Not if they're trying to measure the same thing - then the outputs should agree within experimental errors.
Sorry, but that is simply not so. They are measuring a dynamic system, using instruments that introduce several kinds of known, quantified errors (see error bars.) For instance, each instrument has a static offset and a random error, with latter being represented by its pdf. I'm amazed that I still remember some statistics:-) On top of that, each of these functions may drift over time due to natural and technical reasons.
My example was meant to illustrate this very problem. They are filtering the data using a low-pass filter (roughly so.) But the bandwidth of that filter (cutoff frequency) and the slope of the filter *affect the results* ! You can't do anything about it because that's how math works. You can have the needle dancing between 1 and 3, or you can have the needle glued to 2. And that is only assuming that they don't discard outliers, as they should do, and Kalman filter does exactly that. There are tons of matrices of coefficients that control the smoothing process; these numbers are picked more or less by hand, to trade resolution in time for resolution in amplitude. But the output *will* depend on your choice of smoothing methods and coefficients.
we have no background as to why it was done or what was done with the output
Sure, it's theoretically possible that they just wrote this code for fun and never used it for anything.
However, per ESR (and per your own eyes,) the fudge function matches the hockey stick graph. That is a very serious coincidence. Besides, note the uppercase "VERY ARTIFICIAL" and the comment "fudge factor" - these words aren't what you use when you apply a proper correction. Also note that the correction is awfully broad, it covers decades with just a one or two data points! What on Earth would explain the need for such a correction? That would be something to apply if our Sun were dimming and brightening over decades, but it hadn't done that. So the physical origin of that fudge factor is very suspicious. I'd like to hear their *scientific* explanation of that, and it better not be "Prof. Jones gave me this curve to apply."
If their code was buggy and produced the wrong answer, nobody *would* be able to reproduce the results
Naturally. But there would be other, valid and honest, reasons why two loosely similar algorithms may produce different results. For example, look at the average of this set:
1, 2, 3, 2, 1, 2, 20, 2, 1
There are several correct methods to calculate the result, and they will produce different answers. Simple averaging will give you 3.7(7). But a Kalman filter will give you a much lower value. If you fit a curve onto these points and integrate the curve you will get yet another result that depends on what curve you used. And so on... and that is just a simplest phase of processing of data.
The only way to do a coherent, meaningful analysis of someone's complex code is to have full access to that code. Then you have several options. You can prove correctness of the code. You can repeat the calculations using a different algorithm and prove that the differences are exactly what they should be. But you can never conclusively tell that some black box is correct or incorrect, unless you amass a larger number of competing codebases and then just use them to vote. But that's awfully wasteful.
just what the algorithm was intended to do. That's enough to duplicate the analysis
Much of coding at CRU was done by students with no CS education. They may have had great algorithms in mind, and they could perfectly describe what these algorithms are intended to do... but their code had bugs, and these algorithms produced something else. How in the world would anyone duplicate their analysis?
This is not a contrived scenario. If you look into HARRY_READ_ME file, Harry found tons of such bugs in the codebase that he one day was given to maintain and use. In some cases the code was so wrong that it skipped whole sections of the input data. In other cases the software was so undocumented that you couldn't even figure out how to run the thing. And read about their naming conventions for files... they were an IT disaster.
agree that at some point it is no longer worth it, and that implicitly we do place value on a humans lives. But how much is it worth?
It is worth much more than it would cost to make the launch vehicle safe. The STS problem - and its death toll - is in deliberate design that made emergency escape impossible pretty much in any part of the launch or descent. Capsule based designs could survive both incidents if the capsule is strong enough to perform a ballistic reentry on its own. The problem is that you can't make such a capsule large enough to hold 7 people. STS design went for capacity and payload, at great risk to safety.
The point you are missing is that "bare-metal code" is assembler, regardless of how much effort is involved.
I again have to point you to Linux or *BSD, these OSes have real time drivers in C. I don't recall seeing *any* peripheral driver in Linux that is not C. Practically all assembly code is under arch/ which means bootstrap, memory initialization and main timers. The rest is C.
Go ahead and write a real time driver in C, let me know how it works for you.
I'm doing this right now, and it is a very usual thing for me to do because I work on firmware for slower microcontrollers that run at clock speeds from 1.8 to 16 MHz. I have tons of peripherals in the MCU, and they must be serviced on time. A typical MCU project is a real time design. Sometimes I profile the code by connecting an oscilloscope to some spare pins and check that I have enough time in critical parts of the code. And *all* the code is C, compiled by avr-gcc 4.3.2. I have maybe 0.1% of the code that is in assembler, and that is stock macros that come with avr-gcc.
