My comment (parent post) was modded up to 5 for a little while, then some chicken RMS fanboys gave it a few "overrated" mods to bring it down. I don't care about my karma--it's in the stratosphere anyway--but I noticed this peculiar slashdot bug (?). Here's the moderation history for my comment
Now, 30% + 30% + 20% != 100%. My first guess was that I got a total of 5 mods: 2 informatives, 2 overrateds, and 1 insightful, which, with my +1 Karma bonus, brings me to the current score of 3. The thing is, then it should be 40% / 40% / 20%. Is this just a math bug, or is there some sort of weird secret moderation going on that doesn't show up in the moderation report?
Note to mods: I intentionally did not use my karma bonus on this post, because it's sort of off-topic.
Step down from what? The FSF? He created it, he runs it, and it is his baby.
Yes, but it's time for him to move on. The best thing he could do for the Free Software movement is to step down. If he's a reasonable person with his ego under control, he'll realize that. Right now, he's become a caricature and his mere presence in any sort of official role hurts the movement. See the other post (sibling to yours, I believe) which describes the time he put on a robe and halo during a lecture to businessmen.
And the alternative? Mediocrity? Let me ask you this... Who is pushing back from the other side?
There are two ways to accomplish change: incrementally, or via revolution. Given that copyright never, ever registers in the top ten list of things voters care about, the chance of a revolution at the polls is almost zero. The odds of slashdotters leaving their basements, grabbing their pitchforks, and storming the RIAA are even smaller. Stallman doesn't get this, but the only option left to us is incremental change, and that requires pushing at the edge of the mainstream.
Being the most reasonable person in a public debate is often the way to victory. With the RIAA suing grandmothers, this shouldn't be hard, but Stallman is blowing it for us.
Several years ago, I heard Stallman speak at a lecture at my university. He was clearly very smart, and very driven by ideological goals. On top of all that, there was also a hint of that indefinable quality shared by most crazy people. Something about him was not quite right--I got a sense that his grasp on reality was slipping a bit. Maybe this is necessary for a person to make the kind of sacrifices he has, but it's a dangerous balance.
From reading the recent draft of the GPL v3, and the article attached to this story, I get a sense that he's slipped further. For instance, when he spoke at my university, he recognized that the best way to achieve your goals is to have limited, realistic goals, and focus on those. When people asked him about copyright on music or movies, he diplomatically dodged the question and said it was a separate issue from his Free Software philosophy, and he didn't want to address it. In the interview linked in TFA, he outright attacks copyright for these things. The GPL v3's attack on DRM is similar. Stallman has sacrificed the clarity and readibility of the GPL v2 in order to attack patents and DRM.
Now, maybe you agree with Stallman about copyright for music, etc. Even so, you should recognize that that puts you farther outside the mainstream, and it's much harder to change the mainstream when you're 1,000 miles away. If a bunch of Americans write letters to Congress demanding that copyright be abolished*, they will be ignored. If they ask that copyright law take a step back towards the original constitutional idea of limited (in time and power) protection to promote progress in science and the useful arts, that may actually get somewhere. It is vitally important that we sound reasonable.
Stallman has lost his sense of perspective and his grasp on reality. I think it's possible that he is now harmful to the Free Software movement, and the community needs to think about how to deal with this problem. If the community asked him to step down, would he?
* I know Stallman didn't outright call for the abolition of copyright. Still, the changes he wants (the freedom for anyone to distribute any published work) amount to nearly the same thing.
With option 1, I have some ability to do good. Under option 2, I have no ability to do good.
With option 1, Google does some evil (censorship) in return for the opportunity to do some good (and make a profit). With option 2, they do no evil, but also give up the chance at doing some good.
It's important to note that option 1 is inconsistent with the philosophy of "do no evil", but not necessarily inconsistent with the philosophy of "don't be evil", since the latter could be seen as dependent only on the net moral balance of actions.
With that subtle point made, I'd now like to engage in an egregious violation of Godwin's Law, and liken Google's dilemma to that of a person faced with the choice of Nazi collaboration.
Let's say you are (unwillingly) a Nazi officer. You are given the option of taking command of Auschwitz, or refusing and allowing another to take command. If you were in charge, you could perhaps assist some Jews in escaping, maybe reduce the number executed somewhat, but you would still be involved in the execution of innocent people. Do you refuse?
What I've done is reformulate the fundamental moral dilemma (should you become complicit with evil in order to do some good) in a very different emotional context. It could be argued that the good/evil balance is different in the two cases, but that's not the point here (i.e. adjust the number of people saved from Nazis until the balance is comparable). It's also true that the magnitude of both the good and the evil in the Nazi case is probably greater (although not necessarily--Yahoo's cooperation with China has already gotten a journalist thrown in jail), but this is again not the point. I'm interested in hearing how people react to this dilemma, versus Google's actions.
If Google's failure to willingly cooperate means that they're completely blocked in China, then as far as the West influencing things it's worse.
Influence is a two-way street. Trade with China was supposed to export our values of freedom and democracy to them, but instead we're importing their corruption and contempt for human rights. Before exposure to China, Google was arguably one of the most morally upright major American companies. Now they're turning into scum.
To put it another way, if Google cooperates with China's censorship, they don't have a whole lot of ammunition when the US Department of Justice asks them for people's search records. This does affect you--it's not just about China.
US Representative Chris Smith (R. NJ) is planning to hold hearings into the cooperation of US internet companies with Chinese censorship. I just wrote my congresswoman and asked her to support these hearings. You should consider doing the same, if you are an American.
I also plan to write Google and tell them I am canceling my AdWords campaign because of their practices.
With all this talk of PKI and such, has anyone actually started planning for the collapse of RSA, Diffie-Hellman, and other forms of public-key crypto? We've never had a security proof of these systems (information-theoretic or even computational security), and, since the publication of Shor's algorithm, we now know they can be broken with a quantum computer. Perhaps there's also a classical algorithm for breaking them, but let's assume not. Quantum computers are probably a minimum of ten years away (more like twenty five years, but let's consider the realistic worst case of ten).
I'm guessing many of you have forward security requirements of longer than ten years. Medical records, financial data, and lots of other things probably need to stay secret for longer than that. That means that you need to start now on preparing for the collapse of public-key crypto. Has anyone in IT actually started doing this? Anyone even thought about it?
Disclaimer: IAAQCR (I am a quantum computing researcher).
As another poster pointed out, AltiVec is great (although it's only single-precision, dammit), and there are some nice AltiVec libraries for BLAS and for stuff people in bioinformatics like to play with.
Another advantage is that the G5 is very similar to the processors used in DataStar, the biggest machine at the San Diego Supercomputing Center. This can be handy for doing small-scale testing and optimization before going to the big time.
Still another advantage is OS X--a really nice GUI with Unix goodness under the hood. Finally, there's the convenient 1U xserve enclosure, which is good for DIY small-scale clusters.
Actually, Schneier specifically said in that interview that it makes public key cryptography insecure. He was referring to symmetric, private key cryptography when he stated that it doesn't make all cryptography insecure.
I took another look at the interview, and it's really not clear. In one paragraph, he says that quantum computers render public key crypto insecure, and, a few paragraps later, he makes the statement I quoted about quantum computers effectively only halving key length. He doesn't make any distinction between private or public-key crypto in this paragraph, and it's not clear from context. One could easily read it to mean that quantum computers effectively halve key length, thus making current public key crypto insecure, but allowing for the simple fix of doubling current public key crypto key lengths to compensate (which of course won't really help).
