"The headline may well become true" might be a reason to publish the headline in the future, but it isn't a reason to publish it now. If you want to publish something now, fine, this is an interesting story -- but it needs a different headline.
The sex detection part is some fairly trivial signal processing. You need a band pass filter with a passband of about 0.5-3 Hz (at a guess; better numbers exist, but I haven't tried googling them). Then you need to detect extended signal within that band; there are a variety of options for this, any of which are likely to work.
Making your own instrumentation is common. I've done it in an industry setting, though not in a lab; but I've heard about plenty of labs that build their own special-purpose instruments. The end result is that you have to calibrate it. Obviously you can't use something like this without at least verifying the calibration, but that can be very easy.
Once you've calibrated it, what's not to like about a piece of consumer electronics that does the job?
If you honestly believe that would make more users read it in detail, then I have to conclude that you haven't tried to get users to read things. People will either read the policy or just trust their friends' implicit judgments. Having a clear, well-written policy / settings will help the former group; making the latter group click an extra button won't do much for the vast majority of them (certainly it will for some, but not most).
If you really want to raise awareness among most users, I see only two options: either keep repeating it with stories like this (except that you somehow get them to read), or come up with some brilliant radical departure from what's been done in the past.
Have you read their privacy settings page? It's a lot like what you describe. Except that it's longer, and has more options. Instead of yes / no, it lets you choose what groups can see the data in question. And there are more categories of data.
I still don't particularly trust Facebook with data I care about, largely because I'm well aware I'm not their customer. And I don't like the idea that they can change the policy later, and that I have no way to delete data from their servers. But I think they do a far better job than most of presenting their privacy policy and giving you options.
Of course, given that your friends already have access to whatever you've set to "Only Friends", and every app they install has whatever access they have to your profile, the "Only Friends" setting is a tad misleading.
Huh? Unless I'm mistaken, you can turn that off fairly easily:
If your friend uses an application that you do not use, you can control what types of information the application can access. Please note that applications will always be able to access your publicly available information (Name, Profile Picture, Gender, Current City, Networks, Friend List, and Pages) and information that is visible to Everyone.
On the other hand, Facebook's privacy policy isn't some dense legalese thing that no one reads. It's actually fairly clear. And the policy changes weren't hidden, you log in and they're presented to you. All it takes is reading a bit of perfectly ordinary English. Sure, it's a little long — but that seems to me more like they're being thorough than like they're trying to get you to ignore it.
So if the web site clearly explains their privacy policy, in an obvious, readable form, and people still have expectations based on things other than that document, whose fault is that? If the company makes reasonable efforts to be clear about what they plan to do, and people ignore them, is it the user's fault? The users still get the same message from their friends: Facebook is trustworthy and will respect your information. But when the privacy policy is well documented in regular English, it seems rather hard to say the company is at fault.
Well, seeing as creativity correlates with IQ (though in a slightly odd fashion), I'd say that hardly invalidates IQ as a metric for statistical purposes.
Again, what is the proposed alternative? Are you suggesting that we throw out IQ and simply stop studying statistical relationships involving intelligence? Or are you proposing some alternative measure of intelligence? The fact that IQ is not an ideal measure of intelligence does not mean it is not a useful measure of intelligence.
Of course, IQ does a remarkably good job at what it's intended to do: correlate with the sort of things we normally associate with intelligence, in the context of a statistical study. Sure, there are plenty of people who seem stupid in some ways but have high IQ; on average, though, it works well.
This is yet another case of people who know what IQ is actually supposed to be used for using it that way, and then the uninformed public complaining that it doesn't perfectly match something else.
Did you have some alternate metric that this study could have used in place of IQ that would do a better job?
Well, you could re-sterilize after fixing the robots, provided you can go long enough between fixing them. No need for a fully automated von Neumann factory, though it would certainly help. Of course, a fully automated von Neumann factory would help with lots of things. Just make sure you keep out any self-reproducing machines you don't want that might otherwise contaminate your vats. Perhaps you'll need a seal between the factory proper and the vats, and to disinfect the vats after breaking the seal for maintenance.
(Yes, I know that the phrase comes from the U.S. Declaration of Independence rather than Constitution, and therefore has no legal force. Nonetheless, if one subscribes to the notion of inalienable rights in the first place, they are inherently above laws.)
Actually, I believe you're mistaken there. The Constitution grants the government powers, and explicitly forbids it certain actions. In the process it recognizes some rights that we possess with or without it; that list is not, and was not intended to be, all-inclusive. It has been used in court on that basis, and referenced in decisions. Some googling turns up this summary, for example.