To illustrate, here you can see the lowest level of avr32 support, and you can observe how many LOCs are in.S and how many are in.c files. If still not convinced, visit mm/ and see what language the do_page_fault() is implemented in; that is one of most performance-critical pieces of code. C today is the "bare-metal" language of choice, and it works well in that role.
let me know when they start writing "writing tight, bare-metal code" (i.e. assembler). C isn't that.
Linux is 99% C and 1% assembler. Today it would be foolish to write the entire OS in assembler - at least because the compiler often optimizes the machine code better than you do, unless you really spend time on it. I wrote assembly code years ago, and it is minimally maintainable or reviewable. Imagine how far Linux 1.00 would go if it was all in assembly... and you'd immediately lose portability.
Now, after we find out that much of the experimental and observational basis for Global Warmology is actually a scam...
Yes, all of it. Every single journal or proceedings containing papers relevant to global warming research is now debunked... that was sarcasm.
Despite the sarcasm, it is unclear what you are trying to say. Of course, not everything is automatically tainted; the OP was careful to say "much", not "every single *".
I think that indeed, much of the published results needs a review now. If a publication was based on tainted data, or used improper methods, then indeed it is in doubt. But if a work was done honestly and is based on reasonably good data, there isn't anything wrong with it. I have no idea what the percentage is, but that will become known soon. I'm sure there will be many honest scientists who just were given "adjusted" data, simply because CRU wasn't giving their source data to anyone.
It's clear from the emails that the science community now views any skeptic, no matter how reputable, as 100% hostile
s/science community/liars/
That's where the problem is, and the emails only confirm what was suspected all along. Nobody cares any more what CRU people think - they are done for. Not for their opinions - for what they did. Emails only explain why they concealed data and unethically suppressed critics.
I would be required to leave a backdoor into my machine, so that the police (or whatever government agency) could RDP in if my kid pushes this panic button?
I don't mean to advocate for or against such a thing in principle, I just don't know enough. However from the technical POV you do not need to leave a backdoor into your computer. The IM software only needs to have a plugin that implements the dolphin button, and that plugin, when activated, gives remote access to *that IM session only*. The remote user does not need to access anything above and beyond the IM software. That software may be augmented to provide additional, technical details about the session, so that the other person can be located. The plugin will not be a backdoor; it is even possible to release it as open source, and you can compile one yourself. After all, it's a very simple piece of code. I don't know how many LOCs a plugin wrapper may need, but the TCP session to a certain host won't take more than 100 lines. When the remote session is open the incoming data is sent there too, and the remote input is injected into the IM's outgoing stream. Or you may block the local I/O entirely, if you don't want the child to see what the police is talking about.
The point is the cops told him to do something and he didn't do it...
The order must be lawful. In this case I believe the order to ask the crowd to disperse was lawful, and it doesn't matter if the message is shouted through the bullhorn or sent through Twitter.
The police could themselves tell the crowd to go home, but from every POV it is better if the organizers do that - they have better contact with the crowd. That's why he was approached and asked to cooperate.
Nobody should foist 4.5 MB images onto anyone, this is not a format for distribution regardless of how you do it. The problem needs to be fixed where it originates: you need to configure her camera to default to a more reasonable format. I can understand taking raw images if you are a professional, but then you'd be importing them right into Photoshop and you'd know what to do with them.
But if you or your mother are not professionals that need every pixel then you should change the settings on the camera. You should end up with a JPEG that is anywhere from 200 to 300 kB, and you can't see any loss of quality in that. Also the flash card in the camera will hold more of those.
It would seem that there might be a distinction, and I'm curious where people draw the line.
I think the line is clearly visible:
Case A (a CD): anyone can have a copy for $10-20. The IP owner's business is not in trouble because you bought (or even stole) a music CD. At worst they lost those $10-20 that you didn't pay.
Case B (an IP): nobody can have a copy for any reasonable amount of money. The IP owner's business may be in trouble because the data was stolen, and many people may lose jobs if the business folds.
So to tell those cases apart you only need to ask if the data is ultimately intended for public use or not.
If you understand the risks of what happens on Facebook well enough to make an informed decision to put your stuff up there, you probably understand it well enough to throw up a quick web server.
