In theory you're right, but in practice cars don't seem to improve much in efficiency over a car's lifetime. For instance, a Honda Civic in 1987 got 57 miles per gallon.
Vir says that since many of the links that make it to the front page of the site have already been shared on other social networks, the Digg In The Future software looks at frequently shared URLs from Twitter and gives those added weight.
I challenge Mark to point to a single instance where Intercontinental Ballistic Missiles with Nuclear Warheads were successfully used for nefarious purposes.
Some people have the inkling that there is something wrong with your analogy, but no one has analyzed it in monotonous detail yet, so I'll give it a shot.
When quantifying the danger in something, it's important to consider both the probability of harm and the magnitude of that harm. Something may be a significant risk because it causes minor harm frequently (e.g. a neighbor playing their music too loud) or because it could cause severe harm (e.g. nuclear missile).
In the case we are talking about, someone stealing someone's RFID card is an instance of a relatively minor harm. Thus if it happens relatively infrequently, it's relatively safe. The "fact" that it hasn't happened so far is evidence that it happens relatively infrequently.
On the other hand, a nuclear missile can cause extreme harm. Thus, they are dangerous even if they cause harm only infrequently. The fact that millions of people haven't died due to an ICBM may be evidence that the missiles don't kill people frequently. However, this fact is very poor evidence against the idea that ICBMs don't pose significant danger.
Agree on the GPU part. If you haven't seen them, check out the gnutools and cudaBayesreg R packages. They don't look too easy to use now, but eventually this will become mainstream.
Yep, monte carlo techniques typically converge at O(n^.5), unless possibly if you are low discrepancy sequences/quasi-random numbers. But I guess I don't see how the article will result in hardware that can manipulate distributions in "a few CPU cycles". First of all, manipulating a probability distribution often does not involve many numbers between 0 and 1 (e.g. if it is continuous you are dealing with probability densities, not probabilities).
About convolutions, those are usually most quickly calculated using fast fourier transforms. So anyway, I don't know if some awesome floating point analog computer is possible, I'm just saying that if they made some chip that excelled at Bayesian computation it would also revolutionize many other fields. For instance, how many different disciplines use FFT? MCMC itself was first developed to handle problems in nuclear physics I think.
The article mentions Bayesian calculations. Can these computers really speed up those calculations? Nowadays Bayesian calculations usually involve thousands of iterations of a technique called Markov Chain Monte Carlo (MCMC) unless the distributions in question are conjugate priors. The simulation then converges to the right answer.
The issue I see is that all these techniques are just math. They are either analytic (conjugate priors) or require strict error bounds in order get sensible answers (MCMC). There's no separate system of math that Bayesians use. Like many others, Bayesians just need quick reliable floating point mathematics. So anyway, I don't see how this can help Bayesian statisticians, unless it also revolutionizes engineering, physics, etc.
Yep, there's also Google desktop, and Chrome. (And if you count dropped products, Wave. Although that was probably "too innovative" to be successful.)
Also, it's not like Google search was finished 5-10 years ago. Improving search (or even keeping it from deteriorating) requires tons of innovations (e.g. MapReduce). Doing anything at Google's scale makes innovation a necessity behind the scenes. They've also added new search features like Google scholar which I use a lot.
So now the projects will actually have to have some merit? Sounds good to me.
Ideally you're right, but often which projects have merit in the long term is a self-fulfilling prophesy. A decision to switch to a platform is a long-term commitment, and necessarily is about what will have "merit" 5 years from now.
Google is up against Apple and its Reality Distortion Field and Microsoft with its 90 thousand employees and its Windows/Office cash cow monopoly. Google must convince people that its platforms will have critical mass five years from now—if they don't their projects won't have (long term) merit, no matter how good they are.
Even if Wave was a bad idea, perhaps Google should have continued to support it.
Why throw resources into a bottomless pit? Because time and time again it's not the best technology that wins, it's the one that everyone thinks everyone else is using. (Examples (debatable of course): qwerty keyboards, VHS, SQL, windows, C++, XML, javascript)
In the future, Google will unveil other major initiatives and will try to reach critical mass with them. Now that people know Google is willing to abandon a large project so easily, they will be less likely to commit to future Google projects.
