Oh, I can think of others. Is the sense of humour strict? If not, then the letter is false advertising. Is the act of rejecting an invitation a cease and desist of the parody (for once rejected, it is no longer parody but being parodied)? Is it even lawful for a lawyer to have a sense of humour?
There are 132 neolithic sites at the end of the Wye river, so it took me a while to find. There's a short writeup and some photos at The Modern Antiquarian. The Ordinance Survey coordinates are SO559120, and Yahoo maps places it somewhere along the Gloucester to Monmouth Road which I'm taking to be the Little Dean Road/Speech House Road, although the A4151/A4136 would also fit the description. I'm pretty bad at converting the OS system to long/lat, but if you have a calculator that can do that, you'll be better off using those.
The Long Stone description shows no indication of any archaeological findings and a reference by BBC Gloucester only talks about ley lines.
I seem to recall mentioning something about people not living in the houses. Oh, yes, there it is. I thought so.
The archaeologists are making an association on a basis for which the data they present to support said basis is actually contrary to the association they are using the data to present. This is not a "bald assertion", but comes from the well-known history of the area and from the well-known history of Skara Brae, not to mention the more recent development of DNA analysis to genetically identify populations. You can call me an armchair archaeologist all you like - and you'd have fair cause to, I'm not academically qualified in that discipline - but the fact remains that I am well aware of accepted archaeologically-documented timelines in both regions and also well aware of genetic archaeology conducted in both places. This new find does not obviously contradict any existing theory or existing piece of evidence, so I am asserting that the theories that exist are the more likely.
These archaeologists are proposing not an update of an existing theory, but a complete replacement for all current thinking. They are not tagging on just one more piece of knowledge, they are replacing all existing knowledge. Sometimes this is a good thing, but when it is done without any basis whatsoever, NOBODY should allow them to do so without challenge. If they can back their ideas up, great, but damnit, replacing tried-and-tested theories with something for which the best they can cite is a lack of evidence?! That may be how religions work, but that is NOT how science is done and I would thank them for keeping their paws off if they're not willing to meet the standards of their discipline.
If I mouth off about this, or any other subject, you might want to consider if there might be a reason - particularly when I make it clear that it is not on an emotional basis but on intellectual or professional grounds. I may be wrong - I'm not omniscient, by any means - but if I'm wrong, there will be clear, rational reasons for me being wrong, it will not be because the Gods have decreed it so. If I am right, that will also be for clear, rational reasons, and not because I have a four-digit UID. I find it nice on those rare occasions when someone disagrees with me on an intellectual level, I wish it happened more often.
What's so irrational about feeling a degree of kinship with neolithic geeks? Very hardy neolithic geeks, too. There's also an element of the unknown, too, which is something that also tends to appeal to those who are in any way intelligent or curious. And talking of curiosity, I'll bet you almost anything that nobody has carried out even a basic archaeological survey of the area.
Well, I wouldn't get too caught up in the theories from the archaeologists on this. The buildings are associated with Stonehenge by proximity in time and location, but only just. The area is littered with relics from the past (Avebury - which makes Stonehenge look like a roadside memorial, Silbury Hill - one of the largest man-made hills of ancient times, Woodhenge, the White Horse, the Giant, a veritable forest of longbarrows and roundbarrows, a giant meeting center roughly a hundred feet high and twice the area of a modern American Football field, etc.)
How can we be so sure this has anything to do with Stonehenge? The buildings are described as similar to those in Skara Brae, and those people were as close to indigenous as anyone in Britain gets. Yet we know that Stonehenge's builders displaced the Avebury builders by force and were a relatively modern society. The theory put forward simply doesn't add up. You can get cut shaving with Occam's Razor, but I think it's safe to apply it here - the theory linking this settlement to Stonehenge requires too much unnecessary complexity and should not be accepted.
To answer the question - the buildings were never abandoned as they were never really inhabited. They were summer houses, if you like - places for the rich and famous neolithic people to rub shoulders and get their cave paintings taken. The buildings were simply never returned to, at some point, possibly because the tribe's real home was wiped out in a fight, though it might be that the paparazzi ran out of ochre. The history of the area is confusing, though less because of the facts and more because of a desire to dramatize. There really isn't any need to make things sound more amazing than they really are, and all the archaeologists do when they do that is make themselves look stupid to anyone who knows even a little of the history of the region.
The first mistake is in assuming that the program has to understand concurrency. There are parallelizing compilers which will take standard C code, spot the parallelizable segments, and build those to run in parallel threads. Sure, GCC isn't one of them - OpenMP is manually-described parallelism - but there is no obvious reason why it couldn't do this. It already has a profile round that is followed by a build-to-profile optimization round, so why not have something similar for segmenting the code?
The second mistake is in assuming that all existing programs are, indeed, linear. PVM, MPI and BSP have been around for a long time now, and I spent plenty of time at University learning Occam (now Linuxified in the form of KROC). Oh, and for that matter there's OpenMP, which I've already mentioned. Everything that is based on RPC, Corba, DCL, or similar technologies is also parallel, albeit not quite to tightly-coupled. No, far as I can see, there's no shortage of parallel code. Maybe at IBM there is, but that's IBM's problem.