I'll grant him the benefit of the doubt and assume that he knows the difference. Still, he should be a lot more careful with public statements about this kind of stuff. People listen to him and God knows there are already enough slashdotters confused about quantum computing.
Oh, and Bruce Schneier, if you're reading this, sorry about saying nasty things about you.
This is such bad research that I can't wait until Bruce Schneier [schneier.com] get ahold of this.
I lost all respect for Schneier when he shot his mouth off about quantum computation in a 1999 Slashdot interview. Quoting Schneier:
And when it becomes a reality, it does not destroy all cryptography. Quantum computing reduces the complexity of arbitrary calculations by a factor of a square root. This means that key lengths are effectively halved. 128-bit keys are more than secure enough today; 256-bit keys are more than secure enough against quantum computers.
This is completely wrong. Schneier has confused Grover's algorithm (which does something analogous to linear search, e.g. of a key space), with Shor's algorithm (prime factorization and discrete log). Shor's algorithm gives an exponential speedup over the best known classical algorithm, not a square-root speed-up (which is what you get from Grover's algorithm). To make things really clear, Shor's algorithm scales as O(n^3) of the number of bits of your key, and perhaps better if you optimize it carefully. If you go from 128 bits to 256 bits, your classical computer needs to get about 2^128 times more powerful (give or take a few orders of magnitude), while your quantum computer needs only to get about eight times more powerful.
Shor's algorithm was published around 1994 or 1995. For Schneier, a supposed cryptography expert, to be ignorant of a published attack against most public-key cryptosystems more than four years later is just inexcusable. This attack is not just of theoretical concern, as Schneier suggests. For anyone dealing with secrets where you need forward security of more than fifteen years, you have to worry about quantum computers and Shor's algorithms. I'm sure there are lots of people who want, say, their medical records to remain secret (especially things like abortions, STDs, etc.) for longer than that. If such people listened to Schneier, they'd be badly misinformed.
What's worse than Schneier's ignorance of such a crucial development in his own field of cryptography and cryptanalysis is his arrogance about it. He didn't just punt the question or say that he didn't know enough about quantum computers to be sure. He gave a very confident and very wrong answer.
I will lose all respect of the Canadian government if they actually try to implement such a device. I thought that the German Autobahn should've been a lesson to the world that it's not speed in itself that is dangerous, it's having unsafe cars being driven on unsafe roads by irresponsible people.
I agree that this GPS speed limiter is a tremendously stupid idea. That said, the rest of your comment is off the mark.
So in essence, the Germans are happily driving at 250+ Km/h on their autobahns without having significantly more accidents than us, because they have much higher standards when it comes to issuing drivers' licenses, they have suited roads, and their cars go through a very strict mechanical check-up every year, to make sure they are road-legal.
The fundamental issues here are geography, weather, and population density, not quality of cars and drivers. Canada != Germany. There are very few places in Canada where a significantly faster highway speed limit would be safe. Many of the highways in British Columbia, including a large chunk of the Trans-Canada, are twisty roads through the mountains with only one or two lanes in each direction. The western edge of Alberta is similar. The prairie provinces are flat enough and the roads there straight enough for high-speed travel, but the highways are often only two lane, undivided roads. In northern Ontario, the highways (including the Trans-Canada) are often twisty, undivided, two-lane roads. In southern Ontario, it's all one big traffic jam, and on the rare occasions when the highways are flowing freely, highway traffic travels at about 150 km/h (i.e. everyone ignores the posted speed limit, which is not enforced; this is the one place the limit could perhaps be increased). I haven't driven east of Ontario, but you get the picture. Some of this could be improved by building more/better roads, but the geography imposes severe limits everywhere but the prairies, and low population density makes it hard to justify the enormous expense of upgrading the roads.
Weather is another important factor. In the mountainous regions of B.C. and Alberta, things can change very, very quickly. Many crashes are caused by sudden weather changes and the resulting unsafe road conditions. If you're 200km from anything, and all of a sudden a blizzard kicks up and you have 20m visibilty, what do you do? Drive 25 km/h (roughly the maximum safe speed in those conditions) and spend a very cold night stuck in a blizzard on the highway, also risking getting rear-ended by someone who didn't see your tail-lights? Or do you drive faster so you can make it to a town before the snow builds up and the road becomes impassable, hoping you'll be able to stop in time if there's something on the road ahead? This isn't hypothetical--pretty much this situation happened to me last winter. Fortunately, the storm cleared up and I got home safely.
This isn't to say that bad drivers are not the cause of accidents in Canada--in BC, nearly a third of traffic fatalities are the result of drinking and driving, which is clearly something only very bad drivers do (but not something stricter drivers' license tests would catch). Penalties and enforcement of drinking and driving regulations are pretty strict, notwithstanding Premier Campbell's vacation to Hawaii, but it's still an issue.
My point is that you can't fairly claim that the German system is better or would work in Canada (or perhaps the United States). The circumstances are simply very different.
Quantum crypto is theoretically amusing, but it practice it's boring. It lets you connect two sites together with dedicated fiber optics and run unbreakable-via-physics crypto over it, but that's a lot more expensive and incovenient that using whatever transmission medium supports your communication speeds and running unbreakable-via-mathematics crypto over it.
By "unbreakable-via-mathematics", do you mean information-theoretic secure (e.g. secure against all attacks, independent of physics), or computationally-secure (secure against attacks using a reasonably-sized classical computer)? The issue is that there aren't any information-theoretic algorithms out there for key distribution, only for encryption once you have a shared secret key the size of the message (I'm talking about one-time pad encryption). If you mean computationally secure, then AFAIK there are no public-key algorithms known to be computationally secure against a classical computer, let alone a quantum computer. Even RSA was not known to be computationally secure--people just strongly suspected it was. For all we know, there may be an efficient classical attack against it.
Is all this paranoia? Not if you need forward security (e.g. your secrets still need to be secret in 20 years when we might have quantum computers), or if your secrets are valuable enough that people would be willing to hire genius mathematicians to try to break your cryptosystems. The first case is more common than you might think: consider medical records, for instance. If you're young now, and you get an abortion or whatever, that could come back to haunt you when you run for office in 20 years. Even normal people do need forward security for certain things.
I should say that my background is in quantum computing and physics, with a more limited understanding of cryptography. If you're aware of some algorithms for public-key cryptography that have been proven to be information-theoretic secure or even just computationally secure, I would love to hear about them. I don't know too much about public key beyond RSA, and would be interested to hear more about elliptic curve crypto. It would be interesting to try to see if it's vulnerable to the quantum fourier transform.
the degree of precision required would be many orders of magnitude greater than any observations of any physical laws have ever been in a real experiment)
This is not true. The fault-tolerance threshold--the error threshold below which an arbitrarily long quantum computation can be performed to arbitrary precision with only polynomial overhead--is estimated to be anywhere from 10^-1 to 10^-7, depending on the physical system and the error model.
Now, 10^-7 is pretty hard to reach, but we most certainly have performed physical experiments to that degree of accuracy. Just not quantum computers, yet. It's important to note that there are other approaches to fault-tolerant quantum computation beyond just error correcting codes. Topologically fault-tolerant quantum computing is currently highly theoretical, but has a lot of potential.
In fact, part of the power of quantum computing is that (even without the somewhat less plausible factoring algorithm) we would have real secure encryption
Quantum cryptography, or, more correctly, quantum key distribution (QKD), only does a subset of what today's public key cryptography does. QKD requires that the two parties already have a small shared secret for authentication, and simply allows those parties to produce an arbitrarily large shared secret (e.g. random number which they both know, but is hidden from all other parties). That large shared secret can be used as a one-time pad. In other words, QKD solves the encryption problem, but not the authentication problem.
