but damn was it
Clue #2: NT is an entirely new operating system. It was developed by David Cutler & friends, former architect of VMS at Digital. The design is influenced by VMS but the underpinnings are very modern: highly portable micro-kernel communicating with user processes by IPC, sitting on top of a hardware abstraction layer Compare this with the archaic 1970s technology of monolithic kernels found in Linux.
On top of the kernel, there is an additional layer implementing the Win32 API. (Called the "Windows Subsystem") But there were also OS/2 and Posix subsystems in NT4 which implemented the corresponding APIs. This additional layer is responsible for the high degree of compatibility with Win9X-- but obviously it is not perfect.
Please try calling 1-800-GET-CLUE before you bash NT next time.
bluespower
That is no protection against brute-force at all. Any password scheme operating in the clear (eg without shared secrets and without public-key cryptography) is vulnerable to offline attacks. The adversary eavesdrops on the connection and then mounts a dictionary attack offline-- since there are no secrets beside the password its always possible to guess the password and compare against the actual observed traffic. Perhaps some 14-year old amateur would try to brute-force the account by repeated login attempts but its doubtful the serious attacker might do something that silly. As others pointed out, locking out the account because of incorrect login can easily lead to denial-of-service attacks. BP
This is almost certainly not true.
The better colleges with solid academic-reputation have "need-blind" admission policies.
In other words whether you need financial aid (or how much you need it) *can not* affect your chances of admission.
In fact for government-funded schools (eg state universities) I believe this is legally required-- they are not allowed to discriminate applicants on the basis of income.
Private schools on the other hand can do this; in other words they are under no obligation to have need-blind policies in place. All good schools however state they are need-blind, claiming their admission does not take economic factors into account.
This is particulary true about the "cream of the crop", the schols being bad-mouthed here as elitist: the Ivy league, MIT, Stanford, etc. These schools have huge endowments and constant flood of money from their wealthy alumni to use for their students. (Did you know that for most schools tution covers only half the total expenses? If it werent for the funds from elsewhere the costs would be exorbitantly high in order to break even.)
One could argue ofcourse that admissions without the aid is useless. But in the majority of cases students who are admitted to these schools also receive the full scholarship they ask for.
Consider my case: I am not even a US citizen yet I attended an Ivy with near full scholarship.
That, despite the fact that admission for international students is decidedly *NOT* need-blind. I have seen complete idiots (foreign nationals) pay their way into some of the highly-regarded schools.
If anything one has to be grateful for the generosity here. You would be hard pressed to find schools in Europe which will accommodate such a diverse student body from all economic backgrounds with the necessary financial aid.
Going back to the point, yes rich people might have an early start in the admissions race.
But personal accomplishments (not necessrily academic) will dwarf that difference, making it insignificant and irrelevant.
B
The Alpha architecture is partially under the control of Intel, because they acquired some DEC facilities when Compaq was taking over.
Last summer around July a friend had a meeting with the friendly folks from Intel.
They put it in so many words:
"At Intel we build Intel chips."
So you might need to tone down the pipe-dreams about Alpha competing with Intel.
Secondly, MS dropped the ball on Alpha only because *Compaq* announced it would stop supporting NT on the Alpha. First it was announced that 32b support would go, then the brilliant management decided 64b NT would not be supported either. The engineers at DEC-West (the Pacific Northwest facility which used to do all the NT development and credited with the excellent emulation layer FX!32) dont have jobs anymore.
Wonder of wonders Compaq still supports OpenVMS. Apparently they are trying to position the Alpha platform for Tru64, OpenVMS and Linux.
Except for very high-performance applications, Alpha is just not price-competitive. Don't get me wrong: I am very impresed with the Alpha family. At work I have a dual-proc 533mhz Ultimate WKS. Alpha shines at scientific computing and highend. But for things like web servers, you can get 4 or 8-way PIIIs that support clustering, much cheaper than the Alpha system with comparable performance. Once Merced comes out the gap will be closed even more.
The idea of "Alpha at home" will not fly with consumers.