In the meantime, I'll set the OS to run in one code, the X server in another, the X client in a third, and supertuxkart in a fourth. It won't run any better, but at least I can pompously sneer at IBM's chief architect for their lack of understanding of how either people or the software they have use their machines.
I was thinking we should ask the US to parachute Darl McBride into the borderlands between Afghanistan and Pakistan. The worst that can happen is that he'll use up all of the Taliban's money suing half the middle east for infringing on spice-based intellectual property.
Och aye! Have ye some haggis with a wee dram or five? Nah, ye wurr tellin' me about this fiber you want installed. You wanted how much for it? Ach, you need more scotch. What was that ye said? Ye're sure ye don't want me to send a truck'o haggis to yer headquarters, then? Aboot this price. More scotch for the lad! Can't ye see he's parched! There. Ach, the piper has started his roonds. Ye look pale, lad. Have some more scotch, it'll do ye the world'o good. Gotta go? Come, lad, let's finish this deal. So you pay me a hundred thousand, install the cable all round Dundee, give gigabit to the homes, and free Internet for ten years. There. Signed. Now off ye go!
My understanding is that you are correct. The correlation is, I believe, along several different dimensions and not just a trivial one. The trivial case is that if you change some given number of bits in a message, over some distribution, then the changed bits in the hashed value should not be predictable in either number or distribution. The next-most trivial case is that if you add some new data to the end of a message, then the changed bits in the hashed value should also be unpredictable in number or distribution.
In addition to those two cases, what else can we say? Well, we can say that a cryptographically strong hash is a many-to-one mapping, where the set of cases that map to the same hash value should follow a random distribution AND where the set of differences between the aliases of any two hash values should also follow a random distribution.
(By random, I mean that the deltas aren't constant, cyclic, polynomial or otherwise predictable. You can't use any subset in order to predict any other subset. I also mean that for an infinite number of deltas, all possible delta values will occur with a frequency equal to the probability of that delta. In this case, you would want a flat probability, so all delta values will occur equally.)
Another way of saying all of this is to say that finding the aliases for a given input should be an NP-complete problem - the only way to find them is to look.
Anything else? Well, since most of this is extremely hard to test for, there needs to be a few tests that are more practical. The first would be that the hashing function should be in its simplest form. There should be no way of reducing it - for all inputs or for selected subsets of inputs. The second would be that the hashing function is non-differentiable (otherwise related inputs will produce related outputs). The third is that the function should be sensitive to initial conditions (for much the same reason). The fourth is that for any random selection of inputs where the pool of inputs is statistically significant, a standard statistical test for randomness should yield unimaginably high confidence limits that the distribution is indeed random. The chances of it not being random - ie: the chances of you finding an alias algorithmically and not herustically - should be no better than finding the alias by chance alone. (So, even if the hash could be broken, the chances of you finding the flaw are no better than of you finding the alias in the first place.)
I'm sure there are a million other tests that could be applied, but these are the more "obvious" ones that spring to mind. People can turn anything into cryptographic hashes (FFTs and even cellular automata have been used) - my guess is that chaotic systems might be another area of interest, provided you could guarantee the input was always in a chaotic region, as chaotic systems are only that way for specific conditions and can become periodic or stable outside of those limits, which is exactly what you don't want.
Oh, I completely agree with you and made sure to note in my post that the rules governing the withdrawl of habeus corpus do NOT authorize blanket suspension, even when the conditions are apparently met. And, certainly, as the Geneva Conventions are covered by Constitutional obligations, the Government is not authorized to suspend habeus corpus, no matter whether it is a privilege, right, or pretzel.
This is why it is so vitally important to get away from the idea of the Constitution authorizing freedoms to individuals, because if you look at it that way, such authorization can always be taken away. No, the ONLY safe way to regard what is written is to regard it as a law regarding what the Government may do and what freedoms the Government has. Which means that the Government is legally required to obey the Geneva Conventions. It has not been authorized the freedom to not obey them.
(Whether this applies to those sections that have been signed by enough nations to be regarded as International Law but not by the US specifically is a matter of debate. I would say yes, as it HAS signed up to the agreement that this is the case, and that means the US Government hasn't the authority to exclude those portions. It's a little dodgier in those cases where the US hasn't signed up to any of the agreement at all, but in a case where the US has accepted the bulk of the treaty AND has accepted that the remainder is indeed governing in matters of International Law, then I'm not convinced that its obligations under the Constitution permit it to reject those remaining parts in any way, shape or form.)
Yes and no. Everything in the Constitution is arguably a privilege, if you want to look at it from the perspective of what it bestows on individuals, as there are exceptions to just about every single amendment written, created by the Supreme Court over time. (The 2nd amendment is not regarded as applying to convicted felons, for example. I'm not arguing the rights or wrong of any given exception, the fact that they are considered to exist at all is enough to show that there are no inalienable rights in the United States, no matter what the Declaration of Independence might say on the issue.)