In the case of the death penalty, if your crimes are heinous enough (treason, murder, kidnapping and rape should all qualify IMHO) then I don't see any problems with society putting you out of our collective misery. My only issue with the death penalty is the fact that no justice system is 100% perfect, although I'm not convinced that spending your entire life behind bars for a crime that you didn't commit is anymore humane than being executed for it.
I tend to agree with you; however, the major reason I oppose the death penalty isn't that it's inhumane; it's that we make mistakes. Given an imperfect justice system (as all are), a life sentence made in error can be partially corrected later if new evidence comes to light. It's rare, but there have been a decent number of life sentences later reversed because of new evidence (in particular DNA evidence).
We owe it to the convicted to acknowledge that, in some cases, we make mistakes.
Thermodynamically disadvantaged? Huh? We're talking about what, 1-2 psi to inflate the buttons? At that pressure rise, the adiabatic temp increase is small (order of 10 degrees C). The majority of the work being done goes into P*V, just like it would if you replaced the air with a liquid.
I would assume that either way, the hard part is the micro-pump that can deliver useful pressures. Or maybe that's not that hard after all; I don't really follow the field.
Yeah. It was supposed to be under $100 including rebate according to the announcement, but it isn't. I may get the 80 GB Intel drive if the price drops appreciably before the Kingston one is widely available. I'm also considering waiting a little longer just to let other people find firmware bugs...
What few benchmarks I can find for the i-RAM make it look unimpressive when compared to an SSD. By the time I put 8GiB in it (barely enough for / and/home, and there are other things I'd like on the fast disk too) it costs as much as an X-25 — and the performance isn't that much better, if it all.
I'd rather have the extra space. Any of the good SSDs are fast enough, and a little more beyond that isn't worth a huge sacrifice in capacity (or increase in price). From a practical standpoint, I'd see a bigger performance boost by being able to put more of my data on the SSD than I would from having less of the data on the slightly faster iRAM.
The i-RAM would be a lot more interesting if it was a PCIe card, with performance numbers to match. There's no good reason a DRAM based device should have performance numbers limited by SATA speeds. (And yes, I'm aware such things exist — but the price tag removes them from the comparison for me.)
The drive to compare it to isn't a mechanical disk, but the Intel X-25M (or X-25E). The X-25M does better at small random writes, for a similar cost per GB. If you care about performance per dollar, the Intel drive still wins.
The plot was semi-log: the x-axis values were 1, 2, 4, which is an exponential progression, but were plotted at equal spacing. You would make it a linear plot by having the spacing between the depth=1 and depth=2 points be half the space between the depth=2 and depth=4 points. More interesting to me, I suspect, would be the log-log plot where the y-axis also varies logarithmically; then the linear behavior (doubling of queue depth causes doubling of average time) would show as linear, but so would other common scaling behaviors (doubling of queue depth causes 1.5x increase in average time, for example). However, that's probably harder to interpret for the average reader who isn't used to reading scientific papers and data sheets and the like.
Better graphs would be good; so would having the raw numbers that produced them. The format of the file copy tests is good, but I prefer the line plots for the queue depth data. Perhaps just a link to the raw numbers below the charts?
Speaking of the file copy tests, I just noticed something: the graph makes it look like the drives are uniformly faster at copying small files than large ones (since lower numbers are better, according to the chart heading). In reality, it looks like the majority of the difference in the size of the bars for a single drive comes from the fact that it takes it less time to copy less data! In order to see the effect of changes in file size, it would be nice to have all the data on the same scale, either by having the same number of bytes at each file size, or by giving the data in MB/s transfer rate. I'm not sure which I prefer, but having the numbers for the Colossus be 44.9, 44.7, 23.2, 6.4 is odd; they should really be 44.9, 44.7, 46.4, 64.
No one cares only about the raw random write speed, just like they don't care only about the raw sequential speed. There are times when either matter. That's why I'd like to see both the raw numbers and the numbers for a specific workload.
I find the average transaction time graphs hard to read. The semi-log plot format is a bit odd. It's also a bit hard to see relative changes when everything interesting is in the bottom 25-50% of the graph. Relative performance of two drives is easy enough to see, but between the semi-log plot and the sparse grid lines, it's really hard to see the scaling effects you're talking about.
The 256 GB drive is cheaper -- into the price range of the dedicated power user. OCZ doesn't sell to the server market, they sell to the desktop user. The power user on a desktop machine still cares about random write performance. They're far more likely to *also* care about sequential read / write performance, but they still care about random writes too.
Slashdot Mirror
"The headline may well become true" might be a reason to publish the headline in the future, but it isn't a reason to publish it now. If you want to publish something now, fine, this is an interesting story -- but it needs a different headline.
If that's Chase's policy, they should just explain that and be consistent about it, and far fewer people would be complaining.
