That right there is the problem. I don't think there would be a problem paying a 2x or even a 3x price difference to go with SGI over x86/Linux. But 10x? That's the deal breaker.
You don't know what you're talking about, do you? You could buy 256 cheapo Dell boxes for one tenth that price, but that won't make a cluster.
I recently saw negotiations to buy a Linux cluster from either IBM or Dell. The price they were talking about came to around $100,000 for a 24-processor configuration, with gigabit networking (not Myrinet or anything like that). That's about $4000 per processor (and these are dual-proc machines); the only reason it was this cheap is because it's an academic institution. If you wanted to buy a 256-processor configuration, you'd have to increase the infrastructure by a huge factor. You'd probably need faster networking as well. And it'd still never parallelize as well as the SGI. It would almost certainly be faster for farming out single-processor jobs, and some MPI programs might run very well, but you wouldn't get the ridiculous cost savings you just pulled out of your ass.
Trying to deal with SGI is like a 16-year-old trying to deal with a salesman at a Rolls-Royce dealership.
Our experience with SGI has been far better than with Dell, which misplaced a server order and then took weeks to replace it, and pretended that nothing had gone wrong.
You can set up a cheap-ass render farm for about $250k
Yes, but:
1. Subtract from that at least 10% of the CPU power, because that's going to have downtime. (This is probably a low estimate.)
2. Add quite a few dollars for increased admin time. If you don't care when boxes die, this won't be quite so bad.
3. Add quite a few dollars for increased power consumption.
4. A lot of that money will be spent on networking equipment.
5. Quite a bit will be spent on a central file server as well.
6. Forget about doing *anything* in parallel (though for a render farm, I don't think this is an issue - just render one frame per CPU at a time).
7. You wouldn't make a Beowulf cluster this large; you'd have sophisticated load management software.
8. You will not get 250 processors for $250k. Some collaborators bought a cluster from IBM and managed to push the price down to $100k. That's for 24 Pentium 4/2.4Ghz CPUs in twelve nodes. Dell couldn't do much better. (You might save a bit by assembling it yourself, but I doubt it.) There is almost always at least one node down.
If you do things right, you still save money over buying an SGI supercomputer; this is one task where commodity hardware definitely can do the job. However, your price/performance estimates are totally at odds with reality.
Thanks, dude. I work with both Linux and SGIs, and I can confirm pretty much everything you say. But for my situation it's actually the reverse. SGI workstations were traditionally very big in my field, but we use Linux for almost everything now. However, we now have two Origin 300s for file serving, and the stability and performance has been far superior to anything we could get out of a PC. We don't need raw speed - we need something with bandwidth, that won't crash when NFS goes out of control. (But of course we'd never buy an SGI to do web serving...)
I suspect most of the people on Slashdot who either bash proprietary Unix or pimp Linux clusters have never spent much time using or administering either.
The big difference is that China is consistent in its image and its actions
Nope. This is the country that's starting to allow *gasp* capitalists into its government, and has been allowing more and more limited free enterprise within its borders (particularly within Hong Kong) and whose economic health depends to a great degree on the continued relationship with the evil bourgeoise imperialists over in the USA. There's a great deal of hypocrisy there - I doubt more than a few of the leaders still believe in Communism; they're just trying to stay in power as long as possible.
I asked a Chinese friend of mine why they didn't dump their government, since they knew it was corrupt and oppressive. He told me, "As long as things keep improving, we deal with it. Nobody wants to dump the Communists when the economy keeps getting better."
The same freedoms that citizens enjoy in every other country: everything, except those things forbidden by the laws made by your government.
Wrong. A communist country explicitly subordinates the economic interests of an individual to the economic interests of the people as a whole. It also places extreme limits of freedom of expression that are vastly worse than anything we've ever had in the US. There is no real concept of personal independence. You ever seen Star Trek: TNG? You remember the Borg? After I read "The Communist Manifesto" for the first time, I decided Karl Marx would have loved those guys.
China has become much better in recent years, but they're still a police state.
the recent rise of the US religious right and gradual breakdown in the separation of the church and the state?
