Actually I always wonder why there are so many data centers placed at hot places... after all most of the time we don't care whether our data is sent half a globe in the East-West direction or North-South direction. unlucky for those living at those hot regions, the extra cost of air-conditioning to make it work also means that there are a lot more carbon emission and power consumption than really needed, both works to make their place even hotter.
> From a legal point of view material gained with a secret computer search shouldn't be brought to court, > because there is no way to prove that the evidence isn't faked.
There are lots of activities done by the police which is exactly like that: they know it is impossible to put the direct result to court as evidence for whatever reason ("the court know the police can fake it" is about the most minor one...), but the police would do it anyway. Reason? If they can tap into their suspect and as a result know something would happen in "real life" rather than electronically, they can organize an operation, and the result of that operation can be put to court. No court will be dissatisfied if the operation is done in response to some "undisclosed insiders' information".
Like you, I use NoScript and AdBlock Plus and won't see any of those sh*t. But I'm actually rather confused about why the hell did they do it. Are they really that stupid? Once everybody are forced to learn to do the same there is no business for ads business. Did somebody predict something in the lines that Microsoft will never make piracy checking bullet-proof because they would kill themselves that way by forcing everybody to Linux? Perhaps those Ads companies are really not that clever after all.
This sounds *really* wrong. You can say white-hats should have waited for a few days or even a few weeks after notifying the vendors before disclosing problems, but they should be disclosed eventually, and should be disclosed after giving vendors a reasonable amount of time. There bound to be people not upgrading their Windows, and there bound to be people not upgrading their Redhat or Fedora or Ubuntu or SuSE or FreeBSD or whatever operating system you name (not to mention whatever Firewalls, protocols, applications, etc, etc you name). So we shouldn't be disclosing any vulnerability about any of those?! Who, then, know that their software is vulnerable to black-hats and needs upgrading, and who, then, know which software vendor is more trust-worthy for providing secure software or providing rapid response to security issues? And, more importantly, how developers can learn from the others' mistakes and start writing secure code?
> Imagine running the jpegtopnm program in an "extreme > sandbox" that doesn't let the program do anything other > than read the JPEG le from standard input, write the > bitmap to standard output, and allocate a limited amount of > memory. Existing UNIX tools make this sandbox tolerably > easy for root to create: >... (working on RLIMITs, do chroot, fork, setuid, etc etc)
Hm... even if not allowing the program to reuse the filesystem and other programs in the system is tolerable, it doesn't sound like a good idea that a program that want to *lower* its privilege has to run as root. If nothing else, the system administrators will probably go mad staring at all those suid root programs (much more mad than all the non-standard procedures needed to run the current programs of djb indeed). Unluckily for all of us, POSIX-compliant Unix is still the way to go. I expect that an "extreme sandbox" would be much easier to build (and much more useful!) with a system like Hurd, only that it won't be ready for use outside the Hurd developers' spare PC in any foreseeable future.
Perhaps it's good that the courts "drop the ball"... imagine what Seagate can do now. Mark their drives as 200GB when everybody else mark it as 215GB? Makes no business sense--they will simply lose orders. Make special production lines for the US that mark it as 200GB while every other production line marks it as 200GB? Makes no operational sense--even their own staff will mix up the two. Easiest solution? Well... just stick a tiny "i" (perhaps at a smaller font) at the middle between G and B! At least they can tell their curious customers that "it is mandated by the court, you'll get just the same no worry". Other vendors will follow suit shortly for fear of litigation, there's nothing to lose anyway. People will soon find it not so awkward to utter the sound "gibibyte", and after that there is some reason for the OS to change their ways, too. And our decades long mess of units will finally be past issues. Or perhaps that's just my own wishful thinking.
The problem is probably the title of the article: "Know Any Hardware Needing Better Linux Support?" Nothing in the title (or even the overview!) talks anything about the "kernel". Most people do not RTFA, so most people will be pretty ignorant that the many people working on a driver are actually asking for kernel work rather than userland work. And no, to most people "Linux" means the whole thing based on the Linux kernel, not the kernel itself. If the submitter at least give a try, people will think "is that really a kernel work" before they complain that their video card, webcam or printer does not work.
> Yeah, it does. "View History". And the "nature" involved is mostly > the active core of editors; the "viewing public" that provides occasional > edits but doesn't track articles is more a source of mutations than the > selection mechanism.
This is an interesting interpretation. But my perception is that many articles no longer have such an "active core" of editors who had viewed all the historical edits and select the "best" one, while "viewing public" is abundant in supply who will make modifications to the single version currently active. In your interpretation, in many cases there is only mutation and no selection. Let's not to discuss whether this is happening, but instead think about what will happen if this does happen. My guess is that the article will still be of high quality, showing the social behavior at work.