Well here's a simple example: keeping an EFT (exchange traded fund) in line with its associated basket of stocks. Because the stocks that comprise it change on a millisecond basis, keeping the EFT priced correctly also requires millisecond trades.
Good point, but that power is pretty hard to use. Most managers are more than happy to go with the "devil they know", probably because it would take a year+ to get the new person up to speed. If you replace a lot of (important) people at once, your whole company is put in chaos and enters a doomsday spiral (not to mention the morale problems). So unless your company is facing imminent demise, you're pretty much stuck with the people you have.
My guess is that it was similar in the age of monarchies. King Henry VIII may have been able to "single-handedly grant and relieve citizenship" as you put it, but he still could never change the culture of England to match France (I think William the Conquerer tried that).
I firmly belive that I am among the very few people who have played the entire half-life series, and yet have spent less than 10 minutes playing any source or goldsource multiplayer game.
Nah, I bet your preferences are really common. I've also played all of half-life but never touched multiplayer. My guess is we're the silent majority that you never hear about because game reviewers necessarily like all or most game types, including multiplayer.
Your post got modded very highly, but there's no reason why things have to work that way. Does the U.S. have the culture it does because of Obama? Large groups of people who interactive regularly, including many companies, eventually acquire a culture which cannot be traced directly back to a single individual. It's just an emergent property of large groups of people.
Also, the GP's point wasn't that Microsoft is a bad company. S/he said that MS's culture is compatible with staid business applications, not creativity and style. A good CEO knows the corporate culture at a company and chooses goals accordingly.
Hah, you can be witty and right at the same time. Do you really think that firing guns is no more interesting than using a socket wrench? I know plenty of people (including myself) who have fired a gun in real life and still enjoy first person shooters. Plenty of people also go shooting for fun. No one uses a socket wrench for fun.
I don't have any strong feelings pro- or con- about gun control, but your whole guns = socket wrench position is ridiculous and must be motivated by politics.
Once again, TI is preventing calculator hobbyists from running their own software on calculators they bought and paid for.
So what? As everyone else said, there's no point in buying a TI to do real math and graphing—just buy a laptop and put Sage/Octave/R/whatever on it. The point of a TI is that it's a portable device with presumably circumscribed functionality well known to teachers and proctors. Whether or not I think this is the way exams should be conducted, why shouldn't TI feel free to pursue this market?
Clearly it would be bad if every device were like a TI calculator or iPhone, but as long as they are niche players, I see no problem in their actions. If anything it's the Ndless community that's being counterproductive here.
Wow, Chile must have a great educational system if every non-native English speaker writes English "properly". And you're very clever to catch him in such an obvious lie: he doesn't write English well, clearly he can't be Chilean!
Funny, but the fact that it's funny is also sad (sad that it's so ridiculous to think he'd be interested in science). Glad that what was parody is now sorta true.
Re:So you are taking Economist seriously.
on
Behind Cyberwar FUD
·
· Score: 4, Informative
one can easily say that this time the group they are licking the boots of is RIAA.
The Economist is the world's best weekly newspaper. If you read what they say about the RIAA, including the first article which mentions how the RIAA's agressive tactics aren't working and are a lesson to other industries on what _not_ to do, you'd know that the Economist takes a moderate view on intellectual property.
In particular, they often report on academic research showing that IP laws are too strong. For instance, this article (subscription required) called "Killing Creativity" is about how overly strong IP laws can smother innovation.
Disclaimer: Years ago before getting papers on the internet was pervasive, I was once a TA. One of my students was struggling with coming up with a good paper topic, but in the end she turned in a great paper. When I complimented her on it, she seemed oddly evasive. A few years later I realized she had probably cheated.
While I'm on my soapbox, I'd like to say I think it's pretty sad that this kind of service is useful. If education was done properly, or specifically if students were evaluated in a meaningful and practical way, this service would be useful to maybe a handful of smart but lazy students.
I think plenty of good classes can be hurt by this service. For instance, suppose a student gets assigned to write an essay on <important work>. This essay might be open-ended (to allow the student to be creative) and a take-home assignment (for smart students who aren't necessarily the best under pressure or who can't physically write fast). I don't see why this is education done improperly or how this kind of assignment isn't meaningful. If it's impractical it may be impractical just because of services like this.