As a student of history, I can tell you that your perspective is all wrong. Religious fundamentalists have always been a threat towards freedom and a factor in US politics, but your perception of them is colored by the fact that a large percentage of the American voting public is more liberal and secular now than in previous eras. The fact that we're now positively shocked by attempts to instill prayer in schools, or teach creationism, or prevent gay marriage, is a good measure of how far we've really come. What you see now is a conservative backlash against increasingly dominant liberal, secular ideals. I'm not saying we don't need to fight it, but you're not looking at long-term trends.
A side note: Bush's religiosity seems absurd to many people, but it wouldn't have a century ago. I agree with the sentiment that the Founding Fathers believed in a secular government, but they certainly didn't have any problem with declaring their belief in God. Separation of church and state should prevent Bush's stupid faith-based initiative, or this anti-family-planning bullcrap, but it doesn't mean he's not allowed to be religious or to let his policies be framed by Christian ideals.
As far as I know, there is no easy method to distinguish a gene from other parts of DNA sequence. In order to get such an estimate heurictic algorithms that look for characteristic patterns in the sequence are used.
"Easy", no, but the methods are better than you think. It's been a while since I've read on this in detail, but there are three techniques which may be used, and which are usually combined:
1. De-novo gene finding. Genscan and its successor GenomeScan are good examples - they look for RNA splice sites (among other things).
2. cDNA matching - purified genes are located on the genome.
3. Homology searches. Find what regions of the genome match known proteins in human or other organisms, and satisfy the conditions for being genes. I think GenomeScan might do this as well.
Usually these methods are combined to some degree. There are all sorts of other things you can do as well; people are still trying to refine the yeast data set by integrating all sorts of experimental data. At any rate, you won't get an exact number for quite a while, but the estimates are almost certainly in the right range. The problem is that the number alone isn't particularly meaningful.
Re:The proteins are where the fun is at
on
Genome Surprise
·
· Score: 1
They do everything, so naturaly they are being studied closely by biologists and drug companies.
And have been for years. Studying proteins in parallel ("proteomics"), however, is a fairly recent innovation, just as genomics grew out of molecular genetics. It'll take a while before we can study proteins at the same scale as genomes. There are efforts now to scale up the process of structure determination through what's (rather inappropriately) called "structural genomics" - I've been working in this field recently. The problem is that crystallography and NMR do not lend themselves to high-throughput techniques as well as sequencing does, though some work is being done to address this.
Protiens can fold four different ways, offering many different structural combinations.
This is incorrect. Protein folding is not an absolute (once you involve membranes and chaperonins it gets even more ridiculously complicated) but molecular diversity is not cause by multiple folded states of the same peptide chain. Instead, it results from alternate splicing of mRNA transcripts, i.e. you get significantly different proteins being made based on the same gene. They may share certain tertiary protein domains, but it's possible for a particular splicing to frameshift part of a transcript relative to another version.
Not if you need a simple workstation or webserver. I can see why people would want a Sun box for high-end stuff - we just bought a couple of SGI Origins for file serving, but would have been just as happy with Suns - but I can see few reasons to buy a simple 1U Sun server, other than binary compatibility with larger systems.
In addition, as the article points out, applications that once ran on giant multi-cpu boxes now run on clusters. Clusters can be a fucking nightmare, but the point is that if you need a lot of computer you don't necessarily have to buy it all in one box unless your applications specifically require many processors with shared memory. Google's Linux farm is a good example.
Yeah, but the fact that you know how to dual-boot means you're already more technically astute than 99% of the people out there. I see your point, and feel the exact same way, but people here seem to forget that Linux is a good choice - even, perhaps, the ideal choice - for the desktop for many people here because of its power and customizability, which the n00bs don't care about. I've been using Linux exclusively for several years and I now find Windows downright difficult and obnoxious to use.
My problem is with stupid people who say "Linux is not a desktop OS" and assume that I can't possibly be using it that way, and thus expect me to fix their Windows boxes, send me Word docs, and don't understand why I don't love Windows.
I'd like to point out that the GUI on SGI's Irix machines is X11, and it's certainly fast enough.
