Now we have the opposite, though. The American tech industry of the past few decades has flourished by commercializing those good ideas you mention that had been invented at government labs, academia, etc., but never put outside the labs and commercialized. Now we're running out of good ideas to mine, so our main new tech product is new kinds of websites. Industry just doesn't do the fundamental research necessary to produce anything but incremental advances.
You could say that about any platform, though. OS X sucks because it lacks ports of most major PC games, for example. Maybe it's not Apple's fault, but as someone who plays games, I don't care who is at fault.
(Also, Windows sucks, because to get lots of stuff working you have to deal with half-ported abominations in cygwin.)
Microsoft's court papers cite a number of SGML editors that implement more or less the same features, in various combinations; anything that i4i implements that wasn't already in them is a pretty obvious extension or adaptation. The court didn't find that argument persuasive, because the non-obviousness bar for software patents is so ludicrously low that "clicking once" is a valid patent.
There are a lot of things you might want out of a game, of which immersion is only one. You might also want engagement, fun, thought-provokingness (okay, maybe less from an FPS), and lots of other qualities. There even some research showing that perfect immersion might harm some of these other properties, and may not be the sweet spot--- playing games on some perfectly immersive, like the Star Trek Holodeck, might not actually be what a lot of people want. I know I personally enjoy some mediation between myself and the virtual world; I like to feel that I'm playing a game, not actually in the world. But then i like turn-based and 2d games, too.
I don't think their use of it is wrong. The title is "On the detectability of habitable exomoons", and the abstract clarifies that to detect "habitable exomoons", this research proposes to detect "habitable-zone exomoons" (that phrase with the -zone qualification appears 4 times in the abstract), because presumably the actually habitable moons will be some subset of those.
You're right, it isn't currently known either way.
To review briefly,
P problems are those solvable in polynomial time on a regular computer.
NP problems are (one definition) those verifiable in polynomial time on regular computers. That is, if you gave the answer to the problem, in polynomial time I could tell you if it was the correct one.
QBP problems are those solvable in polynomial time on a quantum computer.
It is not known whether any of these classes are equivalent. However, the possibilities are constrained by,
NP-complete, which are problems in NP to which all other NP problems can be reduced (provably!) in polynomial time.
Traveling salesman is NP-complete. Therefore, if we found a polynomial-time algorithm on regular computers, P = NP. If we found a polynomial-time algorithm on quantum computers, QBP = NP.
Integer factorization is in NP, but not known to be either NP-complete or in P. Therefore, a polynomial-time algorithm on regular computers could exist without P = NP--- but we don't know of one. Shor's algorithm (the subject of this article) is a polynomial-time algorithm for quantum computers, so integer factorization is in QBP. However, since integer factorization isn't NP-complete, this doesn't have any implications for whether QBP = NP or not.
So it's not provably known that integer factorization is easier than traveling salesman on any kind of computer. But on quantum computers, the fastest known integer factorization algorithm is polynomial, while the only way we could do that for traveling salesman is if QBP = NP. On regular computers, no polynomial algorithm is known for either problem. But in a sense it'd be more surprising if one were found for traveling salesman, because that would imply P = NP... while finding one for integer factorization wouldn't have such wide-ranging implications on other problems (though it might have implications for other not-yet-known-to-be-in-P problems, if the technique were transferable).
Currently, they and the traditional techniques can each manipulate 4 non-error-checking qubits. =]
The article argues that their approach is more promising for scaling up, though, and has fewer inherent limits to doing so. That of course is still to be demonstrated, but the result still seems interesting--- basically, here's proof-of-concept of a new method that at least works as well as previous methods, along with some reasons to believe it might scale up better.
That's their claim. The full version of the article says of previous implementations, "these approaches cannot be scaled to a large number of qubits because of purity, size, and stability limitations of these systems". And of theirs: "Although it currently uses an inefficient single photon source and modest efficiency detectors, ongoing progress to address heralded gates and efficient sources and detectors combined with the results presented here will allow large-scale quantum circuits on many qubits to be implemented".