Personally, I don't look at the Constitution as bestowing anything on individuals. I look at it as the law that specifically applies to the US Government. This is what the Government is prohibited from doing, NOT what you are permitted to do. It's an important distinction, precisely because of such issues as illegal immigration, the war on terror, overseas detention, etc. If this is a law that the Government must obey, then the nationality and locality of any individual is of no importance, because it isn't about the individual, it is about the Government. The US Government is the US Government, whether it is dealing with a Mexican in Mexico or a Mexican in New York. Nothing changes, whether a suspect is detained in a US Government facility in San Fransisco or Spain, if viewed this way.
I admit that this is not a popular view of the Constitution, but it does have some basis. The Constitution was derived from the Magna Carta (which was specifically a law on the lawmakers and specifically included penalties that could be exacted on the King or anyone else in government who violated that law), and was specifically drafted for the purpose of establishing a Government. Not a country, not a legal system, not a rights system, a Government.
There are also plenty of reasons for accepting the established view, not least of which being that this is how it has always been interpreted by the Supreme Court. My views on the matter are infinitely less significant than those held by the people who get to rule on such matters, even if I think I'm right and that they're idiots. (They certainly think themselves just as right and would likely consider me no less an idiot.)
If we accept, for a moment, that my interpretation might be valid, then habeas corpus is automatically granted to ALL individuals (no matter where they are, no matter who they are) and must be explicitly suspended, which can only occur in the explicitly-stated conditions. The quoted section does NOT, however, say that it is suspendable under all such conditions or for all individuals at such times. You need to remember that the Revolution was partly on the grounds of improper imprisonment, so clearly the framers were very well aware that a crisis was possible in which habeas corpus should NOT be suspendable.
What does this mean? That we should grant such a privilege to everyone captured? Maybe, maybe no. The Constitution doesn't say that it can't be suspended, only that there are very strict conditions where that is lawful for the Government to do. I believe that this means that a challenge to such suspension through the courts is not merely legal but mandatory. Mandatory? Yes. Remember, I'm looking at this not as a law on individuals but on the Government. In the same way that for many serious crimes, the individual injured need not press charges (and in the case of murder, this would be a remarkable achievement), in the case of a serious governmental crime, the individual should not need to press charges. Why do you think we have separation of powers, if not to allow one such power to hold another accountable for governmental crimes? If no such accountability existed, what need would there be for separation? There would be nothing to account for.
As I said before, this is not a popular view. It means EVERYONE has the same inalienable rights, with respect to the US Government, as EVERYONE else, and virtually everyone can think of someone th
If you have a multi-core processor, you can switch which one you're running 'cpuburn' on and move the heat source. This allows for perfectly even cooking and reduce the burning of the strfry() in your wok.
There is no all-encompasing style for Open Source, Free Software, or any other variety of the beastie. There is no Universal Way, no Grand Master Plan that all must follow, and no guaranteed recipe for either success or failure. There is only code, tended to by a cooperative under the policies of that cooperative, for no benefit other than the scratching of a collective itch.
One of the very reasons the term "Open Source" was so heavily slammed in the early days was that it meant too many damn things to too many people (some of whom might also be damned). People, as a whole, adopted it despite those objections and often belittled those who raised them. Now we're finding out that some of those same people are finding out that Open Source does indeed too many different things to too many people, and that people really are trying to achieve different results. Congratulations. Should I break into applause or just do a Kerr Avon impression and throw these people out the airlock?
True. Now, if you had g(f(x, k), k')=x, you have the potential to have public key encryption with a stream cipher (all public key encryption systems that I know of are all block ciphers). A public key stream cipher could be very interesting.
There are cases where people use time-dependent (or use a dependency on some other parameter external to the data) stream ciphers. In those situations, the functions are not purely a function of x, but a function of x and these other parameters. Ultimately, it must still be possible to reverse the process, but you now have a dynamic mapping between (x, k) and (x', k'). The same input with the same key value would produce a different result for different values of the external parameter. This sort of stuff can seriously screw with people's minds.
It's to do with exponentials. Let's say that a particular snow crystal can form in one of fifteen ways. ((That's all the possibilities depicted on this chart). Then, two such crystals covers 225 possibilities (minus those that simply can't be joined for whatever reason). A snowflake with a hundred crystals would have fifteen to the hundred (ie: one googol) possible permutations.
However, is our starting number of 15 reasonable? The standard snowflake crystals are all formed at temperatures just below freezing under fairly normal conditions. The rate at which the water cools will have a major impact, as will any airborne particles around which the snow crystals can condense. (Particles may cause a break in the symmetry or may force the ice to contain patterns that simply aren't possible when only hexagonal ice crystals are present.) There again, anything dissolved in the water will change the chemistry as well. As not everything freezes at the same temperature, it is entirely possible for snowflakes to acquire bubbles and other oddities where something has remained liquid even as the water froze.