This is because SGI, being a graphics company, did a very, very good job optimizing their X server. Sure, if you try to run Netscape or some other common app on an older box, it'll crawl, but the latest version of Irix still runs fairly well on a ten-year-old machine as long as you have a decent amount of memory. The GUI still feels responsive, despite the slow CPU and (now) slow graphics. You can't say the same about any other OS/hardware combo on the market. The point here is that it is certainly possible to make a full-featured X server extremely fast, without sacrificing features. The problem with this comparison, of course, is that SGI has a very small set of hardware to support, and thus can tune its software very well. XFree86 is in the unfortunate position of trying to support too many graphics chipsets on too many different hardware/OS combinations, which necessarily takes attention away from Linux-x86-specific optimizations.
If I wanted to be pragmatic I wouldn't use Linux in the first place.
I'd wager that most of the surge of interest in Linux in past years has been based on pragmatism, not idealism. I like open source and prefer to use it whenever possible. However, I use Linux because it's the best tool for my job, and Windows is not. Quite a few people think this way, and like it or not they will probably be the ones driving future Linux development and adoption. I respect what the FSF has done for us, but anyone can see how little their idealism has done for getting a complete, usable system working and getting it adopted.
Frankly, the only way I'm going to be able to have any part of my systems be open-source right now is to have some parts be closed-source. I need high-quality 3D acceleration on Unix systems, which pretty much limits me to Linux with proprietary drivers, since my boss is too cheap to buy an SGI (which is proprietary anyway).
Reverse engineering a company's software or protocols isn't scientific nor does it advance science.
Which is why I said "scientific and technical progress." Or perhaps I should have said "science and engineering", since you seem to have missed my point entirely. Reverse engineering isn't science, but it's improved the available technology considerably in the past. The DMCA presents manufacturers the opportunity to cripple third-party product interoperability by forbidding reverse engineering. Whether or not a product is protected by encryption should make absolutely no difference to the rights of whomever purchased it. It's simply a tool to enable greater corporate protectionism than before - companies are protected by the threat of federal charges.
I would also argue that reverse engineering, in a way, is very real science. If you'd ever studied molecular biology you might understand this - I don't consider what I do all that much different from reverse engineering a cell, only we call it "hypothesis-driven science" instead.
I think Edelman needs to stop being a pussy and go ahead and decrypt the list. Then, either find someone willing to defend him, or find a way to leak the list onto the Internet without getting caught.
Heck, he's at Harvard fucking Law School. Maybe he can convince the faculty that this would be a good case study. I doubt N2H2 could take down that big of a target. Besides, they might simply not sue.
How is that giving N2H2 the power to censor the internet?
Because their products are used without any idea of what they actually block. N2H2 thus has wide discretion to block any site its employees find objectionable. Thus the company Solid Oak Software, makers of CyberSitter, decided to block the National Organization for Women, among other public-minded sites (actually I find NOW pretty objectionable, but I also believe in the free exchange of ideas). When this software is being deployed in a tax-payer funded environment, the government is paying N2H2 to decide what the citizens can or can't view on public computers. My objection to library/school filtering is based principally on this fact - if the filtering lists were made public I would not find the laws offensive (though I still think they're unconstitutional).
I would have said something different from the parent poster: scientific and technical progress depends on public disclosure and peer review. It also depends on the freedom to investigate a problem without interference from the government or companies (as long as no one gets hurt!).
the RIAA generally asks that the school shut down the site, cut of network access, and turn over the students name.
My sense was (at least at Yale) that the last is never actually done, and network access is only taken away for repeat offenders. The university will certainly take care of copyright violations, but the RIAA can't force them to take punitive action or turn over student records.
Actually, Yale was one of the first schools targeted by Metallica - we made the mistake of partially blocking Napster due to network traffic, which made it difficult to argue that Yale ITS was simply a common carrier that didn't discriminate between different types of traffic. The administration caved almost immediately - they didn't think it was worth going to court to support students' music piracy, particularly when the network was getting trashed.
Unfortunately, I think the RIAA is perfectly within its rights to target individual file traders. However, I also think it's mean-spirited, stupid, and counterproductive to target some dumb college kids.
Most universities (from what I've read, mine included) will handle the matter internally, making sure that the student stops distributing copyrighted music and then replying to the RIAA that any problem has been resolved. Students who do this repeatedly might be disciplined or lose their network privileges, but they're certainly not turned over to the RIAA.