One of the sibling posts makes an incorrect statement about QKD: The problem with quantum crypto is that it requires a direct transmission of photons from Alice to Bob. You can't have a relay station in-between, unless you are willing to guarantee its security (any relay station would allow for interception of the signal when it isn't entangled - which cannot be detected).
This is most certainly not the case if Alice and Bob have a shared secret, unless you use a particularly boneheaded implementation of QKD. QKD requires only a quantum and a classical channel between Alice and Bob. It doesn't matter if that channel is a single fiber, or a series of repeaters doing quantum teleportation and entanglement swapping, or whatever. The whole point of QKD is to allow Alice and Bob to detect tampering with the quantum channel, and that works regardless of the physical nature of the channel.
The catch is when Alice and Bob don't have a shared secret, as one might imagine in a large quantum network. If Alice and Bob have never met, but both know and trust Charlie, they can use him as a trusted intermediary for QKD. If Charlie cheats, then the channel between Alice and Bob is compromised. Otherwise, it is secure. Once Alice and Bob have successfully communicated (without eavesdropping) via Charlie, they then have a shared secret and no longer need a trusted intermediary.
I say this as someone who is no fan of Microsoft, and who is actually a student at the University of California--Microsoft's loss is bad for open source. Microsoft was simply the biggest fish Eolas could go after. Now they're going to demand royalties from all other browser makers, which could spell big trouble for Mozilla.
I hope I'm wrong--please tell me this isn't going to kill open source web browsers.
The UC/Eolas patent covers "a system allowing a user of a browser program... to access and execute an embedded program object." Sounds like it would cover browser support for Java and perhaps JavaScript embedded in web pages.
There are plenty of European/Western scientists, that most would consider some of the greatest scientists in the world, believed in a Christian view of God
This is very true. Take Charlie Townes, the nobel laureate who invented the maser and essentially also the laser. He also recently won the 2005 Templeton Prize for Progress Toward Research or Discoveries about Spiritual Realities (it's worth about $1.5 million).
Here's an interview with Professor Townes that discusses religion and intelligent design.
It has everything to do with the fact that our other friends who became doctors, lawyers, MBAs, etc are making more money, at younger ages, than we will.
Yes! I'm also a young scientist (physics grad student). You're right, although it's not just the money. It's the sacrifices you have to make in terms of family and having a life. For instance, my supervisor nearly forgot her own (young) son's birthday. She also had to carefully plan her pregnancy to coincide with tenure decisions, and had to wait a long time to have kids (which increases the risk of lots of problems).
Then there is the lack of jobs (if there aren't enough scientists, why aren't there jobs for all the current scientists?). If I wanted to end up as a medium-paid programmer, I wouldn't get a physics PhD to do it. There are much easier ways.
Then there's the slave labor that's expected of many grad students (I have a friend who was working 70+ hours a week who was told he needed to work even more).
One of my professors told me that you should only go into physics* if you love it and can't bear the thought of not doing it. He's right, except that I would add that you shouldn't do it unless you love it more than anything else. I have a feeling I won't make a really good physicist because I refuse to put my career ahead of family. One might say that this is true of many professions, except that you can make a very comfortable living in almost any city as a mediocre doctor or lawyer, whereas you have very few options as a mediocre scientist. You'll be lucky to get a job as an untenured instructor making 40k in Cornfield State University, Generic Midwestern State, and you'll be stuck teaching unmotivated students while having zero time for research, which is probably the reason you got into physics in the first place.
* this probably applies to most other sciences in addition to physics.
This is why we have so few Americans going into science.
People lump all sorts of things together under the label of "anti-science", when they are, in fact, very different things.
Intelligent design, for instance, really is an anti-science viewpoint. People are opposed to evolutionary science because they don't like the conclusions. Opposition to fetal stem cell research, on the other hand, stems from opposition to the methods. It is a self-consistent viewpoint (although one I do not hold) that "personhood" begins at conception, and that it is not acceptable to destroy or kill innocent persons except under extraordinary circumstances. The Terri Shiavo case was also not anti-science, as there were concerns that she wasn't really in a persistent vegitative state (no fMRI had been done), and the appointment of her husband as her guardian was of dubious legality.
In general, holding a broader definition of human life / personhood, and opposing actions that harm such broadly-defined human life, is not an inherently anti-science viewpoint. Until science can give a definitive answer to what does and does not qualify as a person, it is perfectly reasonable to give the benefit of the doubt and have a broad definition of personhood.
My point is that much of what is held up as "anti-science" in the US is in no way such. The anti-science attitudes that do exist towards evolution are by no means new--the creationism vs. Darwin battle is very old, and just has a new name. I don't think things have really gotten worse.
What I do have a problem with is when I only get a 3-4% raise, yet, executives can give themselves 50% raises, 4 million dollar bonuses, etc. There is nothing a CEO can produce that warrants that level of compensation. PERIOD.
Now, there are lots of CEOs out there who don't deserve their salaries, but the idea that no CEO is worth a couple million (or even ten million) a year is crazy. The difference between a really good CEO and a really bad CEO is the difference between going out of business and making record profits of billions of dollars. Just look at Apple before the return of Steve Jobs (when the media were constantly claiming Apple is dying!), versus the wildly-successful, iPod selling Apple we have now. Think it's a fluke? Look at Pixar. The man has the magic touch.
Very roughly, Apple's share price has increased ten-fold since the bad days of the late nineties. AAPL has a market cap of 47.4 billion dollars now, so that means that AAPL increased in value by about 40 billion under the stewardship of Jobs. Let's give him one percent of that--that works out to about 40 million a year.
You'd be hard-pressed to find a janitor who makes that kind of contribution to his company.
As for your contention that hard materials are not expensive, can you please present an appropriate counter-example?
I'm not the original poster, but I would be happy to provide an example. I used to use sapphire wafers as a substrate for experiments I did as an undergrad. The sapphire wafers were about 5 cm in diameter (much bigger than a watch face), and were "research grade". I'm not sure what "research grade" implied as to purity, but one face of the wafer was polished to a couple of nanometers RMS roughness. For comparison, to have a surface appear smooth, an RMS roughness of a hundred nanometers or so would be sufficient. The sapphire wafers cost about $85 in bulk (we paid a bit over a hundred per wafer, since we just ordered a small number). Since the face of a watch is smaller than 5 cm, and the specs required are much less demanding, it should be possible to mass produce sapphire watch faces for $10 - $20, and certainly less than $50. Same goes for iPods. Now, $20 is a lot for a low-margin consumer product like the iPod nano. I just got one, and I probably wouldn't want to pay an extra $20 - $30 for a scratch-proof screen.
As to the question of breakage, sapphire is also incredibly strong. The wafers we had were less than a millimeter thick (either 0.7 or 0.5 mm, I forget), and it was extremely hard to break them with your hands. I used a diamond saw to cut them to the size we needed...
If anyone is really interested, I could probably post some AFM (atomic force microscope) images I took of the wafer surface to show just how smooth it really is. We had to verify the smoothness for the experiments we were doing...
Because that's the whole freakin point of voting in the first place, that everybody gets one vote regardless of whether you agree with them ideologically or not, or whether you like them, or whether they have the most money or power.
You miss the point. People have natural rights. Governments don't.