On the other hand, if the social behavior is lost, and somehow people just go and make random (or plain bad) edits (and do that persistently), and suppose the "active core" is still there. What will happen? Do you still expect a high quality article? My guess is that you won't. The selection will be out-numbered, showing evolution behavior not at work (see what happens to OOXML discussion during the time of voting!).
And don't forget that Wikipedia itself says that "revert to a previous version" is such a powerful device that it shouldn't be used most of the time:
> (From Wikipedia:Etiquette) Avoid reverts whenever possible, and stay within > the three-revert rule except in cases of clear vandalism. Explain reversions > in the edit summary box.
In other words, Wikipedia itself is "suppressing" its "selection" mechanism!
For me, in Wikipedia human being runs both the "mutation" and "selection", so you require social behavior to make the "evolution" work anyway. Instead of saying "evolution" makes Wikipedia successful, I'd rather say that it is the social behavior at work.
> Comparing it to evolution, an edit of Wikipedia might be considered equivalent to a genetic mutation. A > mutation, of course, is non-directed...that is, "random." It could be bad or good, but most of the time > it is bad.
IMNSHO this is simply untrue. If this is true Wikipedia is dead for long: it never keeps a large, visible "pool" of "genes" (different version of the same article) that the "nature" (viewing public) can "select", and the "nature" simply is too busy to "select" them anyway. They have many version of the same article, but there are not many who will go into the version and select to revert to one of those. To me, the success of Wikipedia is that those who don't know much about a subject will normally refrain from editing the subject, so most edits are actually of a rather high quality. It is a social behavior, not an evolution behavior.
> The power consumption is proportional to the square of the voltage, not the frequency.
I suppose the "frequency" here means the frequency of the clock of the chip, rather than the frequency of the channel (the two should be quite strongly related, but not necessarily proportional).
The above is not exactly correct. To understand this, one needs to know why the chip consumes (most of the) power. It consumes power when its transistors switch. When they don't switch, they are at a high impedance state, so nearly no current flow through (and thus consumes nearly no power) although there is a voltage. When they do switch, a current flows for a brief period of time, and of course this leads to energy consumption following V square over R. But since this happens only for a brief moment we need to know the total length of period of time this happens in order to know the total energy consumption. And it isn't hard to see that this is proportional to the frequency f. So the energy consumption is actually proportional to f times squared V.
But this is not the whole story. In order to support a higher frequency, the chip requires a higher voltage (in order to force the states of the transistors to change quickly enough); and the dependency is approximately proportional. So it means that if you are a designer of a chip/product and can adjust both f and V, or if you use a chip which can adjust the voltage dynamically depending on the frequency, the power consumption varies approximately as cube of f (or equivalently cube of V).
So running fast is actually very expensive!
IANAEE. Instead I'm a computer scientist doing research on energy efficiency.
> I find it astounding that people would so willing store so much personal information on the > servers of these companies.... Is it because they're ignorant of the risks?
Can you suggest anywhere to handling our mail in such a way that no risk is involved? For the starter, it is okay if you propose a system without the following risks: that of "mails leaking to somebody who shouldn't be able to read my mails", of "having to pay undue amount of money to just to make sure others can E-mail me", of "getting so much spam that we have to spend undue amount of time just to process inbox", of "not being able to find a mail even though I know it must be somewhere", and of "not being able to access my mailbox within a minute".
I did look at the actual stats, and I see no trend that the growth "stopped" or is "declined". Instead I see it soars in August and return to a more normal growth in September. E.g., edits in the month is 7.40M (May), 7.68M (June), 7.90M (July), 8.80M (Aug) and 8.20M (Sep). September figure is growing quite normally if we neglect the August figure. I believe there is nothing bad going on, students just go on with their life after their summer vacations.
> It's odd that they are even smart enough to do something so nasty when they are so stupid (or brainwashed as > children and never learn critical thinking as happens to most poor "religious" folks) as to believe in this > creationist crapola.
No. They are as intelligent as any of us, if not more intelligent. At least they understand that intelligent beings do not only know how to learn what is true or not, but also how to lead others (at least, a significant portion of others) to believe what is true or not. And of course, they understand that what's important is not the truthfulness of what they are leading others to believe, but instead the worthiness of it. From the point of view of those creationists, it doesn't really matter whether evolution is truth or not. If it shaken the ground that others are joining their religious (the scripture, the bible), they have to find any possible way to lead others not to believe it, using their intelligence.
Actually I'm more interested in using it as a home server. Home server usually operate 24-hours, with screen off most of the time until you want to use it as a desktop, and mostly used for serving your own home page. In such systems, a 1W PC really makes a huge difference. The next question is, of course, (1) is there any power-efficient hard disk, and (2) is there any way to make sure the disk will not be spinned up more than once a few hours.