Truth is, most college administrators and network people can't stand illegal file sharing and certainly don't want the school to get in trouble- when universities started blocking Napster most of the posts to a related Slashdot story were from university net admins who loved the idea. However, they also don't want to see their students persecuted and definitely don't have time to police the network for the music labels.
What is most concerning here is that the RIAA is a de facto trust that is likely breaking the law by monopolization of intellectual property and distribution with respect to music.
Horseshit. They're a trade and lobbying association, just like any other. For the RIAA to be in violation of antitrust law, it'd have to (for instance) make secret agreements between major labels about what new acts get signed, or set industry-wide prohibitions on how they distribute music over the internet. While I certainly find their actions distasteful (especially in this case), I agree in principle that piracy is wrong. And I can't think of any way in which an idiot college student running an mp3z server represents any sort of "innovation".
Intresting you note on Vaclav Havel for he is now having a pretty bad track redord himself.
Really? I know his reputation in the West has been grossly inflated, but I wasn't aware of anything bad that he'd done - care to elaborate?
In any case, Soviet Russia was falling for its last 20 years. And it had surprisingly little to do with the lack of a free market.
Agreed, sort of. The symptoms that plagued Russian industry can be seen in many individual companies in the US. However, in the US the market usually eventually corrects itself. In Russia, the government (and the people's fear of government) keeps things going far too long. If the Party said to increase production, production was increased even if the new rate was unsustainable and the product crap.
I'd say a good comparison would be Enron, except that sooner or later Enron had to collapse. In a communist state, Enron would have been propped up as long as possible.
By now, most people equate "administrative assistant" == secretary, anyways.
I'd say there's a subtle difference. A good "administrative assistant" will actually do quite a few things behind the scenes. I work in academia, and some of the AA's here are very competent and keep stuff running which the professors don't have a clue about. (Plus, there's a wide range in job category - I wouldn't insult the assistant to the chairman of a huge, wealthy, important department by calling her a secretary.) My mom is an administrative assistant to the headmaster of an expensive private high school, and while the routine parts of her job involve taking minutes and ghost-writing letters, she's also had to singlehandedly prepare immense reports and has several people working for her. She's so effective that they keep dumping more crap on her.
Nowadays, however, "secretary" usually denotes someone who does little more than stenography and managing files. Or to take my job, the difference is the same as between "research assistant" (runs experiments) and "technician" (washes glassware).
To be fair, it seems probable that Reagan hastened the fall, but I've never believed that the outcome was planned that way or that America's fiscal irresponsibility during those years was appropriate. One thing I read recently pointed out that Reagan's most lasting contribution was emphasizing human rights in the Soviet bloc, which apparently really did encourage the many dissidents over there like Vaclav Havel. (Ironic given Reagan's abominable track record on human rights in, say, Central America.)
Some of the most interesting articles I've read in the past decade have chronicled the disasters left behind by Russian communism - it's obvious that their entire economic system was barely creaking along by the start of the 1980s. Jonathan Pryce's apartment in Brazil was pretty close to the reality.
I wouldn't describe them as "leftwing"; a better word would be "totalitarian", but since Mao's death it's really just been "authoritarian" with strong socialist underpinnings. It hasn't been a true communist state for some time (though it's nowhere near to being a proper capitalist state).
From what I've read and been told (college history class, etc.), the attitude of the Chinese government can be oversimplified as one of extreme distrust over any mass medium or mass *movement* that they don't control. Tiannamen is the most famous case, but the Falun Gong and indeed any sort of religion are persecuted because they represent popular organization that isn't managed by the government. When Zhou Enlai (China's most famous Communist leader other than Mao - very interesting person) died, many people were genuinely distraught and held a spontaneous wake in Beijing. The government broke it up, because it wasn't under their control. I think the Internet appears the same way to them.
That's just my opinion, but a Chinese coworker thought it made sense when I explained it to him.
If fusion goes big-time, that means that just as with fission reactors, very large quatitites of radioactive waste will be generated.
Huh? Most of the waste from conventional fission plants is spent fuel and its byproducts, like Cesium-137 (one of the worst pollutants from Chernobyl). Protection against neutron radiation has always been through very thick concrete walls, and obviously those don't get thrown away. I don't know anything about the neutron output of fusion, but the principal "byproduct" is helium rather than various nasty heavy isotopes.