Governments only have legitimate authority insofar as they are given authority by citizens. Governments should only be allowed to vote at the UN as representatives of their people. A non-representative government, such as China's, should not have a vote.
I can't stress this point enough, because it's something even democracies are starting to overlook. Take the EU, for instance. The power structures of the EU are increasingly distant from the people of Europe. When French and Dutch voters rejected the EU constitution, politicians talked about how the voters were making a mistake and about how to proceed anyway. The arrogance is incredible.
It's not even that I have to pay the tax, if everyone else had to, it's that now I have to be concerned about not just the tax laws of each state, but the tax laws of each county in each state. It's ridiculous. So much for "state's rights".
I totally agree. My wife is starting a small business selling custom portraits online. This kind of multi-state tax accounting complexity could easily kill it.
We've had one meltdown in the commercial reactors in the US that was due to not following procedure and about ~30 something things going wrong simultaniously. Radiation released to the public was about the amount you'd get on a couple cross country flights. We don't have a problem with this in the US.
No, we have not had any meltdowns in the US. A meltdown is when the reactor core overheats, and you get molten fissile material burning a hole through the bottom of the reactor.
Three Mile Island was not a meltdown, it was a fairly small (intentional) release of radioactive gas*, done to avoid the possibility of an explosion of hydrogen that operators thought might have been generated by high-temperature steam that was released through a series of other problems and errors. Even if the operators hadn't released that gas, and there had been an explosion, it almost certainly still would not have lead to a meltdown. Unlike reactors designed by Soviet communist fools, American reactors do not operate in or near a regime with positive feedback. Canadian CANDU reactors are even safer, as the moderator required for the reaction to happen (heavy water, or D2O) is also the coolant. If something goes wrong, it boils off, and the reaction stops before anything gets too out of hand. Pebble bed reactors are even safer--as I understand it, they operate in a regime where Doppler broadening at high temperatures decreases the neutron capture cross-section enough to stop the reaction. The point is that Three Mile Island wasn't actually a very dangerous failure, and that it wasn't close to being a meltdown. It was bad, but probably also a worst case for a US reactor.
As others have pointed out, it is true that nuclear is more expensive than natural gas-generated power, however the cost of natural gas power depends primarily on the cost of natural gas (whereas uranium is a small portion of the cost with nuclear power). If you also include a reasonable carbon tax, nuclear can start looking pretty good. It's the only serious non-CO2 producing candidate for baseline electricity production. Wind and solar can effectively be used supplementally, but as a baseline source, you'd have to factor in the cost of storing power for use at night or during cloudy or calm periods, and that's going to be extremely expensive. Most reasonable proponents of wind power will tell you it's not ever going to make up more than 10 or 15% of US power, even in a best case. With solar, you can do some simple calculations based on the solar radiation flux and realize that the land area required for it to replace most or all of our energy needs would be absurd (as in, by a couple decades from now, we'd have to cover an area larger than California with solar). Hydro is great, but there are a fixed number of rivers around to dam, and it's very hard politically. There's no way, for instance, that Hetch Hetchy could be dammed today--in fact, that dam may eventually get removed for environmental reasons.
*the amount of radioactive material released during the Three Mile Island incident was such that, if you were standing immediately outside the power plant's outer fence during the whole thing, you would have received a dose equal to a normal year's worth of background radiation in the US, or about 3 months' worth of background in France. Background radiation varies subtantially by geographic region due to naturally occuring radioactive elements in the soil.
One more thing for the spelling Nazis who were picking on someone for spelling Chernobyl as "Chernoble": since Russia and the Ukraine use a different alphabet than we do, English translations of place names are just transliterations. For some names (such as Chebychev, aka Chebyshev, aka Tchebychev), there are several common English spellings. It may well be the case that Chernoble is simply a less common transliteration. I don't know, and I suspect you don't, either, so give the guy a break.
I second this. We run OS X Server 10.4.2 on an Xserve G5 to provide directory, authentication, and home directory / file sharing to about 15 Linux and Windows clients. The Xserve G5 might sound like overkill, but there's some light scientific computing / numerical work that happens on these machines, which can generate substantial loads. It's funny, but OS X Server was actually the best match for providing these services to our heterogenous Windows/Linux environment.
OS X Server 10.4 still isn't perfect. Examples: Software RAID on OSX Server is crap--save yourself the trouble and get a hardware RAID (either an external device, or an Xserve with hardware RAID card). OS X Server is not quite as polished as a longtime Apple user might expect, but better than most alternatives. The 10.3 --> 10.4 upgrade was a bit problematic--the recommended approach is to do a clean install of the new version, followed by re-importing the user database, but this doesn't preserve user passwords.
Even with these flaws, OS X Server is an inexpensive, scalable, reliable, easy-to-setup option. It plays well with Windows, OS X, and most* unix-like OSes. If, at some point, you want to switch to a cobbled-together OSS solution, you'll likely find it much easier to do so from OS X than from a Microsoft setup, since much of OS X Server is already OSS.
* most means not FreeBSD 5.2.1. We encountered a bizarre incompatibility between OS X Server 10.4 and FreeBSD 5.2.1's NFS implementations that caused a race condition that broke KDE logins for users with NFS-shared home directories. KDE developers were stumped, and we ultimately had to switch to Linux from FreeBSD. The problem didn't exist with OS X 10.3.x. We attributed it to looseness in the NFS spec, which allows for different implementations to be compliant yet subtly incompatible.
Yes, OS X Server supports "pushing" policies out of the box for OS X clients. There's a range of methods you can use for managing OS X clients, including NetBooting, managed clients, and mobile home directories.
There may be a way to hack in management features for other OSs, but there's nothing simple I'm aware of.
We run OS X Server 10.4.2 on an Xserve G5 to provide directory, authentication, and home directory / file sharing to about 15 Linux and Windows clients. We used to use FreeBSD instead of Linux, but switched due to a variety of hassles with BSD (didn't play nice with our hardware, nor with OS X Server's NFS). The Xserve G5 might sound like overkill, but there's some light scientific computing / numerical work that happens on these machines, which can generate substantial loads.
As a developer of (specialized, scientific) Free Software, I have to say this sounds obnoxious and stupid, and I predict it leads to a "forking" of the GPL. The article was thin on details, but if it's as it sounds, I will not license my software under the new GPL. The next commit to my project's CVS is going to be changing
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.
to
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License version 2 as published by the Free Software Foundation.
I hate software patents as much as the next guy, and I think a lot of DRM implementations are stupid and evil, but it's not such an absolute that I want to tether my software to an ideological fight against those things.
For one, I *can* see some cases where DRM is not totally evil. Apple's iTunes Music Store (iTMS) is a good example--they struck a decent balance between usability and convincing the music execs the service was safe. Now, I know people have legitimate objections to the iTMS DRM (FairPlay), but if Apple hadn't gone with DRM, the iTMS would never have happened. I don't think Apple deserves to be penalized for that--they've done the best with the cards they were dealt. Apple has also made significant contributions to the Free Software community--do we really want to shut them out?
As another example, I do research in quantum computing. Suppose I develop some sort of new process or technology relating to quantum computing, and my university pushes for a patent on it. In the world of quantum computing, what is "software" and what is "hardware" is very fuzzy. Could I get shut out of using GPL'd software over that? I might not even have a choice about whether to patent or not--I had to sign a patent agreement when I enrolled here.
The FSF should stick to their original mission for the GPL. By trying to make it too broad, they're going to hurt everyone.