> Perl is an horrifically bad language. It's called "write-only" for a reason.
The reason is that there are more people create write-only script on Perl than other language. I've worked in a company which keep half their code in Perl, and I do find much of their code "write-only". But then, when I actually write my own code, and then read it after months, I can no longer justify calling Perl write only, at all. It's just the improperly trained people who make the whole thing that bad.
Why Perl had been so "write-only"? There can be many different reasons. Some of them are (1) they allow very terse code, (2) they do not promote OOP as hard as other languages (and has a strange way to do OOP), (3) they use so many different operators, some at strange precedence (at the first glance), (4) Perl promote using implicit stuff, (5) all the $@%*& hurts, etc.
But when I actually do it myself for some time, I learnt that (1) is actually helpful when you want to write readable code, (2) requires just that you do it your own way (and also is good in the sense that you don't write something which is really procedural to OO), (3) just means you will limit the use of operators in your own programs to those which really makes sense to you (and I use both && and "and", and both || and "or"), (4) just means you do it your own way except for really trivial ones, (5) really just need some time to get accustomed to, etc.
The more important thing is what a menu that the "Perl buffet" provides you with. You get all the convenient things: from running a shell command and catch the result using just a single line of backticks, to reading an XML config file by use a single line of XML::Simple call and doing memoization with just a single line of Memoize, to a built-in way to write lists and hashes so easily that you don't can create one without thinking (and to read one without thinking as well!), to the convenience of writing a simple for loop without any braces (using modifiers), to the power of a real closure facility, to the removal from the need to cater for managing memory in most cases, and to the whole CPAN experience using cpan2rpm. All those are big plus when you want to make your program both readable and maintainable!
At the beginning of that job I start out thinking that "we should move to Python but the pile of old code is in our way". After a few months it is clear to me that "any step trying to move away from Perl is a big wrong step because it just takes away the most important too for us to maintain readability and maintainability". And I've concluded that the only reason why our code had been so horrible is not because of the language. It's instead because of the environment and the improperly trained people.
About the only thing that I think Perl really need fixing is: there really need to be a usable Perl threading environment. Currently Perl threads are even more expensive than processes and thus is not usable at all. That's my only big complain about Perl, nothing more.
Different people of course have different tastes, and different programmers of course have different choice of language. But I think most people complaining about Perl do not taste good Perl, but instead they taste junk Perl and think that's the only taste of Perl. The problem is, there is junk Java, junk Python, junk C++, junk C# and junk Haskell out there. Comparing them makes no sense, they are not what we want. Instead we need to find some premium Perl, premium Python, premium C++, premium C# and premium Haskell to try before we can comment on the language. And if anyone do, and when tasting, taste deeply enough, I think most people will find premium Perl to taste just as good if not supreme.
> Generally speaking hashes are very cpu and cache-inefficient beasts
Um... why you think hashes are inefficient? In a lot of languages (Perl, Python, Javascript, etc) the standard collection is the hash. In Javascript, even a simple array is a hash! Why you think it is inefficient?
My thinking is that it is both CPU and cache efficient: it is CPU efficient because it usually just need one round of computation to get you to the correct result (as compared to a tree, which you need one round per tree level). It is cache efficient because you are usually not lead to somewhere irrelevant to your search (in contrast, any intermediate node in a tree when searching for an item in a binary tree will pollute your cache). Yes, in hash you have the hash table entries themselves which will pollute the cache, but that's not as much, exactly because of what you talk about: (spatial) locality of reference. In a hash all entries are in nearby memory, so it is likely that many searches in the same hash table will end up using very few cache lines. In contrast, in a search tree or a list, different nodes are allocated at different time and are much more likely to use completely different cache lines. At least this should be true until the time you overload it, but then you have extensible hashes.
It doesn't matter. White text on black is not as readable as black text on white, so people will make that up by turning up the brightness, which *does* matter.
If you RTFA, you should understand that the article actually consider fructose to be a carbohydrate that works like fat (or actually, beer) to the body. So the focus of "high quantities of carbohydrates can cause weight increase". Instead, it is "with normal carbohydrates your body would know to stop as soon as you have enough, but with the amount of fructose that an average American is taking, your body start to lose that very ability, and once that happen you are out of control and are highly prone to obesity", which is the problem they claim.
> I also have no idea where ~20-21 came from, as the maximum depth at four million is higher > than that.
That 20-21 number might be wrong for the "depth", but would be about right if he is talking about the "black depth" of the red-black tree (a red-black tree of depth 22 contains 1+2+...+2^21 = 2^22-1 = 4M-1 nodes), which is at least half of the actual depth. So the actual depth is no more than 44 (actually a bit less).