Yes. But that means that 99% of people will use Facebook and 1% of people will NOT use Facebook. Out of that 1% maybe 10% will want to share a file using Unity. So Opera apparently made a business decision to go for the 0.1% of the market. I wish them luck, they will need it.
Considering that most Americans are in the driver's seat for their own retirement these days, that should be basically everyone.
Among my acquaintances and myself, very few actively manage their 401k. In practice it is easier and more profitable to buy a few mutual funds. Many 401k's, like Fidelity, are not set up for active trading; for example, your order will be fulfilled within 24 hours. This is because 401k is intended for long term investments. If you do switch securities, or change your allocations, it is done after careful research online - not based on a random sample from a random issue of WSJ. Mutual funds also offer greater stability (relative to the market) because they buy large number of different securities and they pay attention to what happens to those securities. It's a full time job.
But of course if you want to trade actively then the same Fidelity offers you an account for that too, and it has all the necessary features for trading (such as conditions for the sale.) Whether or not an active investor needs WSJ is a matter of opinion, but a mere 401k contributor most definitely don't need it, IMO. The 401k manager most likely limits his investment choices to a handful of mutual funds anyway. You only can pick the fund, one or many, that buys what you are interested in. That's how things are done at Fidelity 401k at least.
Could be. Perhaps not his own $1B, but how do you know that mspohr is not involved with managing a mutual fund, for example? Who else but an investor would be reading WSJ for any reason?
So why the fuck would anyone want to be on facebook?
People who can think more than 5 minutes ahead don't want to be on facebook, and they aren't. However the majority of people only care of "now", even if that much. They post a message or a photo because they can't be bothered to think what it can cause in the future.
To make things worse, it generally hurts your privacy to have a social life these days. Even if *you* didn't take the photo and didn't post it for all to see, someone else - out of tens of random people you met - can do so and attach your name to the photo, all that without you knowing. And that's how the data trail grows.
Even on/., formerly a bastion of near-paranoid sysadmins, posts appeared (a few years ago, IIRC) saying "the war for privacy is lost, so abandon all hope, drop all defenses and have fun!" And there is indeed a serious push to abandon all hope because it takes more and more effort with every passing day to keep your privacy, and it certainly puts constraints on how you deal with your friends and how you participate in social events. I do not expect majority of the population, especially teenagers, to sacrifice that much at cost of what they value most [at the moment] - their friends and their social life.
Seriously though, how complete and accurate a picture of you and your activities could I create with complete and total access to your [*] ?
You can create a fairly decent picture of a law-abiding person. However a terrorist would present the same profile to you, and you can't tell the difference. He'd be paying cash for his nefarious goods; he will be walking or taking a taxi to meetings of conspirators; he would not use his home computer to contact his terorist superiors, or he'd use a method that is untraceable. If he must use a webmail he will pick a foreign one, not Google. He may even drive to a random residential neighborhood, connect to a non-encrypted AP and send his message, then leave - all within minutes.
This means that a total surveillance society is not any safer than a zero surveillance society. Criminals simply need to work around the surveillance that is there; it may take more work to do it, but they are patient: "The guard may forget that he is watching a prisoner, but the prisoner cannot forget that he is being watched."
In the U.S. at least, using genetic information to deny insurance coverage is illegal.
But it is incredibly profitable, and so it will be done anyway. The insurance company only needs to cite some other factors while rejecting the application.
A business can already, for example, refuse to hire black people (which is illegal) by simply not explaining why they were rejected. And if there is a lawsuit the business owner can always say "it's because some other applicant did better at the interview" and that's that. It would take a smoking gun, like an HR memo spelling out the "no blacks" policy, to lose the court case.
anyone here know where I could get one (or at least, a 30x30x7 (mm))?
Don't know about x7, but here is 30mm L x 30mm H x 6mm W fan. This is a 5V part without tachometer. There is also 259-1327-ND which produces higher airflow (and is noisier, I'd guess.)
Slashdot Mirror
Not if they're trying to measure the same thing - then the outputs should agree within experimental errors.
Sorry, but that is simply not so. They are measuring a dynamic system, using instruments that introduce several kinds of known, quantified errors (see error bars.) For instance, each instrument has a static offset and a random error, with latter being represented by its pdf. I'm amazed that I still remember some statistics :-) On top of that, each of these functions may drift over time due to natural and technical reasons.