Then, there are the exotic states of frozen H2O which are not considered "ice", per se. Water that has frozen under really strange pressures or at extreme rates will not form regular ice crystals, but form other solid states instead. Slashdot has covered a few of these in the past. Is it possible to have a snowflake form from such states? Maybe. Then you add a whole new set of possibilities to the mix, although it would be unlikely that you could get a mixture of regular ice and these exotic states. (Not impossible, though. If the higher-level clouds chucked down snow in the exotic states, which then got added to by regular snowflake crystals, then you could indeed have a mixture. Not sure this could happen on Earth, but there may be planets where this is common.)
This gem first appeared on NASA AMES' webserver. It got featured on Slashdot at the time. After AMES was rebuilt from the wreckage left after their webserver exploded, all copies at NASA were purged. However, the Slashdot archives include the original link and writeup, so proof does exist that this truly is a NASA document.
Modern syncronous stream ciphers usually use XOR operations, but not all. Any symmetric mapping function will do, XOR is merely the simplest and fastest to build electronically. Older stream ciphers that used mechanical devices to do the mapping, or which used base-8 or base-10 electronics (those being fairly common designs in early digital electronic computers) used whatever mapping function the engineers could think of that worked well on their systems.
But, yeah, you could go out and program an FPGA or ASIC tomorrow with a symmetric function f() of your choice where f(f(x))=x and use that to drive a stream cipher. If you stacked nine such functions together, and had a plugboard that allowed you to switch the order of the mappings at will, you would basically have an electronic Enigma Machine.
The Wikipedia article has other thoughts on the subject, but seems to be focussed entirely on base-2 stream ciphers, whereas group theory and other areas of arcane mathematics don't get interesting until you get into the non-trivial group sizes.
Major shift? Conservative? This is the same NASA that broadcast NASA Select over CU-SeeMe and also the Multibone, allowed Donald Becker to develop network drivers for Linux, opened the source of a great many Computational Fluid Dynamics packages, promoted the early development of Beowulf Clusters, published guidelines on how to identify barbarian invaders, has hosted talks by SETI folks, investigated AJAX-style applications in the mid 1990s and was routinely helping the Open Source community years before any business would go near it?
Now, I think they make some extremely stupid decisions at times. I think that half their management is an extreme liability to their operations, the safety of their astronauts and the quality of their science. I also think they are desperately underfunded and have developed something of a siege mentality. However, "conservative" is not a term I'd associate with them, and they are most certainly familiar with "Open Standards" - having either invented them or were early adopters.
This is merely where they should have been all along, based on their own practices and their own connections with the IT industry. Far from calling it revolutionary, I'd consider it merely evolutionary.
No, that's just a pseudo one-time pad, applied to the XOR encryption algorithm. The encryption algorithm is still just the XOR encryption algorithm - there's no additional convolution - and the total pad is derivable from a much smaller input. With block ciphers, the system is called an encryption mode, and there are a multitude of them to choose from - so people usually don't. Always puzzled me why, particularly as the only reason to use ciphers is to secure the data. The more unknowns there are, the harder it is for an attacker to successfully attack.
(This is important - 2DEM's documentation shows that it is possible to inspect encrypted data using the trivial modes that are popular and obtain a significant amount of information about what has been encrypted. This is so not what encryption is supposed to be about.)
You're better off using algorithms that share nothing in common. SHA512 and Whirlpool would be good choices, from that standpoint. Besides, with MD5 effectively broken a long time ago (as hashes go), a collision only requires an attacker to find one flaw, not two overlapping flaws, as would be required with two unbroken hashes.
We know Cthulhu turns into a mist (Call of Cthulhu), we know he can't pass the elder sign and we know that the Chinese can etch entire names onto grains of rice. So, if we hire the entire of China to etch elder signs onto the sand used to make cement, summon Cthulhu into a flooded cavern, run a boat through him, then flood the cavern with the modified cement, you can prevent him reforming and eventually he'll go insane.
...it is vitally important to replace the existing system with something that works. The problem is, nobody really understands what would work - the quality of research is worse than the quality of the elections.
Ultimately, you want a system where true recounts aren't needed (but would be guaranteed, in full, if called for). This requires a system that is essentially non-partisan. There would be no quango (govt. appointee) in charge of running the elections or counting the votes. The separation of powers should ensure that a person cannot be elected by the same people they selected.
Some would argue that this is a case for secure electronic voting, provided the code was formally designed and thoroughly audited, and provided the votes cast were retained in some form (electronically or on paper), not merely tallied. Others would argue that it should require paper ballots but where each ballot box is under supervision so can't "go missing" (as often happens) and where each and every ballot is counted by three or more people - no statistical sampling, no "it fell under the table" and no "oh, I didn't think those mattered".