They only factored the number 15 here as well--- in fact what they implemented was a version of the algorithm compiled to a specialized implementation for the input "15". Their claim from why it's an improvement is (from the full article):
[P]roof-of-principle demonstrations of Shor's algorithm have so far only been possible with liquid-state nuclear magnetic resonance and bulk optical implementations of simplified logic gates, owing to the need for several logic gates operating on several qubits, even for small-scale compiled versions. However, these approaches cannot be scaled to a large number of qubits because of purity, size, and stability limitations of these systems. We demonstrate a compiled version of Shor's algorithm operating on four qubits in which the processing occurs in a photonic circuit of several one- and two-qubit gates fabricated from integrated optical waveguides on a silica-on-silicon chip.
Essentially they claim that, while their demonstration here is as small-scale as the previous ones, it's at least plausible that it might scale up, while the previous demonstrations have inherent limitations that prevent them from scaling up.
If there was some previous result that only 2/3 of filesharers admit it when asked, then an upwards revision by 1/3 in an estimate would be defensible. A "hunch" is not quite as good evidence. of course.
I was objecting mainly to the "how 136 people became 7 million" title, which to my ears reads mainly as a criticism of the sample size. But whatever the problems with this estimate, the sample size wasn't really among them.
It doesn't really make sense to claim "sample size is small" for an 1,100-person sample. If the sampling was done in a random, unbiased manner, that size sample gives a margin of error of +/- 3%. If there are flaws in the sampling method, that's another thing, but the sample size alone doesn't seem problematic, unless you need accuracy better than +/- 3%.
Some of the estimation steps might be sketchy, but the basic practice of estimating a population proportion from a sample of that population is not particularly questionable. That's how almost all studies of populations work, because taking censuses of all people in a country is rarely feasible. We have century-old statistical theory on how to put bounds on the sampling error, too, assuming the sample was indeed random.
You could have a whole slew of these stories if you really objected to that basic methodology, e.g. nearly every estimate of N million people suffering from a disease or disorder is based on a sample.
Not that I'm against animal welfare, mind you, but I think part of respecting animals is realizing they aren't all that interested in climbing Maslow's pyramid [wikipedia.org] all the way to the top.
Yeah, that's a fairly common compatibility workaround, and it's fortunately encapsulated in a number of libraries (e.g. this one), so the individual developer can often avoid having to know about SVGVML mapping issues.
I'm not talking about the patentability of his particular novel claim, but whether the rest of his implementation, in its more mundane parts, will step on some BS patent. That seems to be his main worry, and it's pretty much unavoidable--- there are a whole lot of overbroad patents out there, of the Amazon One-Click and IBM Twitter Updates variety.
It sounds like you're confident that the core of the product is novel and not already patented. So the issue is accidentally stepping on a bunch of overly broad patents for stupid things? Unfortunately, the crux of the patent mess is precisely that: it stifles innovation because there is no good way to know you're not stepping on a bunch of stupid, overly broad patents, that will take a lot of money to litigate even if they turn out to be invalid.
Well, I don't have one myself, so in some sense I must agree with the not-worth-it assessment. But it's not really expensive compared to a netbook, if we're talking about the Kindle 2 (the subject of this article) rather than the Kindle DX. It costs $299, which is basically the going rate for netbooks. So it'd be really deciding on features rather than price.
Kindle wins on: battery life, daylight visibility of the display, weight, free 3G internet
Netbooks win on: hardware (CPU/ram/hdd/etc.), color display, can run a normal OS without heroics
Just depends on what you want, I think. Do you care more about the 1.6 GHz Atom vs. 400 MHz XScale? Do you care more about a weight of 2/3 lbs vs 2 lbs? Etc.
It gives you a better base on which to start customizing the device--- e.g. once there was Linux on the XBox, people started producing software to turn them into media centers. The built-in OS on Kindle can't do much, and it's not easy to modify it to have it do more. For example, even on the Kindle's hardware it should be possible to have a better web browser than the really bad one that's built in.
The Kindle isn't limited to accessing Wikipedia and Amazon even with the default OS-- there's a web browser under the "experimental" features in the default menu. Amazon doesn't play it up much partly because it's not very good, and partly because presumably they'd lose money if people bought Kindles just for web browsing.
That isn't the definition generally used, though. Something like, "I think [racial epithet]s are genetically incapable of intelligent thought" is hate speech under many speech codes, but is not advocating or inciting unlawful conduct to be directed at any person or identifiable group of people.
In the legal sense, no. But the freedom of intellectual inquiry that's a bedrock of the university requires tolerance for a diversity of views, which is unfortunately not popular among the current crop of students (and some professors).