Slashdot Mirror
That right there is the problem. I don't think there would be a problem paying a 2x or even a 3x price difference to go with SGI over x86/Linux. But 10x? That's the deal breaker.
You don't know what you're talking about, do you? You could buy 256 cheapo Dell boxes for one tenth that price, but that won't make a cluster.
I recently saw negotiations to buy a Linux cluster from either IBM or Dell. The price they were talking about came to around $100,000 for a 24-processor configuration, with gigabit networking (not Myrinet or anything like that). That's about $4000 per processor (and these are dual-proc machines); the only reason it was this cheap is because it's an academic institution. If you wanted to buy a 256-processor configuration, you'd have to increase the infrastructure by a huge factor. You'd probably need faster networking as well. And it'd still never parallelize as well as the SGI. It would almost certainly be faster for farming out single-processor jobs, and some MPI programs might run very well, but you wouldn't get the ridiculous cost savings you just pulled out of your ass.
Trying to deal with SGI is like a 16-year-old trying to deal with a salesman at a Rolls-Royce dealership.
Our experience with SGI has been far better than with Dell, which misplaced a server order and then took weeks to replace it, and pretended that nothing had gone wrong.
You can set up a cheap-ass render farm for about $250k
Yes, but:
1. Subtract from that at least 10% of the CPU power, because that's going to have downtime. (This is probably a low estimate.)
2. Add quite a few dollars for increased admin time. If you don't care when boxes die, this won't be quite so bad.
3. Add quite a few dollars for increased power consumption.
4. A lot of that money will be spent on networking equipment.
5. Quite a bit will be spent on a central file server as well.
6. Forget about doing *anything* in parallel (though for a render farm, I don't think this is an issue - just render one frame per CPU at a time).
7. You wouldn't make a Beowulf cluster this large; you'd have sophisticated load management software.
8. You will not get 250 processors for $250k. Some collaborators bought a cluster from IBM and managed to push the price down to $100k. That's for 24 Pentium 4/2.4Ghz CPUs in twelve nodes. Dell couldn't do much better. (You might save a bit by assembling it yourself, but I doubt it.) There is almost always at least one node down.
If you do things right, you still save money over buying an SGI supercomputer; this is one task where commodity hardware definitely can do the job. However, your price/performance estimates are totally at odds with reality.
Thanks, dude. I work with both Linux and SGIs, and I can confirm pretty much everything you say. But for my situation it's actually the reverse. SGI workstations were traditionally very big in my field, but we use Linux for almost everything now. However, we now have two Origin 300s for file serving, and the stability and performance has been far superior to anything we could get out of a PC. We don't need raw speed - we need something with bandwidth, that won't crash when NFS goes out of control. (But of course we'd never buy an SGI to do web serving...)
I suspect most of the people on Slashdot who either bash proprietary Unix or pimp Linux clusters have never spent much time using or administering either.
The big difference is that China is consistent in its image and its actions
Nope. This is the country that's starting to allow *gasp* capitalists into its government, and has been allowing more and more limited free enterprise within its borders (particularly within Hong Kong) and whose economic health depends to a great degree on the continued relationship with the evil bourgeoise imperialists over in the USA. There's a great deal of hypocrisy there - I doubt more than a few of the leaders still believe in Communism; they're just trying to stay in power as long as possible.
I asked a Chinese friend of mine why they didn't dump their government, since they knew it was corrupt and oppressive. He told me, "As long as things keep improving, we deal with it. Nobody wants to dump the Communists when the economy keeps getting better."
The same freedoms that citizens enjoy in every other country: everything, except those things forbidden by the laws made by your government.
Wrong. A communist country explicitly subordinates the economic interests of an individual to the economic interests of the people as a whole. It also places extreme limits of freedom of expression that are vastly worse than anything we've ever had in the US. There is no real concept of personal independence. You ever seen Star Trek: TNG? You remember the Borg? After I read "The Communist Manifesto" for the first time, I decided Karl Marx would have loved those guys.
China has become much better in recent years, but they're still a police state.
the recent rise of the US religious right and gradual breakdown in the separation of the church and the state?