(In all cases, postal ballots should absolutely NEVER be handed to a politically aligned group for forwarding. In fact, if we're going to go with electronic ballots, postal ballots should not exist - you should be able to vote totally securely and totally anonymously - say via a tor-like setup - from any Internet-capable location. What we should not have is political parties able to dump ballots they don't like, which has happened and which will no doubt happen again.)
Cod in that part of the world is virtually extinct, due to overfishing. It's at 5% of the pre-1960s stock and at 1% of the pre-1900s size (based on assorted studies on decline in fish populations and variation that the BBC has linked to from time to time). On that basis, there simply aren't enough fish to satisfy the Icelandic medical industry and the demand for fresh cod, particularly by Britain and Spain.
The governments across Europe will do what any normal person would do when faced with the choice of a potentially life-saving cure for foreigners they will never know and a nice, fat check they can cash the same day. They'll make life hard on the researchers to stop them putting the country over-quota. They have to. If they don't and cod proves a useful source for new treatments, they risk losing out on the bribes they're taking. They won't make money off the new treatments, only the American drug companies will. They have zero incentive to allow this R&D and many reasons to prevent it.
(I might sound cynical, but that's only an illusion. This is text, so there's really no sound at all.)
Yes, no and maybe, but not necessarily in that order.
I attended the PCI SIG conference on virtualization for the new spec. There are two forms of virtualization that will (eventually) be supported - multiple operating systems on the same machine having access to their own private virtual PCI bus, and multi-mastered PCI busses where you can have multiple mainboards driving a virtual PCI bus that spans multiple machines.
The latter is a godsend for cluster builders - why bother with having tightly-coupled NICs on the far side of the PCI bus, when you can simply have the PCI bus carry everything directly to the endpoint? There's less work, less conversion, so less latency and fewer possibilities of data errors. Since it's 5 Gb/s per lane, you can also get speeds far in excess of those offered by most NIC vendors, if you're careful about how the bus divides up the bandwidth between nodes.
So, yes, I imagine the high-end HPC market will have machines that can do PCIe 2 in fairly short order.
Will anyone else see the benefits? Oh, I imagine high-end data centers will be interested. They can now double the width of striped disk arrays and not worry about bandwidth. Microsoft and VMWare will be urging rapid adoption, because of the virtualization abilities - it'll be faster than software virtualization and Microsoft gets to blame someone else if something goes wrong.
Intel will almost certainly be pushing the technology, as the spec allows for proprietary extensions. This means that they can build controllers that can work only with vendors they approve of (and AMD is unlikely to be one of them). Doesn't matter if Intel would or would not do that, what matters is that they can, which means if they're first to implement, they get a BIG stick to ward off rivals.
Slashdot Mirror
Oh, I can think of others. Is the sense of humour strict? If not, then the letter is false advertising. Is the act of rejecting an invitation a cease and desist of the parody (for once rejected, it is no longer parody but being parodied)? Is it even lawful for a lawyer to have a sense of humour?
The Long Stone description shows no indication of any archaeological findings and a reference by BBC Gloucester only talks about ley lines.
The archaeologists are making an association on a basis for which the data they present to support said basis is actually contrary to the association they are using the data to present. This is not a "bald assertion", but comes from the well-known history of the area and from the well-known history of Skara Brae, not to mention the more recent development of DNA analysis to genetically identify populations. You can call me an armchair archaeologist all you like - and you'd have fair cause to, I'm not academically qualified in that discipline - but the fact remains that I am well aware of accepted archaeologically-documented timelines in both regions and also well aware of genetic archaeology conducted in both places. This new find does not obviously contradict any existing theory or existing piece of evidence, so I am asserting that the theories that exist are the more likely.
These archaeologists are proposing not an update of an existing theory, but a complete replacement for all current thinking. They are not tagging on just one more piece of knowledge, they are replacing all existing knowledge. Sometimes this is a good thing, but when it is done without any basis whatsoever, NOBODY should allow them to do so without challenge. If they can back their ideas up, great, but damnit, replacing tried-and-tested theories with something for which the best they can cite is a lack of evidence?! That may be how religions work, but that is NOT how science is done and I would thank them for keeping their paws off if they're not willing to meet the standards of their discipline.
If I mouth off about this, or any other subject, you might want to consider if there might be a reason - particularly when I make it clear that it is not on an emotional basis but on intellectual or professional grounds. I may be wrong - I'm not omniscient, by any means - but if I'm wrong, there will be clear, rational reasons for me being wrong, it will not be because the Gods have decreed it so. If I am right, that will also be for clear, rational reasons, and not because I have a four-digit UID. I find it nice on those rare occasions when someone disagrees with me on an intellectual level, I wish it happened more often.
What's so irrational about feeling a degree of kinship with neolithic geeks? Very hardy neolithic geeks, too. There's also an element of the unknown, too, which is something that also tends to appeal to those who are in any way intelligent or curious. And talking of curiosity, I'll bet you almost anything that nobody has carried out even a basic archaeological survey of the area.