Fun fact: when shareholders vote no on executive pay, the companies often pay it anyway, because companies are run by lawless boards and executives, not actually beholden to their nominal owners. Until legal reforms happen that give shareholders a truly legally enforceable "say on pay", that will continue to be the case.
Slashdot Mirror
Now we have the opposite, though. The American tech industry of the past few decades has flourished by commercializing those good ideas you mention that had been invented at government labs, academia, etc., but never put outside the labs and commercialized. Now we're running out of good ideas to mine, so our main new tech product is new kinds of websites. Industry just doesn't do the fundamental research necessary to produce anything but incremental advances.
You could say that about any platform, though. OS X sucks because it lacks ports of most major PC games, for example. Maybe it's not Apple's fault, but as someone who plays games, I don't care who is at fault.
(Also, Windows sucks, because to get lots of stuff working you have to deal with half-ported abominations in cygwin.)
Microsoft's court papers cite a number of SGML editors that implement more or less the same features, in various combinations; anything that i4i implements that wasn't already in them is a pretty obvious extension or adaptation. The court didn't find that argument persuasive, because the non-obviousness bar for software patents is so ludicrously low that "clicking once" is a valid patent.
There are a lot of things you might want out of a game, of which immersion is only one. You might also want engagement, fun, thought-provokingness (okay, maybe less from an FPS), and lots of other qualities. There even some research showing that perfect immersion might harm some of these other properties, and may not be the sweet spot--- playing games on some perfectly immersive, like the Star Trek Holodeck, might not actually be what a lot of people want. I know I personally enjoy some mediation between myself and the virtual world; I like to feel that I'm playing a game, not actually in the world. But then i like turn-based and 2d games, too.
I don't think their use of it is wrong. The title is "On the detectability of habitable exomoons", and the abstract clarifies that to detect "habitable exomoons", this research proposes to detect "habitable-zone exomoons" (that phrase with the -zone qualification appears 4 times in the abstract), because presumably the actually habitable moons will be some subset of those.
You're right, it isn't currently known either way.
To review briefly,
P problems are those solvable in polynomial time on a regular computer.
NP problems are (one definition) those verifiable in polynomial time on regular computers. That is, if you gave the answer to the problem, in polynomial time I could tell you if it was the correct one.
QBP problems are those solvable in polynomial time on a quantum computer.
It is not known whether any of these classes are equivalent. However, the possibilities are constrained by,
NP-complete, which are problems in NP to which all other NP problems can be reduced (provably!) in polynomial time.
Traveling salesman is NP-complete. Therefore, if we found a polynomial-time algorithm on regular computers, P = NP. If we found a polynomial-time algorithm on quantum computers, QBP = NP.
Integer factorization is in NP, but not known to be either NP-complete or in P. Therefore, a polynomial-time algorithm on regular computers could exist without P = NP--- but we don't know of one. Shor's algorithm (the subject of this article) is a polynomial-time algorithm for quantum computers, so integer factorization is in QBP. However, since integer factorization isn't NP-complete, this doesn't have any implications for whether QBP = NP or not.
So it's not provably known that integer factorization is easier than traveling salesman on any kind of computer. But on quantum computers, the fastest known integer factorization algorithm is polynomial, while the only way we could do that for traveling salesman is if QBP = NP. On regular computers, no polynomial algorithm is known for either problem. But in a sense it'd be more surprising if one were found for traveling salesman, because that would imply P = NP... while finding one for integer factorization wouldn't have such wide-ranging implications on other problems (though it might have implications for other not-yet-known-to-be-in-P problems, if the technique were transferable).
Currently, they and the traditional techniques can each manipulate 4 non-error-checking qubits. =]
The article argues that their approach is more promising for scaling up, though, and has fewer inherent limits to doing so. That of course is still to be demonstrated, but the result still seems interesting--- basically, here's proof-of-concept of a new method that at least works as well as previous methods, along with some reasons to believe it might scale up better.
That's their claim. The full version of the article says of previous implementations, "these approaches cannot be scaled to a large number of qubits because of purity, size, and stability limitations of these systems". And of theirs: "Although it currently uses an inefficient single photon source and modest efficiency detectors, ongoing progress to address heralded gates and efficient sources and detectors combined with the results presented here will allow large-scale quantum circuits on many qubits to be implemented".