As a student of history, I can tell you that your perspective is all wrong. Religious fundamentalists have always been a threat towards freedom and a factor in US politics, but your perception of them is colored by the fact that a large percentage of the American voting public is more liberal and secular now than in previous eras. The fact that we're now positively shocked by attempts to instill prayer in schools, or teach creationism, or prevent gay marriage, is a good measure of how far we've really come. What you see now is a conservative backlash against increasingly dominant liberal, secular ideals. I'm not saying we don't need to fight it, but you're not looking at long-term trends.
A side note: Bush's religiosity seems absurd to many people, but it wouldn't have a century ago. I agree with the sentiment that the Founding Fathers believed in a secular government, but they certainly didn't have any problem with declaring their belief in God. Separation of church and state should prevent Bush's stupid faith-based initiative, or this anti-family-planning bullcrap, but it doesn't mean he's not allowed to be religious or to let his policies be framed by Christian ideals.
As far as I know, there is no easy method to distinguish a gene from other parts of DNA sequence. In order to get such an estimate heurictic algorithms that look for characteristic patterns in the sequence are used.
"Easy", no, but the methods are better than you think. It's been a while since I've read on this in detail, but there are three techniques which may be used, and which are usually combined:
1. De-novo gene finding. Genscan and its successor GenomeScan are good examples - they look for RNA splice sites (among other things).
2. cDNA matching - purified genes are located on the genome.
3. Homology searches. Find what regions of the genome match known proteins in human or other organisms, and satisfy the conditions for being genes. I think GenomeScan might do this as well.
Usually these methods are combined to some degree. There are all sorts of other things you can do as well; people are still trying to refine the yeast data set by integrating all sorts of experimental data. At any rate, you won't get an exact number for quite a while, but the estimates are almost certainly in the right range. The problem is that the number alone isn't particularly meaningful.
They do everything, so naturaly they are being studied closely by biologists and drug companies.
And have been for years. Studying proteins in parallel ("proteomics"), however, is a fairly recent innovation, just as genomics grew out of molecular genetics. It'll take a while before we can study proteins at the same scale as genomes. There are efforts now to scale up the process of structure determination through what's (rather inappropriately) called "structural genomics" - I've been working in this field recently. The problem is that crystallography and NMR do not lend themselves to high-throughput techniques as well as sequencing does, though some work is being done to address this.
Protiens can fold four different ways, offering many different structural combinations.
This is incorrect. Protein folding is not an absolute (once you involve membranes and chaperonins it gets even more ridiculously complicated) but molecular diversity is not cause by multiple folded states of the same peptide chain. Instead, it results from alternate splicing of mRNA transcripts, i.e. you get significantly different proteins being made based on the same gene. They may share certain tertiary protein domains, but it's possible for a particular splicing to frameshift part of a transcript relative to another version.
Linux is an alternative to Windows, not Sun.
Not if you need a simple workstation or webserver. I can see why people would want a Sun box for high-end stuff - we just bought a couple of SGI Origins for file serving, but would have been just as happy with Suns - but I can see few reasons to buy a simple 1U Sun server, other than binary compatibility with larger systems.
In addition, as the article points out, applications that once ran on giant multi-cpu boxes now run on clusters. Clusters can be a fucking nightmare, but the point is that if you need a lot of computer you don't necessarily have to buy it all in one box unless your applications specifically require many processors with shared memory. Google's Linux farm is a good example.
Yeah, but the fact that you know how to dual-boot means you're already more technically astute than 99% of the people out there. I see your point, and feel the exact same way, but people here seem to forget that Linux is a good choice - even, perhaps, the ideal choice - for the desktop for many people here because of its power and customizability, which the n00bs don't care about. I've been using Linux exclusively for several years and I now find Windows downright difficult and obnoxious to use.
My problem is with stupid people who say "Linux is not a desktop OS" and assume that I can't possibly be using it that way, and thus expect me to fix their Windows boxes, send me Word docs, and don't understand why I don't love Windows.
I'd like to point out that the GUI on SGI's Irix machines is X11, and it's certainly fast enough.