How can we be so sure this has anything to do with Stonehenge? The buildings are described as similar to those in Skara Brae, and those people were as close to indigenous as anyone in Britain gets. Yet we know that Stonehenge's builders displaced the Avebury builders by force and were a relatively modern society. The theory put forward simply doesn't add up. You can get cut shaving with Occam's Razor, but I think it's safe to apply it here - the theory linking this settlement to Stonehenge requires too much unnecessary complexity and should not be accepted.
To answer the question - the buildings were never abandoned as they were never really inhabited. They were summer houses, if you like - places for the rich and famous neolithic people to rub shoulders and get their cave paintings taken. The buildings were simply never returned to, at some point, possibly because the tribe's real home was wiped out in a fight, though it might be that the paparazzi ran out of ochre. The history of the area is confusing, though less because of the facts and more because of a desire to dramatize. There really isn't any need to make things sound more amazing than they really are, and all the archaeologists do when they do that is make themselves look stupid to anyone who knows even a little of the history of the region.
The second mistake is in assuming that all existing programs are, indeed, linear. PVM, MPI and BSP have been around for a long time now, and I spent plenty of time at University learning Occam (now Linuxified in the form of KROC). Oh, and for that matter there's OpenMP, which I've already mentioned. Everything that is based on RPC, Corba, DCL, or similar technologies is also parallel, albeit not quite to tightly-coupled. No, far as I can see, there's no shortage of parallel code. Maybe at IBM there is, but that's IBM's problem.
In the meantime, I'll set the OS to run in one code, the X server in another, the X client in a third, and supertuxkart in a fourth. It won't run any better, but at least I can pompously sneer at IBM's chief architect for their lack of understanding of how either people or the software they have use their machines.
I was thinking we should ask the US to parachute Darl McBride into the borderlands between Afghanistan and Pakistan. The worst that can happen is that he'll use up all of the Taliban's money suing half the middle east for infringing on spice-based intellectual property.
Och aye! Have ye some haggis with a wee dram or five? Nah, ye wurr tellin' me about this fiber you want installed. You wanted how much for it? Ach, you need more scotch. What was that ye said? Ye're sure ye don't want me to send a truck'o haggis to yer headquarters, then? Aboot this price. More scotch for the lad! Can't ye see he's parched! There. Ach, the piper has started his roonds. Ye look pale, lad. Have some more scotch, it'll do ye the world'o good. Gotta go? Come, lad, let's finish this deal. So you pay me a hundred thousand, install the cable all round Dundee, give gigabit to the homes, and free Internet for ten years. There. Signed. Now off ye go!
In addition to those two cases, what else can we say? Well, we can say that a cryptographically strong hash is a many-to-one mapping, where the set of cases that map to the same hash value should follow a random distribution AND where the set of differences between the aliases of any two hash values should also follow a random distribution.
(By random, I mean that the deltas aren't constant, cyclic, polynomial or otherwise predictable. You can't use any subset in order to predict any other subset. I also mean that for an infinite number of deltas, all possible delta values will occur with a frequency equal to the probability of that delta. In this case, you would want a flat probability, so all delta values will occur equally.)
Another way of saying all of this is to say that finding the aliases for a given input should be an NP-complete problem - the only way to find them is to look.
Anything else? Well, since most of this is extremely hard to test for, there needs to be a few tests that are more practical. The first would be that the hashing function should be in its simplest form. There should be no way of reducing it - for all inputs or for selected subsets of inputs. The second would be that the hashing function is non-differentiable (otherwise related inputs will produce related outputs). The third is that the function should be sensitive to initial conditions (for much the same reason). The fourth is that for any random selection of inputs where the pool of inputs is statistically significant, a standard statistical test for randomness should yield unimaginably high confidence limits that the distribution is indeed random. The chances of it not being random - ie: the chances of you finding an alias algorithmically and not herustically - should be no better than finding the alias by chance alone. (So, even if the hash could be broken, the chances of you finding the flaw are no better than of you finding the alias in the first place.)
I'm sure there are a million other tests that could be applied, but these are the more "obvious" ones that spring to mind. People can turn anything into cryptographic hashes (FFTs and even cellular automata have been used) - my guess is that chaotic systems might be another area of interest, provided you could guarantee the input was always in a chaotic region, as chaotic systems are only that way for specific conditions and can become periodic or stable outside of those limits, which is exactly what you don't want.
Oh, I completely agree with you and made sure to note in my post that the rules governing the withdrawl of habeus corpus do NOT authorize blanket suspension, even when the conditions are apparently met. And, certainly, as the Geneva Conventions are covered by Constitutional obligations, the Government is not authorized to suspend habeus corpus, no matter whether it is a privilege, right, or pretzel.
This is why it is so vitally important to get away from the idea of the Constitution authorizing freedoms to individuals, because if you look at it that way, such authorization can always be taken away. No, the ONLY safe way to regard what is written is to regard it as a law regarding what the Government may do and what freedoms the Government has. Which means that the Government is legally required to obey the Geneva Conventions. It has not been authorized the freedom to not obey them.