The title of the story seems to make a lot of sense. Computer science should be the science about how to make computer does the work that people want, and it should not be just a mathematical science; and given the wealth of knowledges that mathematics is already offering, and the lack of trusted knowledges from other parts of equation like programming process and creative thinking, I think it should worth the click. It doesn't. Just see the following quote straight from TFA:
A logic circuit is not a sequence of operations. An operating system is not supposed to terminate, nor does it yield a singular solution. An operating system cannot be deterministic because it must relate to uncoordinated inputs from the outside world. Any program utilising random input to carry out its process, such...is not an algorithm.
So the article isn't about all those. Instead, it is because that stupid guy didn't know that mathematics do not end at computing a function. He thinks there is no mathematics in programs just because they deviate slightly from the simplest model of computation that put some input to the RAM tape and have an algorithm put something into the output tape. He ignores that mathematics is rich enough to capture computations that lasts more than providing a single value. He ignores the wealth of knowledge of mathematics that deals with coordinating other computation. He ignores the wealth of knowledge that deals with random process and, indeed, the source of randomness, which is algorithmically created. And then he claim that mathematics has no place in it, ignoring that once you have any sort of idea, it requires mathematical arguments to make sure nothing wrong can happen to it. I can't help but believing that this is one of the most clueless article I've ever seen in/.
> This is really a question of statistics not of mathematics. Having done experiments on MBA > students, we found that a well written multiple choice question is more accurate than 4 well > written essays. The fact that we can easilly have 50 multiple choice questions and a maximum > of 8 essays makes it a no brainer that multiple choice is much more accurate.
I don't know how you judge whether a question is well written or not. In my experience, multiple choice questions are very easy to write wrongly. A wrongly worded essay question easily have exactly the opposite effect as you want: you reward the ones who know the subject less (it seldom just give you random noise). Worse, you won't know it happened before you're told. I've read many exam MC questions during exam paper review meetings, my feeling from reading such questions for 4 years is that one in 4-6 MC questions are poor enough this way. In contrast, a wrongly worded essay question will present students some real-life trouble (the questions that they will face will be full of inaccuracies!), and when marking them you know the question is written wrongly, but at the same time you know whether the students are good anyway.
But the real problem of multiple choice questions is that it doesn't present the student any real world test. In the real world, nobody would tell you that "You are in situation, you can do A, B, C or D. Please choose one". Instead, what they see is "Somebody is in this situation. Please advice." Being good in multiple choice question usually has doubtful utility in the real world. And education systems will have to align with the judgement system, so at the end the teachers train their students the wrong technique as well.
Of course, there are benefits of MC questions: they can be marked mechanically, which means that (1) they lessen the workload of markers, (2) they are marked with perfect consistency, and (3) their markings are free from language or hand-writing proficiency. I don't think "accuracy" is one of those, though, since MC questions are just testing the wrong ability.
Wrong. No matter which branch of science you're talking about, once you have the experiments done and have the papers out, you want a consensus. It is because there are many ways to interpret your experiment result, and there are people who will put doubt into the integrity of your experiment. It requires a consensus from the scientific community, whether you are talking about evolution, relativity, global warming or string theory. Scientific findings only becomes accepted theories when most scientists in the field understood the findings, trusted that if they are given the funding and repeat the experiment they will get the same result, and tried and found that any alternative explanation they can think of for the experiment will lead to inconsistency with previous experiments they trust or highly improbable events happening. Consensus are part of science. It's only that the consensus are taken from the ones who actually know how to interpret the experiment and perhaps to repeat it, rather than from the general public.
> and that's exactly what the Global Warming debate is about: politics
True. Global warming is about politics. Not because it is not known whether global warming is a fact or not. It is, and in scientific community it is no longer debated about whether it is happening, or that human is a primary reason why that happen, or even whether this is bad or not. Instead, global warming is politics because we need to do something about it, and it takes people to do something about it. After a problem arise, it will not disappear because "somebody studied that and identified the reason". Somebody must find a way to solve it, and the people have to use the method. Unluckily, the global warming problem is not one which the first one using a solution will be the one to gain most. In the contrary, it is a problem which the first one using a solution will be the one to give most: they will be the one to spend money and restrict people. So the Nash equilibrium is a state in which everybody suffers. Now it becomes a political problem, to change the landscape so that the first one using a solution will at least not be giving the most, perhaps just give a bit in terms of money and gain a bit in terms of political power, or long term desirability, or whatever. The sad fact is that some largest country of the world, with the largest and most powerful budget, tries to be somebody who will gain in the short term by delaying the implementation of a solution, gambling that the long term cost to do that is less, and ignoring the fact that it means every party will suffer by this gamble.
> But lets see, e.g. Wikipedia example of RAII. The file objects cannot be put into STL > containers because they are not copyable. Making them copyable... just to define how > should e.g. write behave is a big task. Not worth it.