Oh, and "cultural flat fee" with internet access? What the hell? This is like Canada's levy on blank recordable media that goes to pay Canadian "artists" like Celine Dion. I can't think of a worse system for music than putting the government in charge of it. If you thought the RIAA was bad, just wait until Uncle Sam is in charge. Get ready for the "War on Inappropriate Rap Lyrics", the "War on Peer-to-Peer Networks", etc.
Slashdot Mirror
Now, 30% + 30% + 20% != 100%. My first guess was that I got a total of 5 mods: 2 informatives, 2 overrateds, and 1 insightful, which, with my +1 Karma bonus, brings me to the current score of 3. The thing is, then it should be 40% / 40% / 20%. Is this just a math bug, or is there some sort of weird secret moderation going on that doesn't show up in the moderation report?
Note to mods: I intentionally did not use my karma bonus on this post, because it's sort of off-topic.
Step down from what? The FSF? He created it, he runs it, and it is his baby.
Yes, but it's time for him to move on. The best thing he could do for the Free Software movement is to step down. If he's a reasonable person with his ego under control, he'll realize that. Right now, he's become a caricature and his mere presence in any sort of official role hurts the movement. See the other post (sibling to yours, I believe) which describes the time he put on a robe and halo during a lecture to businessmen.
And the alternative? Mediocrity? Let me ask you this... Who is pushing back from the other side?
There are two ways to accomplish change: incrementally, or via revolution. Given that copyright never, ever registers in the top ten list of things voters care about, the chance of a revolution at the polls is almost zero. The odds of slashdotters leaving their basements, grabbing their pitchforks, and storming the RIAA are even smaller. Stallman doesn't get this, but the only option left to us is incremental change, and that requires pushing at the edge of the mainstream.
Being the most reasonable person in a public debate is often the way to victory. With the RIAA suing grandmothers, this shouldn't be hard, but Stallman is blowing it for us.
Several years ago, I heard Stallman speak at a lecture at my university. He was clearly very smart, and very driven by ideological goals. On top of all that, there was also a hint of that indefinable quality shared by most crazy people. Something about him was not quite right--I got a sense that his grasp on reality was slipping a bit. Maybe this is necessary for a person to make the kind of sacrifices he has, but it's a dangerous balance.
From reading the recent draft of the GPL v3, and the article attached to this story, I get a sense that he's slipped further. For instance, when he spoke at my university, he recognized that the best way to achieve your goals is to have limited, realistic goals, and focus on those. When people asked him about copyright on music or movies, he diplomatically dodged the question and said it was a separate issue from his Free Software philosophy, and he didn't want to address it. In the interview linked in TFA, he outright attacks copyright for these things. The GPL v3's attack on DRM is similar. Stallman has sacrificed the clarity and readibility of the GPL v2 in order to attack patents and DRM.
Now, maybe you agree with Stallman about copyright for music, etc. Even so, you should recognize that that puts you farther outside the mainstream, and it's much harder to change the mainstream when you're 1,000 miles away. If a bunch of Americans write letters to Congress demanding that copyright be abolished*, they will be ignored. If they ask that copyright law take a step back towards the original constitutional idea of limited (in time and power) protection to promote progress in science and the useful arts, that may actually get somewhere. It is vitally important that we sound reasonable.
Stallman has lost his sense of perspective and his grasp on reality. I think it's possible that he is now harmful to the Free Software movement, and the community needs to think about how to deal with this problem. If the community asked him to step down, would he?
* I know Stallman didn't outright call for the abolition of copyright. Still, the changes he wants (the freedom for anyone to distribute any published work) amount to nearly the same thing.
With option 1, I have some ability to do good.
Under option 2, I have no ability to do good.
With option 1, Google does some evil (censorship) in return for the opportunity to do some good (and make a profit). With option 2, they do no evil, but also give up the chance at doing some good.
It's important to note that option 1 is inconsistent with the philosophy of "do no evil", but not necessarily inconsistent with the philosophy of "don't be evil", since the latter could be seen as dependent only on the net moral balance of actions.
With that subtle point made, I'd now like to engage in an egregious violation of Godwin's Law, and liken Google's dilemma to that of a person faced with the choice of Nazi collaboration.
Let's say you are (unwillingly) a Nazi officer. You are given the option of taking command of Auschwitz, or refusing and allowing another to take command. If you were in charge, you could perhaps assist some Jews in escaping, maybe reduce the number executed somewhat, but you would still be involved in the execution of innocent people. Do you refuse?
What I've done is reformulate the fundamental moral dilemma (should you become complicit with evil in order to do some good) in a very different emotional context. It could be argued that the good/evil balance is different in the two cases, but that's not the point here (i.e. adjust the number of people saved from Nazis until the balance is comparable). It's also true that the magnitude of both the good and the evil in the Nazi case is probably greater (although not necessarily--Yahoo's cooperation with China has already gotten a journalist thrown in jail), but this is again not the point. I'm interested in hearing how people react to this dilemma, versus Google's actions.
If Google's failure to willingly cooperate means that they're completely blocked in China, then as far as the West influencing things it's worse.
Influence is a two-way street. Trade with China was supposed to export our values of freedom and democracy to them, but instead we're importing their corruption and contempt for human rights. Before exposure to China, Google was arguably one of the most morally upright major American companies. Now they're turning into scum.
To put it another way, if Google cooperates with China's censorship, they don't have a whole lot of ammunition when the US Department of Justice asks them for people's search records. This does affect you--it's not just about China.
US Representative Chris Smith (R. NJ) is planning to hold hearings into the cooperation of US internet companies with Chinese censorship. I just wrote my congresswoman and asked her to support these hearings. You should consider doing the same, if you are an American.
I also plan to write Google and tell them I am canceling my AdWords campaign because of their practices.
With all this talk of PKI and such, has anyone actually started planning for the collapse of RSA, Diffie-Hellman, and other forms of public-key crypto? We've never had a security proof of these systems (information-theoretic or even computational security), and, since the publication of Shor's algorithm, we now know they can be broken with a quantum computer. Perhaps there's also a classical algorithm for breaking them, but let's assume not. Quantum computers are probably a minimum of ten years away (more like twenty five years, but let's consider the realistic worst case of ten).
I'm guessing many of you have forward security requirements of longer than ten years. Medical records, financial data, and lots of other things probably need to stay secret for longer than that. That means that you need to start now on preparing for the collapse of public-key crypto. Has anyone in IT actually started doing this? Anyone even thought about it?
Disclaimer: IAAQCR (I am a quantum computing researcher).
As another poster pointed out, AltiVec is great (although it's only single-precision, dammit), and there are some nice AltiVec libraries for BLAS and for stuff people in bioinformatics like to play with.
Another advantage is that the G5 is very similar to the processors used in DataStar, the biggest machine at the San Diego Supercomputing Center. This can be handy for doing small-scale testing and optimization before going to the big time.
Still another advantage is OS X--a really nice GUI with Unix goodness under the hood. Finally, there's the convenient 1U xserve enclosure, which is good for DIY small-scale clusters.
Actually, Schneier specifically said in that interview that it makes public key cryptography insecure. He was referring to symmetric, private key cryptography when he stated that it doesn't make all cryptography insecure.
I took another look at the interview, and it's really not clear. In one paragraph, he says that quantum computers render public key crypto insecure, and, a few paragraps later, he makes the statement I quoted about quantum computers effectively only halving key length. He doesn't make any distinction between private or public-key crypto in this paragraph, and it's not clear from context. One could easily read it to mean that quantum computers effectively halve key length, thus making current public key crypto insecure, but allowing for the simple fix of doubling current public key crypto key lengths to compensate (which of course won't really help).