(Whether this applies to those sections that have been signed by enough nations to be regarded as International Law but not by the US specifically is a matter of debate. I would say yes, as it HAS signed up to the agreement that this is the case, and that means the US Government hasn't the authority to exclude those portions. It's a little dodgier in those cases where the US hasn't signed up to any of the agreement at all, but in a case where the US has accepted the bulk of the treaty AND has accepted that the remainder is indeed governing in matters of International Law, then I'm not convinced that its obligations under the Constitution permit it to reject those remaining parts in any way, shape or form.)
Personally, I don't look at the Constitution as bestowing anything on individuals. I look at it as the law that specifically applies to the US Government. This is what the Government is prohibited from doing, NOT what you are permitted to do. It's an important distinction, precisely because of such issues as illegal immigration, the war on terror, overseas detention, etc. If this is a law that the Government must obey, then the nationality and locality of any individual is of no importance, because it isn't about the individual, it is about the Government. The US Government is the US Government, whether it is dealing with a Mexican in Mexico or a Mexican in New York. Nothing changes, whether a suspect is detained in a US Government facility in San Fransisco or Spain, if viewed this way.
I admit that this is not a popular view of the Constitution, but it does have some basis. The Constitution was derived from the Magna Carta (which was specifically a law on the lawmakers and specifically included penalties that could be exacted on the King or anyone else in government who violated that law), and was specifically drafted for the purpose of establishing a Government. Not a country, not a legal system, not a rights system, a Government.
There are also plenty of reasons for accepting the established view, not least of which being that this is how it has always been interpreted by the Supreme Court. My views on the matter are infinitely less significant than those held by the people who get to rule on such matters, even if I think I'm right and that they're idiots. (They certainly think themselves just as right and would likely consider me no less an idiot.)
If we accept, for a moment, that my interpretation might be valid, then habeas corpus is automatically granted to ALL individuals (no matter where they are, no matter who they are) and must be explicitly suspended, which can only occur in the explicitly-stated conditions. The quoted section does NOT, however, say that it is suspendable under all such conditions or for all individuals at such times. You need to remember that the Revolution was partly on the grounds of improper imprisonment, so clearly the framers were very well aware that a crisis was possible in which habeas corpus should NOT be suspendable.
What does this mean? That we should grant such a privilege to everyone captured? Maybe, maybe no. The Constitution doesn't say that it can't be suspended, only that there are very strict conditions where that is lawful for the Government to do. I believe that this means that a challenge to such suspension through the courts is not merely legal but mandatory. Mandatory? Yes. Remember, I'm looking at this not as a law on individuals but on the Government. In the same way that for many serious crimes, the individual injured need not press charges (and in the case of murder, this would be a remarkable achievement), in the case of a serious governmental crime, the individual should not need to press charges. Why do you think we have separation of powers, if not to allow one such power to hold another accountable for governmental crimes? If no such accountability existed, what need would there be for separation? There would be nothing to account for.
As I said before, this is not a popular view. It means EVERYONE has the same inalienable rights, with respect to the US Government, as EVERYONE else, and virtually everyone can think of someone th
If you have a multi-core processor, you can switch which one you're running 'cpuburn' on and move the heat source. This allows for perfectly even cooking and reduce the burning of the strfry() in your wok.
One of the very reasons the term "Open Source" was so heavily slammed in the early days was that it meant too many damn things to too many people (some of whom might also be damned). People, as a whole, adopted it despite those objections and often belittled those who raised them. Now we're finding out that some of those same people are finding out that Open Source does indeed too many different things to too many people, and that people really are trying to achieve different results. Congratulations. Should I break into applause or just do a Kerr Avon impression and throw these people out the airlock?
Well, if they are all different, then they are all alike in being different, and therefore identical.
There are cases where people use time-dependent (or use a dependency on some other parameter external to the data) stream ciphers. In those situations, the functions are not purely a function of x, but a function of x and these other parameters. Ultimately, it must still be possible to reverse the process, but you now have a dynamic mapping between (x, k) and (x', k'). The same input with the same key value would produce a different result for different values of the external parameter. This sort of stuff can seriously screw with people's minds.
However, is our starting number of 15 reasonable? The standard snowflake crystals are all formed at temperatures just below freezing under fairly normal conditions. The rate at which the water cools will have a major impact, as will any airborne particles around which the snow crystals can condense. (Particles may cause a break in the symmetry or may force the ice to contain patterns that simply aren't possible when only hexagonal ice crystals are present.) There again, anything dissolved in the water will change the chemistry as well. As not everything freezes at the same temperature, it is entirely possible for snowflakes to acquire bubbles and other oddities where something has remained liquid even as the water froze.
Then, there are the exotic states of frozen H2O which are not considered "ice", per se. Water that has frozen under really strange pressures or at extreme rates will not form regular ice crystals, but form other solid states instead. Slashdot has covered a few of these in the past. Is it possible to have a snowflake form from such states? Maybe. Then you add a whole new set of possibilities to the mix, although it would be unlikely that you could get a mixture of regular ice and these exotic states. (Not impossible, though. If the higher-level clouds chucked down snow in the exotic states, which then got added to by regular snowflake crystals, then you could indeed have a mixture. Not sure this could happen on Earth, but there may be planets where this is common.)