Not just "not worth it", but instead "absolute nonsense". If file objects cannot be copied, it is for its good reason. If you need to put something into a container that is anything related with a file, just store a shared_ptr of it.
Slashdot Mirror
Actually I always wonder why there are so many data centers placed at hot places... after all most of the time we don't care whether our data is sent half a globe in the East-West direction or North-South direction. unlucky for those living at those hot regions, the extra cost of air-conditioning to make it work also means that there are a lot more carbon emission and power consumption than really needed, both works to make their place even hotter.
> From a legal point of view material gained with a secret computer search shouldn't be brought to court,
> because there is no way to prove that the evidence isn't faked.
There are lots of activities done by the police which is exactly like that: they know it is impossible to put the direct result to court as evidence for whatever reason ("the court know the police can fake it" is about the most minor one...), but the police would do it anyway. Reason? If they can tap into their suspect and as a result know something would happen in "real life" rather than electronically, they can organize an operation, and the result of that operation can be put to court. No court will be dissatisfied if the operation is done in response to some "undisclosed insiders' information".
Like you, I use NoScript and AdBlock Plus and won't see any of those sh*t. But I'm actually rather confused about why the hell did they do it. Are they really that stupid? Once everybody are forced to learn to do the same there is no business for ads business. Did somebody predict something in the lines that Microsoft will never make piracy checking bullet-proof because they would kill themselves that way by forcing everybody to Linux? Perhaps those Ads companies are really not that clever after all.
This sounds *really* wrong. You can say white-hats should have waited for a few days or even a few weeks after notifying the vendors before disclosing problems, but they should be disclosed eventually, and should be disclosed after giving vendors a reasonable amount of time. There bound to be people not upgrading their Windows, and there bound to be people not upgrading their Redhat or Fedora or Ubuntu or SuSE or FreeBSD or whatever operating system you name (not to mention whatever Firewalls, protocols, applications, etc, etc you name). So we shouldn't be disclosing any vulnerability about any of those?! Who, then, know that their software is vulnerable to black-hats and needs upgrading, and who, then, know which software vendor is more trust-worthy for providing secure software or providing rapid response to security issues? And, more importantly, how developers can learn from the others' mistakes and start writing secure code?
SETI did successfully find something already!
> Imagine running the jpegtopnm program in an "extreme ... (working on RLIMITs, do chroot, fork, setuid, etc etc)
> sandbox" that doesn't let the program do anything other
> than read the JPEG le from standard input, write the
> bitmap to standard output, and allocate a limited amount of
> memory. Existing UNIX tools make this sandbox tolerably
> easy for root to create:
>
Hm... even if not allowing the program to reuse the filesystem and other programs in the system is tolerable, it doesn't sound like a good idea that a program that want to *lower* its privilege has to run as root. If nothing else, the system administrators will probably go mad staring at all those suid root programs (much more mad than all the non-standard procedures needed to run the current programs of djb indeed). Unluckily for all of us, POSIX-compliant Unix is still the way to go. I expect that an "extreme sandbox" would be much easier to build (and much more useful!) with a system like Hurd, only that it won't be ready for use outside the Hurd developers' spare PC in any foreseeable future.
Perhaps it's good that the courts "drop the ball"... imagine what Seagate can do now. Mark their drives as 200GB when everybody else mark it as 215GB? Makes no business sense--they will simply lose orders. Make special production lines for the US that mark it as 200GB while every other production line marks it as 200GB? Makes no operational sense--even their own staff will mix up the two. Easiest solution? Well... just stick a tiny "i" (perhaps at a smaller font) at the middle between G and B! At least they can tell their curious customers that "it is mandated by the court, you'll get just the same no worry". Other vendors will follow suit shortly for fear of litigation, there's nothing to lose anyway. People will soon find it not so awkward to utter the sound "gibibyte", and after that there is some reason for the OS to change their ways, too. And our decades long mess of units will finally be past issues. Or perhaps that's just my own wishful thinking.
The problem is probably the title of the article: "Know Any Hardware Needing Better Linux Support?" Nothing in the title (or even the overview!) talks anything about the "kernel". Most people do not RTFA, so most people will be pretty ignorant that the many people working on a driver are actually asking for kernel work rather than userland work. And no, to most people "Linux" means the whole thing based on the Linux kernel, not the kernel itself. If the submitter at least give a try, people will think "is that really a kernel work" before they complain that their video card, webcam or printer does not work.
> Yeah, it does. "View History". And the "nature" involved is mostly
> the active core of editors; the "viewing public" that provides occasional
> edits but doesn't track articles is more a source of mutations than the
> selection mechanism.
This is an interesting interpretation. But my perception is that many articles no longer have such an "active core" of editors who had viewed all the historical edits and select the "best" one, while "viewing public" is abundant in supply who will make modifications to the single version currently active. In your interpretation, in many cases there is only mutation and no selection. Let's not to discuss whether this is happening, but instead think about what will happen if this does happen. My guess is that the article will still be of high quality, showing the social behavior at work.