That, and the final reason would be if I DO charge something, and forget to pay the bill that month, it can be the difference between say, a buck in interest, or five bucks.
Uhh, not to say that you didn't have a very legitimate reason for switching banks, but if your interest rate moved from 9.9% to 17%, that's a 72% increase. So the difference in interest in not paying for a month would be more like a buck vs $1.72.
Slashdot Mirror
In theory you're right, but in practice cars don't seem to improve much in efficiency over a car's lifetime. For instance, a Honda Civic in 1987 got 57 miles per gallon.
Some people have the inkling that there is something wrong with your analogy, but no one has analyzed it in monotonous detail yet, so I'll give it a shot.
When quantifying the danger in something, it's important to consider both the probability of harm and the magnitude of that harm. Something may be a significant risk because it causes minor harm frequently (e.g. a neighbor playing their music too loud) or because it could cause severe harm (e.g. nuclear missile).
In the case we are talking about, someone stealing someone's RFID card is an instance of a relatively minor harm. Thus if it happens relatively infrequently, it's relatively safe. The "fact" that it hasn't happened so far is evidence that it happens relatively infrequently.
On the other hand, a nuclear missile can cause extreme harm. Thus, they are dangerous even if they cause harm only infrequently. The fact that millions of people haven't died due to an ICBM may be evidence that the missiles don't kill people frequently. However, this fact is very poor evidence against the idea that ICBMs don't pose significant danger.
Agree on the GPU part. If you haven't seen them, check out the gnutools and cudaBayesreg R packages. They don't look too easy to use now, but eventually this will become mainstream.
Yep, monte carlo techniques typically converge at O(n^.5), unless possibly if you are low discrepancy sequences/quasi-random numbers. But I guess I don't see how the article will result in hardware that can manipulate distributions in "a few CPU cycles". First of all, manipulating a probability distribution often does not involve many numbers between 0 and 1 (e.g. if it is continuous you are dealing with probability densities, not probabilities).
About convolutions, those are usually most quickly calculated using fast fourier transforms. So anyway, I don't know if some awesome floating point analog computer is possible, I'm just saying that if they made some chip that excelled at Bayesian computation it would also revolutionize many other fields. For instance, how many different disciplines use FFT? MCMC itself was first developed to handle problems in nuclear physics I think.
The article mentions Bayesian calculations. Can these computers really speed up those calculations? Nowadays Bayesian calculations usually involve thousands of iterations of a technique called Markov Chain Monte Carlo (MCMC) unless the distributions in question are conjugate priors. The simulation then converges to the right answer.
The issue I see is that all these techniques are just math. They are either analytic (conjugate priors) or require strict error bounds in order get sensible answers (MCMC). There's no separate system of math that Bayesians use. Like many others, Bayesians just need quick reliable floating point mathematics. So anyway, I don't see how this can help Bayesian statisticians, unless it also revolutionizes engineering, physics, etc.
Yep, the U.S. was really free in during the Cold War war with its the Jim Crow laws and McCarthyism. Truly a great time to be a white male.
Yep, there's also Google desktop, and Chrome. (And if you count dropped products, Wave. Although that was probably "too innovative" to be successful.)
Also, it's not like Google search was finished 5-10 years ago. Improving search (or even keeping it from deteriorating) requires tons of innovations (e.g. MapReduce). Doing anything at Google's scale makes innovation a necessity behind the scenes. They've also added new search features like Google scholar which I use a lot.
Ideally you're right, but often which projects have merit in the long term is a self-fulfilling prophesy. A decision to switch to a platform is a long-term commitment, and necessarily is about what will have "merit" 5 years from now.
Google is up against Apple and its Reality Distortion Field and Microsoft with its 90 thousand employees and its Windows/Office cash cow monopoly. Google must convince people that its platforms will have critical mass five years from now—if they don't their projects won't have (long term) merit, no matter how good they are.
Even if Wave was a bad idea, perhaps Google should have continued to support it.
Why throw resources into a bottomless pit? Because time and time again it's not the best technology that wins, it's the one that everyone thinks everyone else is using. (Examples (debatable of course): qwerty keyboards, VHS, SQL, windows, C++, XML, javascript)
In the future, Google will unveil other major initiatives and will try to reach critical mass with them. Now that people know Google is willing to abandon a large project so easily, they will be less likely to commit to future Google projects.