This gem first appeared on NASA AMES' webserver. It got featured on Slashdot at the time. After AMES was rebuilt from the wreckage left after their webserver exploded, all copies at NASA were purged. However, the Slashdot archives include the original link and writeup, so proof does exist that this truly is a NASA document.
But, yeah, you could go out and program an FPGA or ASIC tomorrow with a symmetric function f() of your choice where f(f(x))=x and use that to drive a stream cipher. If you stacked nine such functions together, and had a plugboard that allowed you to switch the order of the mappings at will, you would basically have an electronic Enigma Machine.
The Wikipedia article has other thoughts on the subject, but seems to be focussed entirely on base-2 stream ciphers, whereas group theory and other areas of arcane mathematics don't get interesting until you get into the non-trivial group sizes.
Now, I think they make some extremely stupid decisions at times. I think that half their management is an extreme liability to their operations, the safety of their astronauts and the quality of their science. I also think they are desperately underfunded and have developed something of a siege mentality. However, "conservative" is not a term I'd associate with them, and they are most certainly familiar with "Open Standards" - having either invented them or were early adopters.
This is merely where they should have been all along, based on their own practices and their own connections with the IT industry. Far from calling it revolutionary, I'd consider it merely evolutionary.
(This is important - 2DEM's documentation shows that it is possible to inspect encrypted data using the trivial modes that are popular and obtain a significant amount of information about what has been encrypted. This is so not what encryption is supposed to be about.)
You're better off using algorithms that share nothing in common. SHA512 and Whirlpool would be good choices, from that standpoint. Besides, with MD5 effectively broken a long time ago (as hashes go), a collision only requires an attacker to find one flaw, not two overlapping flaws, as would be required with two unbroken hashes.
Oh.
Ultimately, you want a system where true recounts aren't needed (but would be guaranteed, in full, if called for). This requires a system that is essentially non-partisan. There would be no quango (govt. appointee) in charge of running the elections or counting the votes. The separation of powers should ensure that a person cannot be elected by the same people they selected.
Some would argue that this is a case for secure electronic voting, provided the code was formally designed and thoroughly audited, and provided the votes cast were retained in some form (electronically or on paper), not merely tallied. Others would argue that it should require paper ballots but where each ballot box is under supervision so can't "go missing" (as often happens) and where each and every ballot is counted by three or more people - no statistical sampling, no "it fell under the table" and no "oh, I didn't think those mattered".
(In all cases, postal ballots should absolutely NEVER be handed to a politically aligned group for forwarding. In fact, if we're going to go with electronic ballots, postal ballots should not exist - you should be able to vote totally securely and totally anonymously - say via a tor-like setup - from any Internet-capable location. What we should not have is political parties able to dump ballots they don't like, which has happened and which will no doubt happen again.)
The governments across Europe will do what any normal person would do when faced with the choice of a potentially life-saving cure for foreigners they will never know and a nice, fat check they can cash the same day. They'll make life hard on the researchers to stop them putting the country over-quota. They have to. If they don't and cod proves a useful source for new treatments, they risk losing out on the bribes they're taking. They won't make money off the new treatments, only the American drug companies will. They have zero incentive to allow this R&D and many reasons to prevent it.
(I might sound cynical, but that's only an illusion. This is text, so there's really no sound at all.)
I attended the PCI SIG conference on virtualization for the new spec. There are two forms of virtualization that will (eventually) be supported - multiple operating systems on the same machine having access to their own private virtual PCI bus, and multi-mastered PCI busses where you can have multiple mainboards driving a virtual PCI bus that spans multiple machines.
The latter is a godsend for cluster builders - why bother with having tightly-coupled NICs on the far side of the PCI bus, when you can simply have the PCI bus carry everything directly to the endpoint? There's less work, less conversion, so less latency and fewer possibilities of data errors. Since it's 5 Gb/s per lane, you can also get speeds far in excess of those offered by most NIC vendors, if you're careful about how the bus divides up the bandwidth between nodes.
So, yes, I imagine the high-end HPC market will have machines that can do PCIe 2 in fairly short order.
Will anyone else see the benefits? Oh, I imagine high-end data centers will be interested. They can now double the width of striped disk arrays and not worry about bandwidth. Microsoft and VMWare will be urging rapid adoption, because of the virtualization abilities - it'll be faster than software virtualization and Microsoft gets to blame someone else if something goes wrong.
Intel will almost certainly be pushing the technology, as the spec allows for proprietary extensions. This means that they can build controllers that can work only with vendors they approve of (and AMD is unlikely to be one of them). Doesn't matter if Intel would or would not do that, what matters is that they can, which means if they're first to implement, they get a BIG stick to ward off rivals.