This is because SGI, being a graphics company, did a very, very good job optimizing their X server. Sure, if you try to run Netscape or some other common app on an older box, it'll crawl, but the latest version of Irix still runs fairly well on a ten-year-old machine as long as you have a decent amount of memory. The GUI still feels responsive, despite the slow CPU and (now) slow graphics. You can't say the same about any other OS/hardware combo on the market. The point here is that it is certainly possible to make a full-featured X server extremely fast, without sacrificing features. The problem with this comparison, of course, is that SGI has a very small set of hardware to support, and thus can tune its software very well. XFree86 is in the unfortunate position of trying to support too many graphics chipsets on too many different hardware/OS combinations, which necessarily takes attention away from Linux-x86-specific optimizations.
If I wanted to be pragmatic I wouldn't use Linux in the first place.
I'd wager that most of the surge of interest in Linux in past years has been based on pragmatism, not idealism. I like open source and prefer to use it whenever possible. However, I use Linux because it's the best tool for my job, and Windows is not. Quite a few people think this way, and like it or not they will probably be the ones driving future Linux development and adoption. I respect what the FSF has done for us, but anyone can see how little their idealism has done for getting a complete, usable system working and getting it adopted.
Frankly, the only way I'm going to be able to have any part of my systems be open-source right now is to have some parts be closed-source. I need high-quality 3D acceleration on Unix systems, which pretty much limits me to Linux with proprietary drivers, since my boss is too cheap to buy an SGI (which is proprietary anyway).
Reverse engineering a company's software or protocols isn't scientific nor does it advance science.
Which is why I said "scientific and technical progress." Or perhaps I should have said "science and engineering", since you seem to have missed my point entirely. Reverse engineering isn't science, but it's improved the available technology considerably in the past. The DMCA presents manufacturers the opportunity to cripple third-party product interoperability by forbidding reverse engineering. Whether or not a product is protected by encryption should make absolutely no difference to the rights of whomever purchased it. It's simply a tool to enable greater corporate protectionism than before - companies are protected by the threat of federal charges.
I would also argue that reverse engineering, in a way, is very real science. If you'd ever studied molecular biology you might understand this - I don't consider what I do all that much different from reverse engineering a cell, only we call it "hypothesis-driven science" instead.
I think Edelman needs to stop being a pussy and go ahead and decrypt the list. Then, either find someone willing to defend him, or find a way to leak the list onto the Internet without getting caught.
Heck, he's at Harvard fucking Law School. Maybe he can convince the faculty that this would be a good case study. I doubt N2H2 could take down that big of a target. Besides, they might simply not sue.
How is that giving N2H2 the power to censor the internet?
Because their products are used without any idea of what they actually block. N2H2 thus has wide discretion to block any site its employees find objectionable. Thus the company Solid Oak Software, makers of CyberSitter, decided to block the National Organization for Women, among other public-minded sites (actually I find NOW pretty objectionable, but I also believe in the free exchange of ideas). When this software is being deployed in a tax-payer funded environment, the government is paying N2H2 to decide what the citizens can or can't view on public computers. My objection to library/school filtering is based principally on this fact - if the filtering lists were made public I would not find the laws offensive (though I still think they're unconstitutional).
I would have said something different from the parent poster: scientific and technical progress depends on public disclosure and peer review. It also depends on the freedom to investigate a problem without interference from the government or companies (as long as no one gets hurt!).
the RIAA generally asks that the school shut down the site, cut of network access, and turn over the students name.
My sense was (at least at Yale) that the last is never actually done, and network access is only taken away for repeat offenders. The university will certainly take care of copyright violations, but the RIAA can't force them to take punitive action or turn over student records.
Actually, Yale was one of the first schools targeted by Metallica - we made the mistake of partially blocking Napster due to network traffic, which made it difficult to argue that Yale ITS was simply a common carrier that didn't discriminate between different types of traffic. The administration caved almost immediately - they didn't think it was worth going to court to support students' music piracy, particularly when the network was getting trashed.
Unfortunately, I think the RIAA is perfectly within its rights to target individual file traders. However, I also think it's mean-spirited, stupid, and counterproductive to target some dumb college kids.
Most universities (from what I've read, mine included) will handle the matter internally, making sure that the student stops distributing copyrighted music and then replying to the RIAA that any problem has been resolved. Students who do this repeatedly might be disciplined or lose their network privileges, but they're certainly not turned over to the RIAA.
Truth is, most college administrators and network people can't stand illegal file sharing and certainly don't want the school to get in trouble- when universities started blocking Napster most of the posts to a related Slashdot story were from university net admins who loved the idea. However, they also don't want to see their students persecuted and definitely don't have time to police the network for the music labels.