In your example, 150 is known as the Dunbar number.
Well here's a simple example: keeping an EFT (exchange traded fund) in line with its associated basket of stocks. Because the stocks that comprise it change on a millisecond basis, keeping the EFT priced correctly also requires millisecond trades.
Good point, but that power is pretty hard to use. Most managers are more than happy to go with the "devil they know", probably because it would take a year+ to get the new person up to speed. If you replace a lot of (important) people at once, your whole company is put in chaos and enters a doomsday spiral (not to mention the morale problems). So unless your company is facing imminent demise, you're pretty much stuck with the people you have.
My guess is that it was similar in the age of monarchies. King Henry VIII may have been able to "single-handedly grant and relieve citizenship" as you put it, but he still could never change the culture of England to match France (I think William the Conquerer tried that).
Nah, I bet your preferences are really common. I've also played all of half-life but never touched multiplayer. My guess is we're the silent majority that you never hear about because game reviewers necessarily like all or most game types, including multiplayer.
Your post got modded very highly, but there's no reason why things have to work that way. Does the U.S. have the culture it does because of Obama? Large groups of people who interactive regularly, including many companies, eventually acquire a culture which cannot be traced directly back to a single individual. It's just an emergent property of large groups of people.
Also, the GP's point wasn't that Microsoft is a bad company. S/he said that MS's culture is compatible with staid business applications, not creativity and style. A good CEO knows the corporate culture at a company and chooses goals accordingly.
Hah, you can be witty and right at the same time. Do you really think that firing guns is no more interesting than using a socket wrench? I know plenty of people (including myself) who have fired a gun in real life and still enjoy first person shooters. Plenty of people also go shooting for fun. No one uses a socket wrench for fun.
I don't have any strong feelings pro- or con- about gun control, but your whole guns = socket wrench position is ridiculous and must be motivated by politics.
Yep, kids find guns boring. That's why in all the computer games now you run around in first-person mode with a socket wrench.
So what? As everyone else said, there's no point in buying a TI to do real math and graphing—just buy a laptop and put Sage/Octave/R/whatever on it. The point of a TI is that it's a portable device with presumably circumscribed functionality well known to teachers and proctors. Whether or not I think this is the way exams should be conducted, why shouldn't TI feel free to pursue this market?
Clearly it would be bad if every device were like a TI calculator or iPhone, but as long as they are niche players, I see no problem in their actions. If anything it's the Ndless community that's being counterproductive here.
Wow, Chile must have a great educational system if every non-native English speaker writes English "properly". And you're very clever to catch him in such an obvious lie: he doesn't write English well, clearly he can't be Chilean!
http://www.theonion.com/articles/bush-finds-error-in-fermilab-calculations,1463/
Funny, but the fact that it's funny is also sad (sad that it's so ridiculous to think he'd be interested in science). Glad that what was parody is now sorta true.
The Economist is the world's best weekly newspaper. If you read what they say about the RIAA, including the first article which mentions how the RIAA's agressive tactics aren't working and are a lesson to other industries on what _not_ to do, you'd know that the Economist takes a moderate view on intellectual property.
In particular, they often report on academic research showing that IP laws are too strong. For instance, this article (subscription required) called "Killing Creativity" is about how overly strong IP laws can smother innovation.
Good point. Also, I hear their janitors only have BAs, so that's another way in which they rely on non-PHds.
Disclaimer: Years ago before getting papers on the internet was pervasive, I was once a TA. One of my students was struggling with coming up with a good paper topic, but in the end she turned in a great paper. When I complimented her on it, she seemed oddly evasive. A few years later I realized she had probably cheated.
I think plenty of good classes can be hurt by this service. For instance, suppose a student gets assigned to write an essay on <important work>. This essay might be open-ended (to allow the student to be creative) and a take-home assignment (for smart students who aren't necessarily the best under pressure or who can't physically write fast). I don't see why this is education done improperly or how this kind of assignment isn't meaningful. If it's impractical it may be impractical just because of services like this.
Uhh, not to say that you didn't have a very legitimate reason for switching banks, but if your interest rate moved from 9.9% to 17%, that's a 72% increase. So the difference in interest in not paying for a month would be more like a buck vs $1.72.