On the other hand, if the social behavior is lost, and somehow people just go and make random (or plain bad) edits (and do that persistently), and suppose the "active core" is still there. What will happen? Do you still expect a high quality article? My guess is that you won't. The selection will be out-numbered, showing evolution behavior not at work (see what happens to OOXML discussion during the time of voting!).
And don't forget that Wikipedia itself says that "revert to a previous version" is such a powerful device that it shouldn't be used most of the time:
> (From Wikipedia:Etiquette) Avoid reverts whenever possible, and stay within
> the three-revert rule except in cases of clear vandalism. Explain reversions
> in the edit summary box.
In other words, Wikipedia itself is "suppressing" its "selection" mechanism!
For me, in Wikipedia human being runs both the "mutation" and "selection", so you require social behavior to make the "evolution" work anyway. Instead of saying "evolution" makes Wikipedia successful, I'd rather say that it is the social behavior at work.
> Comparing it to evolution, an edit of Wikipedia might be considered equivalent to a genetic mutation. A
> mutation, of course, is non-directed...that is, "random." It could be bad or good, but most of the time
> it is bad.
IMNSHO this is simply untrue. If this is true Wikipedia is dead for long: it never keeps a large, visible "pool" of "genes" (different version of the same article) that the "nature" (viewing public) can "select", and the "nature" simply is too busy to "select" them anyway. They have many version of the same article, but there are not many who will go into the version and select to revert to one of those. To me, the success of Wikipedia is that those who don't know much about a subject will normally refrain from editing the subject, so most edits are actually of a rather high quality. It is a social behavior, not an evolution behavior.
> The power consumption is proportional to the square of the voltage, not the frequency.
I suppose the "frequency" here means the frequency of the clock of the chip, rather than the frequency of the channel (the two should be quite strongly related, but not necessarily proportional).
The above is not exactly correct. To understand this, one needs to know why the chip consumes (most of the) power. It consumes power when its transistors switch. When they don't switch, they are at a high impedance state, so nearly no current flow through (and thus consumes nearly no power) although there is a voltage. When they do switch, a current flows for a brief period of time, and of course this leads to energy consumption following V square over R. But since this happens only for a brief moment we need to know the total length of period of time this happens in order to know the total energy consumption. And it isn't hard to see that this is proportional to the frequency f. So the energy consumption is actually proportional to f times squared V.
But this is not the whole story. In order to support a higher frequency, the chip requires a higher voltage (in order to force the states of the transistors to change quickly enough); and the dependency is approximately proportional. So it means that if you are a designer of a chip/product and can adjust both f and V, or if you use a chip which can adjust the voltage dynamically depending on the frequency, the power consumption varies approximately as cube of f (or equivalently cube of V).
So running fast is actually very expensive!
IANAEE. Instead I'm a computer scientist doing research on energy efficiency.
> I find it astounding that people would so willing store so much personal information on the ... Is it because they're ignorant of the risks?
> servers of these companies.
Can you suggest anywhere to handling our mail in such a way that no risk is involved? For the starter, it is okay if you propose a system without the following risks: that of "mails leaking to somebody who shouldn't be able to read my mails", of "having to pay undue amount of money to just to make sure others can E-mail me", of "getting so much spam that we have to spend undue amount of time just to process inbox", of "not being able to find a mail even though I know it must be somewhere", and of "not being able to access my mailbox within a minute".
I did look at the actual stats, and I see no trend that the growth "stopped" or is "declined". Instead I see it soars in August and return to a more normal growth in September. E.g., edits in the month is 7.40M (May), 7.68M (June), 7.90M (July), 8.80M (Aug) and 8.20M (Sep). September figure is growing quite normally if we neglect the August figure. I believe there is nothing bad going on, students just go on with their life after their summer vacations.
> It's odd that they are even smart enough to do something so nasty when they are so stupid (or brainwashed as
> children and never learn critical thinking as happens to most poor "religious" folks) as to believe in this
> creationist crapola.
No. They are as intelligent as any of us, if not more intelligent. At least they understand that intelligent beings do not only know how to learn what is true or not, but also how to lead others (at least, a significant portion of others) to believe what is true or not. And of course, they understand that what's important is not the truthfulness of what they are leading others to believe, but instead the worthiness of it. From the point of view of those creationists, it doesn't really matter whether evolution is truth or not. If it shaken the ground that others are joining their religious (the scripture, the bible), they have to find any possible way to lead others not to believe it, using their intelligence.