What is most concerning here is that the RIAA is a de facto trust that is likely breaking the law by monopolization of intellectual property and distribution with respect to music.
Horseshit. They're a trade and lobbying association, just like any other. For the RIAA to be in violation of antitrust law, it'd have to (for instance) make secret agreements between major labels about what new acts get signed, or set industry-wide prohibitions on how they distribute music over the internet. While I certainly find their actions distasteful (especially in this case), I agree in principle that piracy is wrong. And I can't think of any way in which an idiot college student running an mp3z server represents any sort of "innovation".
Intresting you note on Vaclav Havel for he is now having a pretty bad track redord himself.
Really? I know his reputation in the West has been grossly inflated, but I wasn't aware of anything bad that he'd done - care to elaborate?
In any case, Soviet Russia was falling for its last 20 years. And it had surprisingly little to do with the lack of a free market.
Agreed, sort of. The symptoms that plagued Russian industry can be seen in many individual companies in the US. However, in the US the market usually eventually corrects itself. In Russia, the government (and the people's fear of government) keeps things going far too long. If the Party said to increase production, production was increased even if the new rate was unsustainable and the product crap.
I'd say a good comparison would be Enron, except that sooner or later Enron had to collapse. In a communist state, Enron would have been propped up as long as possible.
By now, most people equate "administrative assistant" == secretary, anyways.
I'd say there's a subtle difference. A good "administrative assistant" will actually do quite a few things behind the scenes. I work in academia, and some of the AA's here are very competent and keep stuff running which the professors don't have a clue about. (Plus, there's a wide range in job category - I wouldn't insult the assistant to the chairman of a huge, wealthy, important department by calling her a secretary.) My mom is an administrative assistant to the headmaster of an expensive private high school, and while the routine parts of her job involve taking minutes and ghost-writing letters, she's also had to singlehandedly prepare immense reports and has several people working for her. She's so effective that they keep dumping more crap on her.
Nowadays, however, "secretary" usually denotes someone who does little more than stenography and managing files. Or to take my job, the difference is the same as between "research assistant" (runs experiments) and "technician" (washes glassware).
the untenable nature of a command economy
To be fair, it seems probable that Reagan hastened the fall, but I've never believed that the outcome was planned that way or that America's fiscal irresponsibility during those years was appropriate. One thing I read recently pointed out that Reagan's most lasting contribution was emphasizing human rights in the Soviet bloc, which apparently really did encourage the many dissidents over there like Vaclav Havel. (Ironic given Reagan's abominable track record on human rights in, say, Central America.)
Some of the most interesting articles I've read in the past decade have chronicled the disasters left behind by Russian communism - it's obvious that their entire economic system was barely creaking along by the start of the 1980s. Jonathan Pryce's apartment in Brazil was pretty close to the reality.
beeing a tad more leftwing than most others
I wouldn't describe them as "leftwing"; a better word would be "totalitarian", but since Mao's death it's really just been "authoritarian" with strong socialist underpinnings. It hasn't been a true communist state for some time (though it's nowhere near to being a proper capitalist state).
From what I've read and been told (college history class, etc.), the attitude of the Chinese government can be oversimplified as one of extreme distrust over any mass medium or mass *movement* that they don't control. Tiannamen is the most famous case, but the Falun Gong and indeed any sort of religion are persecuted because they represent popular organization that isn't managed by the government. When Zhou Enlai (China's most famous Communist leader other than Mao - very interesting person) died, many people were genuinely distraught and held a spontaneous wake in Beijing. The government broke it up, because it wasn't under their control. I think the Internet appears the same way to them.
That's just my opinion, but a Chinese coworker thought it made sense when I explained it to him.
If fusion goes big-time, that means that just as with fission reactors, very large quatitites of radioactive waste will be generated.
Huh? Most of the waste from conventional fission plants is spent fuel and its byproducts, like Cesium-137 (one of the worst pollutants from Chernobyl). Protection against neutron radiation has always been through very thick concrete walls, and obviously those don't get thrown away. I don't know anything about the neutron output of fusion, but the principal "byproduct" is helium rather than various nasty heavy isotopes.