I'll grant him the benefit of the doubt and assume that he knows the difference. Still, he should be a lot more careful with public statements about this kind of stuff. People listen to him and God knows there are already enough slashdotters confused about quantum computing.
Oh, and Bruce Schneier, if you're reading this, sorry about saying nasty things about you.
I lost all respect for Schneier when he shot his mouth off about quantum computation in a 1999 Slashdot interview. Quoting Schneier:
This is completely wrong. Schneier has confused Grover's algorithm (which does something analogous to linear search, e.g. of a key space), with Shor's algorithm (prime factorization and discrete log). Shor's algorithm gives an exponential speedup over the best known classical algorithm, not a square-root speed-up (which is what you get from Grover's algorithm). To make things really clear, Shor's algorithm scales as O(n^3) of the number of bits of your key, and perhaps better if you optimize it carefully. If you go from 128 bits to 256 bits, your classical computer needs to get about 2^128 times more powerful (give or take a few orders of magnitude), while your quantum computer needs only to get about eight times more powerful.
Shor's algorithm was published around 1994 or 1995. For Schneier, a supposed cryptography expert, to be ignorant of a published attack against most public-key cryptosystems more than four years later is just inexcusable. This attack is not just of theoretical concern, as Schneier suggests. For anyone dealing with secrets where you need forward security of more than fifteen years, you have to worry about quantum computers and Shor's algorithms. I'm sure there are lots of people who want, say, their medical records to remain secret (especially things like abortions, STDs, etc.) for longer than that. If such people listened to Schneier, they'd be badly misinformed.
What's worse than Schneier's ignorance of such a crucial development in his own field of cryptography and cryptanalysis is his arrogance about it. He didn't just punt the question or say that he didn't know enough about quantum computers to be sure. He gave a very confident and very wrong answer.
I will lose all respect of the Canadian government if they actually try to implement such a device. I thought that the German Autobahn should've been a lesson to the world that it's not speed in itself that is dangerous, it's having unsafe cars being driven on unsafe roads by irresponsible people.
I agree that this GPS speed limiter is a tremendously stupid idea. That said, the rest of your comment is off the mark.
So in essence, the Germans are happily driving at 250+ Km/h on their autobahns without having significantly more accidents than us, because they have much higher standards when it comes to issuing drivers' licenses, they have suited roads, and their cars go through a very strict mechanical check-up every year, to make sure they are road-legal.
The fundamental issues here are geography, weather, and population density, not quality of cars and drivers. Canada != Germany. There are very few places in Canada where a significantly faster highway speed limit would be safe. Many of the highways in British Columbia, including a large chunk of the Trans-Canada, are twisty roads through the mountains with only one or two lanes in each direction. The western edge of Alberta is similar. The prairie provinces are flat enough and the roads there straight enough for high-speed travel, but the highways are often only two lane, undivided roads. In northern Ontario, the highways (including the Trans-Canada) are often twisty, undivided, two-lane roads. In southern Ontario, it's all one big traffic jam, and on the rare occasions when the highways are flowing freely, highway traffic travels at about 150 km/h (i.e. everyone ignores the posted speed limit, which is not enforced; this is the one place the limit could perhaps be increased). I haven't driven east of Ontario, but you get the picture. Some of this could be improved by building more/better roads, but the geography imposes severe limits everywhere but the prairies, and low population density makes it hard to justify the enormous expense of upgrading the roads.
Weather is another important factor. In the mountainous regions of B.C. and Alberta, things can change very, very quickly. Many crashes are caused by sudden weather changes and the resulting unsafe road conditions. If you're 200km from anything, and all of a sudden a blizzard kicks up and you have 20m visibilty, what do you do? Drive 25 km/h (roughly the maximum safe speed in those conditions) and spend a very cold night stuck in a blizzard on the highway, also risking getting rear-ended by someone who didn't see your tail-lights? Or do you drive faster so you can make it to a town before the snow builds up and the road becomes impassable, hoping you'll be able to stop in time if there's something on the road ahead? This isn't hypothetical--pretty much this situation happened to me last winter. Fortunately, the storm cleared up and I got home safely.
This isn't to say that bad drivers are not the cause of accidents in Canada--in BC, nearly a third of traffic fatalities are the result of drinking and driving, which is clearly something only very bad drivers do (but not something stricter drivers' license tests would catch). Penalties and enforcement of drinking and driving regulations are pretty strict, notwithstanding Premier Campbell's vacation to Hawaii, but it's still an issue.
My point is that you can't fairly claim that the German system is better or would work in Canada (or perhaps the United States). The circumstances are simply very different.
Quantum crypto is theoretically amusing, but it practice it's boring. It lets you connect two sites together with dedicated fiber optics and run unbreakable-via-physics crypto over it, but that's a lot more expensive and incovenient that using whatever transmission medium supports your communication speeds and running unbreakable-via-mathematics crypto over it.
By "unbreakable-via-mathematics", do you mean information-theoretic secure (e.g. secure against all attacks, independent of physics), or computationally-secure (secure against attacks using a reasonably-sized classical computer)? The issue is that there aren't any information-theoretic algorithms out there for key distribution, only for encryption once you have a shared secret key the size of the message (I'm talking about one-time pad encryption). If you mean computationally secure, then AFAIK there are no public-key algorithms known to be computationally secure against a classical computer, let alone a quantum computer. Even RSA was not known to be computationally secure--people just strongly suspected it was. For all we know, there may be an efficient classical attack against it.
Is all this paranoia? Not if you need forward security (e.g. your secrets still need to be secret in 20 years when we might have quantum computers), or if your secrets are valuable enough that people would be willing to hire genius mathematicians to try to break your cryptosystems. The first case is more common than you might think: consider medical records, for instance. If you're young now, and you get an abortion or whatever, that could come back to haunt you when you run for office in 20 years. Even normal people do need forward security for certain things.
I should say that my background is in quantum computing and physics, with a more limited understanding of cryptography. If you're aware of some algorithms for public-key cryptography that have been proven to be information-theoretic secure or even just computationally secure, I would love to hear about them. I don't know too much about public key beyond RSA, and would be interested to hear more about elliptic curve crypto. It would be interesting to try to see if it's vulnerable to the quantum fourier transform.
the degree of precision required would be many orders of magnitude greater than any observations of any physical laws have ever been in a real experiment)
This is not true. The fault-tolerance threshold--the error threshold below which an arbitrarily long quantum computation can be performed to arbitrary precision with only polynomial overhead--is estimated to be anywhere from 10^-1 to 10^-7, depending on the physical system and the error model.
Now, 10^-7 is pretty hard to reach, but we most certainly have performed physical experiments to that degree of accuracy. Just not quantum computers, yet. It's important to note that there are other approaches to fault-tolerant quantum computation beyond just error correcting codes. Topologically fault-tolerant quantum computing is currently highly theoretical, but has a lot of potential.
In fact, part of the power of quantum computing is that (even without the somewhat less plausible factoring algorithm) we would have real secure encryption
Quantum cryptography, or, more correctly, quantum key distribution (QKD), only does a subset of what today's public key cryptography does. QKD requires that the two parties already have a small shared secret for authentication, and simply allows those parties to produce an arbitrarily large shared secret (e.g. random number which they both know, but is hidden from all other parties). That large shared secret can be used as a one-time pad. In other words, QKD solves the encryption problem, but not the authentication problem.