And the thought of people's lives being ruined over doing something that did no harm to anyone doesn't sicken you?
The sex detection part is some fairly trivial signal processing. You need a band pass filter with a passband of about 0.5-3 Hz (at a guess; better numbers exist, but I haven't tried googling them). Then you need to detect extended signal within that band; there are a variety of options for this, any of which are likely to work.
Making your own instrumentation is common. I've done it in an industry setting, though not in a lab; but I've heard about plenty of labs that build their own special-purpose instruments. The end result is that you have to calibrate it. Obviously you can't use something like this without at least verifying the calibration, but that can be very easy.
Once you've calibrated it, what's not to like about a piece of consumer electronics that does the job?
If you honestly believe that would make more users read it in detail, then I have to conclude that you haven't tried to get users to read things. People will either read the policy or just trust their friends' implicit judgments. Having a clear, well-written policy / settings will help the former group; making the latter group click an extra button won't do much for the vast majority of them (certainly it will for some, but not most).
If you really want to raise awareness among most users, I see only two options: either keep repeating it with stories like this (except that you somehow get them to read), or come up with some brilliant radical departure from what's been done in the past.
Have you read their privacy settings page? It's a lot like what you describe. Except that it's longer, and has more options. Instead of yes / no, it lets you choose what groups can see the data in question. And there are more categories of data.
I still don't particularly trust Facebook with data I care about, largely because I'm well aware I'm not their customer. And I don't like the idea that they can change the policy later, and that I have no way to delete data from their servers. But I think they do a far better job than most of presenting their privacy policy and giving you options.
Of course, given that your friends already have access to whatever you've set to "Only Friends", and every app they install has whatever access they have to your profile, the "Only Friends" setting is a tad misleading.
Huh? Unless I'm mistaken, you can turn that off fairly easily:
If your friend uses an application that you do not use, you can control what types of information the application can access. Please note that applications will always be able to access your publicly available information (Name, Profile Picture, Gender, Current City, Networks, Friend List, and Pages) and information that is visible to Everyone.
On the other hand, Facebook's privacy policy isn't some dense legalese thing that no one reads. It's actually fairly clear. And the policy changes weren't hidden, you log in and they're presented to you. All it takes is reading a bit of perfectly ordinary English. Sure, it's a little long — but that seems to me more like they're being thorough than like they're trying to get you to ignore it.
So if the web site clearly explains their privacy policy, in an obvious, readable form, and people still have expectations based on things other than that document, whose fault is that? If the company makes reasonable efforts to be clear about what they plan to do, and people ignore them, is it the user's fault? The users still get the same message from their friends: Facebook is trustworthy and will respect your information. But when the privacy policy is well documented in regular English, it seems rather hard to say the company is at fault.
The kind slashdot editors like?
Well, seeing as creativity correlates with IQ (though in a slightly odd fashion), I'd say that hardly invalidates IQ as a metric for statistical purposes.
Again, what is the proposed alternative? Are you suggesting that we throw out IQ and simply stop studying statistical relationships involving intelligence? Or are you proposing some alternative measure of intelligence? The fact that IQ is not an ideal measure of intelligence does not mean it is not a useful measure of intelligence.
Of course, IQ does a remarkably good job at what it's intended to do: correlate with the sort of things we normally associate with intelligence, in the context of a statistical study. Sure, there are plenty of people who seem stupid in some ways but have high IQ; on average, though, it works well.
This is yet another case of people who know what IQ is actually supposed to be used for using it that way, and then the uninformed public complaining that it doesn't perfectly match something else.
Did you have some alternate metric that this study could have used in place of IQ that would do a better job?
Other research has shown a correlation between lack of ability and overestimation of ability in self-assessment.
Unskilled and unaware (pdf) is one such study. Very interesting stuff.
Well, you could re-sterilize after fixing the robots, provided you can go long enough between fixing them. No need for a fully automated von Neumann factory, though it would certainly help. Of course, a fully automated von Neumann factory would help with lots of things. Just make sure you keep out any self-reproducing machines you don't want that might otherwise contaminate your vats. Perhaps you'll need a seal between the factory proper and the vats, and to disinfect the vats after breaking the seal for maintenance.
(Yes, I know that the phrase comes from the U.S. Declaration of Independence rather than Constitution, and therefore has no legal force. Nonetheless, if one subscribes to the notion of inalienable rights in the first place, they are inherently above laws.)
Actually, I believe you're mistaken there. The Constitution grants the government powers, and explicitly forbids it certain actions. In the process it recognizes some rights that we possess with or without it; that list is not, and was not intended to be, all-inclusive. It has been used in court on that basis, and referenced in decisions. Some googling turns up this summary, for example.
Actually, by the time you count increased legal bills from appeals and such, life imprisonment is normally cheaper.