Actually I'm more interested in using it as a home server. Home server usually operate 24-hours, with screen off most of the time until you want to use it as a desktop, and mostly used for serving your own home page. In such systems, a 1W PC really makes a huge difference. The next question is, of course, (1) is there any power-efficient hard disk, and (2) is there any way to make sure the disk will not be spinned up more than once a few hours.
> Perl is an horrifically bad language. It's called "write-only" for a reason.
The reason is that there are more people create write-only script on Perl than other language. I've worked in a company which keep half their code in Perl, and I do find much of their code "write-only". But then, when I actually write my own code, and then read it after months, I can no longer justify calling Perl write only, at all. It's just the improperly trained people who make the whole thing that bad.
Why Perl had been so "write-only"? There can be many different reasons. Some of them are (1) they allow very terse code, (2) they do not promote OOP as hard as other languages (and has a strange way to do OOP), (3) they use so many different operators, some at strange precedence (at the first glance), (4) Perl promote using implicit stuff, (5) all the $@%*& hurts, etc.
But when I actually do it myself for some time, I learnt that (1) is actually helpful when you want to write readable code, (2) requires just that you do it your own way (and also is good in the sense that you don't write something which is really procedural to OO), (3) just means you will limit the use of operators in your own programs to those which really makes sense to you (and I use both && and "and", and both || and "or"), (4) just means you do it your own way except for really trivial ones, (5) really just need some time to get accustomed to, etc.
The more important thing is what a menu that the "Perl buffet" provides you with. You get all the convenient things: from running a shell command and catch the result using just a single line of backticks, to reading an XML config file by use a single line of XML::Simple call and doing memoization with just a single line of Memoize, to a built-in way to write lists and hashes so easily that you don't can create one without thinking (and to read one without thinking as well!), to the convenience of writing a simple for loop without any braces (using modifiers), to the power of a real closure facility, to the removal from the need to cater for managing memory in most cases, and to the whole CPAN experience using cpan2rpm. All those are big plus when you want to make your program both readable and maintainable!
At the beginning of that job I start out thinking that "we should move to Python but the pile of old code is in our way". After a few months it is clear to me that "any step trying to move away from Perl is a big wrong step because it just takes away the most important too for us to maintain readability and maintainability". And I've concluded that the only reason why our code had been so horrible is not because of the language. It's instead because of the environment and the improperly trained people.
About the only thing that I think Perl really need fixing is: there really need to be a usable Perl threading environment. Currently Perl threads are even more expensive than processes and thus is not usable at all. That's my only big complain about Perl, nothing more.
Different people of course have different tastes, and different programmers of course have different choice of language. But I think most people complaining about Perl do not taste good Perl, but instead they taste junk Perl and think that's the only taste of Perl. The problem is, there is junk Java, junk Python, junk C++, junk C# and junk Haskell out there. Comparing them makes no sense, they are not what we want. Instead we need to find some premium Perl, premium Python, premium C++, premium C# and premium Haskell to try before we can comment on the language. And if anyone do, and when tasting, taste deeply enough, I think most people will find premium Perl to taste just as good if not supreme.
> Generally speaking hashes are very cpu and cache-inefficient beasts
Um... why you think hashes are inefficient? In a lot of languages (Perl, Python, Javascript, etc) the standard collection is the hash. In Javascript, even a simple array is a hash! Why you think it is inefficient?
My thinking is that it is both CPU and cache efficient: it is CPU efficient because it usually just need one round of computation to get you to the correct result (as compared to a tree, which you need one round per tree level). It is cache efficient because you are usually not lead to somewhere irrelevant to your search (in contrast, any intermediate node in a tree when searching for an item in a binary tree will pollute your cache). Yes, in hash you have the hash table entries themselves which will pollute the cache, but that's not as much, exactly because of what you talk about: (spatial) locality of reference. In a hash all entries are in nearby memory, so it is likely that many searches in the same hash table will end up using very few cache lines. In contrast, in a search tree or a list, different nodes are allocated at different time and are much more likely to use completely different cache lines. At least this should be true until the time you overload it, but then you have extensible hashes.
> AFAIK, dark is the active state
It doesn't matter. White text on black is not as readable as black text on white, so people will make that up by turning up the brightness, which *does* matter.
If you RTFA, you should understand that the article actually consider fructose to be a carbohydrate that works like fat (or actually, beer) to the body. So the focus of "high quantities of carbohydrates can cause weight increase". Instead, it is "with normal carbohydrates your body would know to stop as soon as you have enough, but with the amount of fructose that an average American is taking, your body start to lose that very ability, and once that happen you are out of control and are highly prone to obesity", which is the problem they claim.
> I also have no idea where ~20-21 came from, as the maximum depth at four million is higher
> than that.
That 20-21 number might be wrong for the "depth", but would be about right if he is talking about the "black depth" of the red-black tree (a red-black tree of depth 22 contains 1+2+...+2^21 = 2^22-1 = 4M-1 nodes), which is at least half of the actual depth. So the actual depth is no more than 44 (actually a bit less).