One of the sibling posts makes an incorrect statement about QKD:
The problem with quantum crypto is that it requires a direct transmission of photons from Alice to Bob. You can't have a relay station in-between, unless you are willing to guarantee its security (any relay station would allow for interception of the signal when it isn't entangled - which cannot be detected).
This is most certainly not the case if Alice and Bob have a shared secret, unless you use a particularly boneheaded implementation of QKD. QKD requires only a quantum and a classical channel between Alice and Bob. It doesn't matter if that channel is a single fiber, or a series of repeaters doing quantum teleportation and entanglement swapping, or whatever. The whole point of QKD is to allow Alice and Bob to detect tampering with the quantum channel, and that works regardless of the physical nature of the channel.
The catch is when Alice and Bob don't have a shared secret, as one might imagine in a large quantum network. If Alice and Bob have never met, but both know and trust Charlie, they can use him as a trusted intermediary for QKD. If Charlie cheats, then the channel between Alice and Bob is compromised. Otherwise, it is secure. Once Alice and Bob have successfully communicated (without eavesdropping) via Charlie, they then have a shared secret and no longer need a trusted intermediary.
References:
Aharonov et al
Raussendorf et al
Silva et al
Szkopek et al
Knill et al
Aliferis et al
You must be one old fratboy
Grad student, actually. Like a fratboy, but without the fun and drunkeness...
I say this as someone who is no fan of Microsoft, and who is actually a student at the University of California--Microsoft's loss is bad for open source. Microsoft was simply the biggest fish Eolas could go after. Now they're going to demand royalties from all other browser makers, which could spell big trouble for Mozilla.
... to access and execute an embedded program object." Sounds like it would cover browser support for Java and perhaps JavaScript embedded in web pages.
I hope I'm wrong--please tell me this isn't going to kill open source web browsers.
The UC/Eolas patent covers "a system allowing a user of a browser program
Crap.
There are plenty of European/Western scientists, that most would consider some of the greatest scientists in the world, believed in a Christian view of God
This is very true. Take Charlie Townes, the nobel laureate who invented the maser and essentially also the laser. He also recently won the 2005 Templeton Prize for Progress Toward Research or Discoveries about Spiritual Realities (it's worth about $1.5 million).
Here's an interview with Professor Townes that discusses religion and intelligent design.
It has everything to do with the fact that our other friends who became doctors, lawyers, MBAs, etc are making more money, at younger ages, than we will.
Yes! I'm also a young scientist (physics grad student). You're right, although it's not just the money. It's the sacrifices you have to make in terms of family and having a life. For instance, my supervisor nearly forgot her own (young) son's birthday. She also had to carefully plan her pregnancy to coincide with tenure decisions, and had to wait a long time to have kids (which increases the risk of lots of problems).
Then there is the lack of jobs (if there aren't enough scientists, why aren't there jobs for all the current scientists?). If I wanted to end up as a medium-paid programmer, I wouldn't get a physics PhD to do it. There are much easier ways.
Then there's the slave labor that's expected of many grad students (I have a friend who was working 70+ hours a week who was told he needed to work even more).
One of my professors told me that you should only go into physics* if you love it and can't bear the thought of not doing it. He's right, except that I would add that you shouldn't do it unless you love it more than anything else. I have a feeling I won't make a really good physicist because I refuse to put my career ahead of family. One might say that this is true of many professions, except that you can make a very comfortable living in almost any city as a mediocre doctor or lawyer, whereas you have very few options as a mediocre scientist. You'll be lucky to get a job as an untenured instructor making 40k in Cornfield State University, Generic Midwestern State, and you'll be stuck teaching unmotivated students while having zero time for research, which is probably the reason you got into physics in the first place.
* this probably applies to most other sciences in addition to physics.
This is why we have so few Americans going into science.
The U.S. is not becoming anti-science.
People lump all sorts of things together under the label of "anti-science", when they are, in fact, very different things.
Intelligent design, for instance, really is an anti-science viewpoint. People are opposed to evolutionary science because they don't like the conclusions. Opposition to fetal stem cell research, on the other hand, stems from opposition to the methods. It is a self-consistent viewpoint (although one I do not hold) that "personhood" begins at conception, and that it is not acceptable to destroy or kill innocent persons except under extraordinary circumstances. The Terri Shiavo case was also not anti-science, as there were concerns that she wasn't really in a persistent vegitative state (no fMRI had been done), and the appointment of her husband as her guardian was of dubious legality.
In general, holding a broader definition of human life / personhood, and opposing actions that harm such broadly-defined human life, is not an inherently anti-science viewpoint. Until science can give a definitive answer to what does and does not qualify as a person, it is perfectly reasonable to give the benefit of the doubt and have a broad definition of personhood.
My point is that much of what is held up as "anti-science" in the US is in no way such. The anti-science attitudes that do exist towards evolution are by no means new--the creationism vs. Darwin battle is very old, and just has a new name. I don't think things have really gotten worse.
What I do have a problem with is when I only get a 3-4% raise, yet, executives can give themselves 50% raises, 4 million dollar bonuses, etc. There is nothing a CEO can produce that warrants that level of compensation. PERIOD.
Now, there are lots of CEOs out there who don't deserve their salaries, but the idea that no CEO is worth a couple million (or even ten million) a year is crazy. The difference between a really good CEO and a really bad CEO is the difference between going out of business and making record profits of billions of dollars. Just look at Apple before the return of Steve Jobs (when the media were constantly claiming Apple is dying!), versus the wildly-successful, iPod selling Apple we have now. Think it's a fluke? Look at Pixar. The man has the magic touch.
Very roughly, Apple's share price has increased ten-fold since the bad days of the late nineties. AAPL has a market cap of 47.4 billion dollars now, so that means that AAPL increased in value by about 40 billion under the stewardship of Jobs. Let's give him one percent of that--that works out to about 40 million a year.
You'd be hard-pressed to find a janitor who makes that kind of contribution to his company.
As for your contention that hard materials are not expensive, can you please present an appropriate counter-example?
I'm not the original poster, but I would be happy to provide an example. I used to use sapphire wafers as a substrate for experiments I did as an undergrad. The sapphire wafers were about 5 cm in diameter (much bigger than a watch face), and were "research grade". I'm not sure what "research grade" implied as to purity, but one face of the wafer was polished to a couple of nanometers RMS roughness. For comparison, to have a surface appear smooth, an RMS roughness of a hundred nanometers or so would be sufficient. The sapphire wafers cost about $85 in bulk (we paid a bit over a hundred per wafer, since we just ordered a small number). Since the face of a watch is smaller than 5 cm, and the specs required are much less demanding, it should be possible to mass produce sapphire watch faces for $10 - $20, and certainly less than $50. Same goes for iPods. Now, $20 is a lot for a low-margin consumer product like the iPod nano. I just got one, and I probably wouldn't want to pay an extra $20 - $30 for a scratch-proof screen.
As to the question of breakage, sapphire is also incredibly strong. The wafers we had were less than a millimeter thick (either 0.7 or 0.5 mm, I forget), and it was extremely hard to break them with your hands. I used a diamond saw to cut them to the size we needed...
If anyone is really interested, I could probably post some AFM (atomic force microscope) images I took of the wafer surface to show just how smooth it really is. We had to verify the smoothness for the experiments we were doing...
Because that's the whole freakin point of voting in the first place, that everybody gets one vote regardless of whether you agree with them ideologically or not, or whether you like them, or whether they have the most money or power.
You miss the point. People have natural rights. Governments don't.