They only factored the number 15 here as well--- in fact what they implemented was a version of the algorithm compiled to a specialized implementation for the input "15". Their claim from why it's an improvement is (from the full article):
Essentially they claim that, while their demonstration here is as small-scale as the previous ones, it's at least plausible that it might scale up, while the previous demonstrations have inherent limitations that prevent them from scaling up.
Basically, except that the confidence level for the interval is 95%, not 50%. Should've quoted that, but 95% is the usual assumed one.
If there was some previous result that only 2/3 of filesharers admit it when asked, then an upwards revision by 1/3 in an estimate would be defensible. A "hunch" is not quite as good evidence. of course.
I was objecting mainly to the "how 136 people became 7 million" title, which to my ears reads mainly as a criticism of the sample size. But whatever the problems with this estimate, the sample size wasn't really among them.
It doesn't really make sense to claim "sample size is small" for an 1,100-person sample. If the sampling was done in a random, unbiased manner, that size sample gives a margin of error of +/- 3%. If there are flaws in the sampling method, that's another thing, but the sample size alone doesn't seem problematic, unless you need accuracy better than +/- 3%.
Some of the estimation steps might be sketchy, but the basic practice of estimating a population proportion from a sample of that population is not particularly questionable. That's how almost all studies of populations work, because taking censuses of all people in a country is rarely feasible. We have century-old statistical theory on how to put bounds on the sampling error, too, assuming the sample was indeed random.
You could have a whole slew of these stories if you really objected to that basic methodology, e.g. nearly every estimate of N million people suffering from a disease or disorder is based on a sample.
And precious few humans are, either!
Yeah, that's a fairly common compatibility workaround, and it's fortunately encapsulated in a number of libraries (e.g. this one), so the individual developer can often avoid having to know about SVGVML mapping issues.
I'm not talking about the patentability of his particular novel claim, but whether the rest of his implementation, in its more mundane parts, will step on some BS patent. That seems to be his main worry, and it's pretty much unavoidable--- there are a whole lot of overbroad patents out there, of the Amazon One-Click and IBM Twitter Updates variety.
It sounds like you're confident that the core of the product is novel and not already patented. So the issue is accidentally stepping on a bunch of overly broad patents for stupid things? Unfortunately, the crux of the patent mess is precisely that: it stifles innovation because there is no good way to know you're not stepping on a bunch of stupid, overly broad patents, that will take a lot of money to litigate even if they turn out to be invalid.
Well, to get a better web browser, for one.
They don't lose money on the hardware, but they're selling you the hardware and a few years of unlimited 3G connectivity for that $299.
Well, I don't have one myself, so in some sense I must agree with the not-worth-it assessment. But it's not really expensive compared to a netbook, if we're talking about the Kindle 2 (the subject of this article) rather than the Kindle DX. It costs $299, which is basically the going rate for netbooks. So it'd be really deciding on features rather than price.
Kindle wins on: battery life, daylight visibility of the display, weight, free 3G internet
Netbooks win on: hardware (CPU/ram/hdd/etc.), color display, can run a normal OS without heroics
Just depends on what you want, I think. Do you care more about the 1.6 GHz Atom vs. 400 MHz XScale? Do you care more about a weight of 2/3 lbs vs 2 lbs? Etc.
It gives you a better base on which to start customizing the device--- e.g. once there was Linux on the XBox, people started producing software to turn them into media centers. The built-in OS on Kindle can't do much, and it's not easy to modify it to have it do more. For example, even on the Kindle's hardware it should be possible to have a better web browser than the really bad one that's built in.
The Kindle isn't limited to accessing Wikipedia and Amazon even with the default OS-- there's a web browser under the "experimental" features in the default menu. Amazon doesn't play it up much partly because it's not very good, and partly because presumably they'd lose money if people bought Kindles just for web browsing.
That isn't the definition generally used, though. Something like, "I think [racial epithet]s are genetically incapable of intelligent thought" is hate speech under many speech codes, but is not advocating or inciting unlawful conduct to be directed at any person or identifiable group of people.
In the legal sense, no. But the freedom of intellectual inquiry that's a bedrock of the university requires tolerance for a diversity of views, which is unfortunately not popular among the current crop of students (and some professors).
Fun fact: when shareholders vote no on executive pay, the companies often pay it anyway, because companies are run by lawless boards and executives, not actually beholden to their nominal owners. Until legal reforms happen that give shareholders a truly legally enforceable "say on pay", that will continue to be the case.