"Universal preference" means that all men have such a preference.
I don't think in Psychology or any other social science, "universal" means "all". They nearly always means "a significant majority".
So the article isn't about all those. Instead, it is because that stupid guy didn't know that mathematics do not end at computing a function. He thinks there is no mathematics in programs just because they deviate slightly from the simplest model of computation that put some input to the RAM tape and have an algorithm put something into the output tape. He ignores that mathematics is rich enough to capture computations that lasts more than providing a single value. He ignores the wealth of knowledge of mathematics that deals with coordinating other computation. He ignores the wealth of knowledge that deals with random process and, indeed, the source of randomness, which is algorithmically created. And then he claim that mathematics has no place in it, ignoring that once you have any sort of idea, it requires mathematical arguments to make sure nothing wrong can happen to it. I can't help but believing that this is one of the most clueless article I've ever seen in
> This is really a question of statistics not of mathematics. Having done experiments on MBA
> students, we found that a well written multiple choice question is more accurate than 4 well
> written essays. The fact that we can easilly have 50 multiple choice questions and a maximum
> of 8 essays makes it a no brainer that multiple choice is much more accurate.
I don't know how you judge whether a question is well written or not. In my experience, multiple choice questions are very easy to write wrongly. A wrongly worded essay question easily have exactly the opposite effect as you want: you reward the ones who know the subject less (it seldom just give you random noise). Worse, you won't know it happened before you're told. I've read many exam MC questions during exam paper review meetings, my feeling from reading such questions for 4 years is that one in 4-6 MC questions are poor enough this way. In contrast, a wrongly worded essay question will present students some real-life trouble (the questions that they will face will be full of inaccuracies!), and when marking them you know the question is written wrongly, but at the same time you know whether the students are good anyway.
But the real problem of multiple choice questions is that it doesn't present the student any real world test. In the real world, nobody would tell you that "You are in situation, you can do A, B, C or D. Please choose one". Instead, what they see is "Somebody is in this situation. Please advice." Being good in multiple choice question usually has doubtful utility in the real world. And education systems will have to align with the judgement system, so at the end the teachers train their students the wrong technique as well.
Of course, there are benefits of MC questions: they can be marked mechanically, which means that (1) they lessen the workload of markers, (2) they are marked with perfect consistency, and (3) their markings are free from language or hand-writing proficiency. I don't think "accuracy" is one of those, though, since MC questions are just testing the wrong ability.
> Consensus science isn't science, it's politics,
Wrong. No matter which branch of science you're talking about, once you have the experiments done and have the papers out, you want a consensus. It is because there are many ways to interpret your experiment result, and there are people who will put doubt into the integrity of your experiment. It requires a consensus from the scientific community, whether you are talking about evolution, relativity, global warming or string theory. Scientific findings only becomes accepted theories when most scientists in the field understood the findings, trusted that if they are given the funding and repeat the experiment they will get the same result, and tried and found that any alternative explanation they can think of for the experiment will lead to inconsistency with previous experiments they trust or highly improbable events happening. Consensus are part of science. It's only that the consensus are taken from the ones who actually know how to interpret the experiment and perhaps to repeat it, rather than from the general public.
> and that's exactly what the Global Warming debate is about: politics
True. Global warming is about politics. Not because it is not known whether global warming is a fact or not. It is, and in scientific community it is no longer debated about whether it is happening, or that human is a primary reason why that happen, or even whether this is bad or not. Instead, global warming is politics because we need to do something about it, and it takes people to do something about it. After a problem arise, it will not disappear because "somebody studied that and identified the reason". Somebody must find a way to solve it, and the people have to use the method. Unluckily, the global warming problem is not one which the first one using a solution will be the one to gain most. In the contrary, it is a problem which the first one using a solution will be the one to give most: they will be the one to spend money and restrict people. So the Nash equilibrium is a state in which everybody suffers. Now it becomes a political problem, to change the landscape so that the first one using a solution will at least not be giving the most, perhaps just give a bit in terms of money and gain a bit in terms of political power, or long term desirability, or whatever. The sad fact is that some largest country of the world, with the largest and most powerful budget, tries to be somebody who will gain in the short term by delaying the implementation of a solution, gambling that the long term cost to do that is less, and ignoring the fact that it means every party will suffer by this gamble.
> But lets see, e.g. Wikipedia example of RAII. The file objects cannot be put into STL ... just to define how
> containers because they are not copyable. Making them copyable
> should e.g. write behave is a big task. Not worth it.
Not just "not worth it", but instead "absolute nonsense". If file objects cannot be copied, it is for its good reason. If you need to put something into a container that is anything related with a file, just store a shared_ptr of it.