Governments only have legitimate authority insofar as they are given authority by citizens. Governments should only be allowed to vote at the UN as representatives of their people. A non-representative government, such as China's, should not have a vote.
I can't stress this point enough, because it's something even democracies are starting to overlook. Take the EU, for instance. The power structures of the EU are increasingly distant from the people of Europe. When French and Dutch voters rejected the EU constitution, politicians talked about how the voters were making a mistake and about how to proceed anyway. The arrogance is incredible.
It's not even that I have to pay the tax, if everyone else had to, it's that now I have to be concerned about not just the tax laws of each state, but the tax laws of each county in each state. It's ridiculous. So much for "state's rights".
I totally agree. My wife is starting a small business selling custom portraits online. This kind of multi-state tax accounting complexity could easily kill it.
We've had one meltdown in the commercial reactors in the US that was due to not following procedure and about ~30 something things going wrong simultaniously. Radiation released to the public was about the amount you'd get on a couple cross country flights. We don't have a problem with this in the US.
No, we have not had any meltdowns in the US. A meltdown is when the reactor core overheats, and you get molten fissile material burning a hole through the bottom of the reactor.
Three Mile Island was not a meltdown, it was a fairly small (intentional) release of radioactive gas*, done to avoid the possibility of an explosion of hydrogen that operators thought might have been generated by high-temperature steam that was released through a series of other problems and errors. Even if the operators hadn't released that gas, and there had been an explosion, it almost certainly still would not have lead to a meltdown. Unlike reactors designed by Soviet communist fools, American reactors do not operate in or near a regime with positive feedback. Canadian CANDU reactors are even safer, as the moderator required for the reaction to happen (heavy water, or D2O) is also the coolant. If something goes wrong, it boils off, and the reaction stops before anything gets too out of hand. Pebble bed reactors are even safer--as I understand it, they operate in a regime where Doppler broadening at high temperatures decreases the neutron capture cross-section enough to stop the reaction. The point is that Three Mile Island wasn't actually a very dangerous failure, and that it wasn't close to being a meltdown. It was bad, but probably also a worst case for a US reactor.
As others have pointed out, it is true that nuclear is more expensive than natural gas-generated power, however the cost of natural gas power depends primarily on the cost of natural gas (whereas uranium is a small portion of the cost with nuclear power). If you also include a reasonable carbon tax, nuclear can start looking pretty good. It's the only serious non-CO2 producing candidate for baseline electricity production. Wind and solar can effectively be used supplementally, but as a baseline source, you'd have to factor in the cost of storing power for use at night or during cloudy or calm periods, and that's going to be extremely expensive. Most reasonable proponents of wind power will tell you it's not ever going to make up more than 10 or 15% of US power, even in a best case. With solar, you can do some simple calculations based on the solar radiation flux and realize that the land area required for it to replace most or all of our energy needs would be absurd (as in, by a couple decades from now, we'd have to cover an area larger than California with solar). Hydro is great, but there are a fixed number of rivers around to dam, and it's very hard politically. There's no way, for instance, that Hetch Hetchy could be dammed today--in fact, that dam may eventually get removed for environmental reasons.
*the amount of radioactive material released during the Three Mile Island incident was such that, if you were standing immediately outside the power plant's outer fence during the whole thing, you would have received a dose equal to a normal year's worth of background radiation in the US, or about 3 months' worth of background in France. Background radiation varies subtantially by geographic region due to naturally occuring radioactive elements in the soil.
One more thing for the spelling Nazis who were picking on someone for spelling Chernobyl as "Chernoble": since Russia and the Ukraine use a different alphabet than we do, English translations of place names are just transliterations. For some names (such as Chebychev, aka Chebyshev, aka Tchebychev), there are several common English spellings. It may well be the case that Chernoble is simply a less common transliteration. I don't know, and I suspect you don't, either, so give the guy a break.
I second this. We run OS X Server 10.4.2 on an Xserve G5 to provide directory, authentication, and home directory / file sharing to about 15 Linux and Windows clients. The Xserve G5 might sound like overkill, but there's some light scientific computing / numerical work that happens on these machines, which can generate substantial loads. It's funny, but OS X Server was actually the best match for providing these services to our heterogenous Windows/Linux environment.
OS X Server 10.4 still isn't perfect. Examples:
Software RAID on OSX Server is crap--save yourself the trouble and get a hardware RAID (either an external device, or an Xserve with hardware RAID card). OS X Server is not quite as polished as a longtime Apple user might expect, but better than most alternatives. The 10.3 --> 10.4 upgrade was a bit problematic--the recommended approach is to do a clean install of the new version, followed by re-importing the user database, but this doesn't preserve user passwords.
Even with these flaws, OS X Server is an inexpensive, scalable, reliable, easy-to-setup option. It plays well with Windows, OS X, and most* unix-like OSes. If, at some point, you want to switch to a cobbled-together OSS solution, you'll likely find it much easier to do so from OS X than from a Microsoft setup, since much of OS X Server is already OSS.
* most means not FreeBSD 5.2.1. We encountered a bizarre incompatibility between OS X Server 10.4 and FreeBSD 5.2.1's NFS implementations that caused a race condition that broke KDE logins for users with NFS-shared home directories. KDE developers were stumped, and we ultimately had to switch to Linux from FreeBSD. The problem didn't exist with OS X 10.3.x. We attributed it to looseness in the NFS spec, which allows for different implementations to be compliant yet subtly incompatible.
Yes, OS X Server supports "pushing" policies out of the box for OS X clients. There's a range of methods you can use for managing OS X clients, including NetBooting, managed clients, and mobile home directories.
There may be a way to hack in management features for other OSs, but there's nothing simple I'm aware of.
We run OS X Server 10.4.2 on an Xserve G5 to provide directory, authentication, and home directory / file sharing to about 15 Linux and Windows clients. We used to use FreeBSD instead of Linux, but switched due to a variety of hassles with BSD (didn't play nice with our hardware, nor with OS X Server's NFS). The Xserve G5 might sound like overkill, but there's some light scientific computing / numerical work that happens on these machines, which can generate substantial loads.
to
I hate software patents as much as the next guy, and I think a lot of DRM implementations are stupid and evil, but it's not such an absolute that I want to tether my software to an ideological fight against those things.
For one, I *can* see some cases where DRM is not totally evil. Apple's iTunes Music Store (iTMS) is a good example--they struck a decent balance between usability and convincing the music execs the service was safe. Now, I know people have legitimate objections to the iTMS DRM (FairPlay), but if Apple hadn't gone with DRM, the iTMS would never have happened. I don't think Apple deserves to be penalized for that--they've done the best with the cards they were dealt. Apple has also made significant contributions to the Free Software community--do we really want to shut them out?
As another example, I do research in quantum computing. Suppose I develop some sort of new process or technology relating to quantum computing, and my university pushes for a patent on it. In the world of quantum computing, what is "software" and what is "hardware" is very fuzzy. Could I get shut out of using GPL'd software over that? I might not even have a choice about whether to patent or not--I had to sign a patent agreement when I enrolled here.
The FSF should stick to their original mission for the GPL. By trying to make it too broad, they're going to hurt everyone.
Oh, and "cultural flat fee" with internet access? What the hell? This is like Canada's levy on blank recordable media that goes to pay Canadian "artists" like Celine Dion. I can't think of a worse system for music than putting the government in charge of it. If you thought the RIAA was bad, just wait until Uncle Sam is in charge. Get ready for the "War on Inappropriate Rap Lyrics", the "War on Peer-to-Peer Networks", etc.