In the case of the death penalty, if your crimes are heinous enough (treason, murder, kidnapping and rape should all qualify IMHO) then I don't see any problems with society putting you out of our collective misery. My only issue with the death penalty is the fact that no justice system is 100% perfect, although I'm not convinced that spending your entire life behind bars for a crime that you didn't commit is anymore humane than being executed for it.
I tend to agree with you; however, the major reason I oppose the death penalty isn't that it's inhumane; it's that we make mistakes. Given an imperfect justice system (as all are), a life sentence made in error can be partially corrected later if new evidence comes to light. It's rare, but there have been a decent number of life sentences later reversed because of new evidence (in particular DNA evidence).
We owe it to the convicted to acknowledge that, in some cases, we make mistakes.
Thermodynamically disadvantaged? Huh? We're talking about what, 1-2 psi to inflate the buttons? At that pressure rise, the adiabatic temp increase is small (order of 10 degrees C). The majority of the work being done goes into P*V, just like it would if you replaced the air with a liquid.
I would assume that either way, the hard part is the micro-pump that can deliver useful pressures. Or maybe that's not that hard after all; I don't really follow the field.
Yeah. It was supposed to be under $100 including rebate according to the announcement, but it isn't. I may get the 80 GB Intel drive if the price drops appreciably before the Kingston one is widely available. I'm also considering waiting a little longer just to let other people find firmware bugs...
What few benchmarks I can find for the i-RAM make it look unimpressive when compared to an SSD. By the time I put 8GiB in it (barely enough for / and /home, and there are other things I'd like on the fast disk too) it costs as much as an X-25 — and the performance isn't that much better, if it all.
I'd rather have the extra space. Any of the good SSDs are fast enough, and a little more beyond that isn't worth a huge sacrifice in capacity (or increase in price). From a practical standpoint, I'd see a bigger performance boost by being able to put more of my data on the SSD than I would from having less of the data on the slightly faster iRAM.
The i-RAM would be a lot more interesting if it was a PCIe card, with performance numbers to match. There's no good reason a DRAM based device should have performance numbers limited by SATA speeds. (And yes, I'm aware such things exist — but the price tag removes them from the comparison for me.)
The drive to compare it to isn't a mechanical disk, but the Intel X-25M (or X-25E). The X-25M does better at small random writes, for a similar cost per GB. If you care about performance per dollar, the Intel drive still wins.
The plot was semi-log: the x-axis values were 1, 2, 4, which is an exponential progression, but were plotted at equal spacing. You would make it a linear plot by having the spacing between the depth=1 and depth=2 points be half the space between the depth=2 and depth=4 points. More interesting to me, I suspect, would be the log-log plot where the y-axis also varies logarithmically; then the linear behavior (doubling of queue depth causes doubling of average time) would show as linear, but so would other common scaling behaviors (doubling of queue depth causes 1.5x increase in average time, for example). However, that's probably harder to interpret for the average reader who isn't used to reading scientific papers and data sheets and the like.
Better graphs would be good; so would having the raw numbers that produced them. The format of the file copy tests is good, but I prefer the line plots for the queue depth data. Perhaps just a link to the raw numbers below the charts?
Speaking of the file copy tests, I just noticed something: the graph makes it look like the drives are uniformly faster at copying small files than large ones (since lower numbers are better, according to the chart heading). In reality, it looks like the majority of the difference in the size of the bars for a single drive comes from the fact that it takes it less time to copy less data! In order to see the effect of changes in file size, it would be nice to have all the data on the same scale, either by having the same number of bytes at each file size, or by giving the data in MB/s transfer rate. I'm not sure which I prefer, but having the numbers for the Colossus be 44.9, 44.7, 23.2, 6.4 is odd; they should really be 44.9, 44.7, 46.4, 64.
I'm glad you find the feedback useful.
Yes, different queue depths is good too.
No one cares only about the raw random write speed, just like they don't care only about the raw sequential speed. There are times when either matter. That's why I'd like to see both the raw numbers and the numbers for a specific workload.
I find the average transaction time graphs hard to read. The semi-log plot format is a bit odd. It's also a bit hard to see relative changes when everything interesting is in the bottom 25-50% of the graph. Relative performance of two drives is easy enough to see, but between the semi-log plot and the sparse grid lines, it's really hard to see the scaling effects you're talking about.
That's my plan. I've got the cheap, big disk... I just have to buy the small, fast SSD.
The 256 GB drive is cheaper -- into the price range of the dedicated power user. OCZ doesn't sell to the server market, they sell to the desktop user. The power user on a desktop machine still cares about random write performance. They're far more likely to *also* care about sequential read / write performance, but they still care about random writes too.