My example was meant to illustrate this very problem. They are filtering the data using a low-pass filter (roughly so.) But the bandwidth of that filter (cutoff frequency) and the slope of the filter *affect the results* ! You can't do anything about it because that's how math works. You can have the needle dancing between 1 and 3, or you can have the needle glued to 2. And that is only assuming that they don't discard outliers, as they should do, and Kalman filter does exactly that. There are tons of matrices of coefficients that control the smoothing process; these numbers are picked more or less by hand, to trade resolution in time for resolution in amplitude. But the output *will* depend on your choice of smoothing methods and coefficients.
we have no background as to why it was done or what was done with the output
Sure, it's theoretically possible that they just wrote this code for fun and never used it for anything.
However, per ESR (and per your own eyes,) the fudge function matches the hockey stick graph. That is a very serious coincidence. Besides, note the uppercase "VERY ARTIFICIAL" and the comment "fudge factor" - these words aren't what you use when you apply a proper correction. Also note that the correction is awfully broad, it covers decades with just a one or two data points! What on Earth would explain the need for such a correction? That would be something to apply if our Sun were dimming and brightening over decades, but it hadn't done that. So the physical origin of that fudge factor is very suspicious. I'd like to hear their *scientific* explanation of that, and it better not be "Prof. Jones gave me this curve to apply."
If their code was buggy and produced the wrong answer, nobody *would* be able to reproduce the results
Naturally. But there would be other, valid and honest, reasons why two loosely similar algorithms may produce different results. For example, look at the average of this set:
There are several correct methods to calculate the result, and they will produce different answers. Simple averaging will give you 3.7(7). But a Kalman filter will give you a much lower value. If you fit a curve onto these points and integrate the curve you will get yet another result that depends on what curve you used. And so on... and that is just a simplest phase of processing of data.
The only way to do a coherent, meaningful analysis of someone's complex code is to have full access to that code. Then you have several options. You can prove correctness of the code. You can repeat the calculations using a different algorithm and prove that the differences are exactly what they should be. But you can never conclusively tell that some black box is correct or incorrect, unless you amass a larger number of competing codebases and then just use them to vote. But that's awfully wasteful.
No obvious "cooking" was there, at least to my eyes
Many eyes make all bugs shallow. Read what ESR has to say about *specifically* cooking the data. Now your eyes see it too.
just what the algorithm was intended to do. That's enough to duplicate the analysis
Much of coding at CRU was done by students with no CS education. They may have had great algorithms in mind, and they could perfectly describe what these algorithms are intended to do ... but their code had bugs, and these algorithms produced something else. How in the world would anyone duplicate their analysis?
This is not a contrived scenario. If you look into HARRY_READ_ME file, Harry found tons of such bugs in the codebase that he one day was given to maintain and use. In some cases the code was so wrong that it skipped whole sections of the input data. In other cases the software was so undocumented that you couldn't even figure out how to run the thing. And read about their naming conventions for files... they were an IT disaster.
agree that at some point it is no longer worth it, and that implicitly we do place value on a humans lives. But how much is it worth?
It is worth much more than it would cost to make the launch vehicle safe. The STS problem - and its death toll - is in deliberate design that made emergency escape impossible pretty much in any part of the launch or descent. Capsule based designs could survive both incidents if the capsule is strong enough to perform a ballistic reentry on its own. The problem is that you can't make such a capsule large enough to hold 7 people. STS design went for capacity and payload, at great risk to safety.
The point you are missing is that "bare-metal code" is assembler, regardless of how much effort is involved.
I again have to point you to Linux or *BSD, these OSes have real time drivers in C. I don't recall seeing *any* peripheral driver in Linux that is not C. Practically all assembly code is under arch/ which means bootstrap, memory initialization and main timers. The rest is C.
Go ahead and write a real time driver in C, let me know how it works for you.
I'm doing this right now, and it is a very usual thing for me to do because I work on firmware for slower microcontrollers that run at clock speeds from 1.8 to 16 MHz. I have tons of peripherals in the MCU, and they must be serviced on time. A typical MCU project is a real time design. Sometimes I profile the code by connecting an oscilloscope to some spare pins and check that I have enough time in critical parts of the code. And *all* the code is C, compiled by avr-gcc 4.3.2. I have maybe 0.1% of the code that is in assembler, and that is stock macros that come with avr-gcc.
To illustrate, here you can see the lowest level of avr32 support, and you can observe how many LOCs are in .S and how many are in .c files. If still not convinced, visit mm/ and see what language the do_page_fault() is implemented in; that is one of most performance-critical pieces of code. C today is the "bare-metal" language of choice, and it works well in that role.
let me know when they start writing "writing tight, bare-metal code" (i.e. assembler). C isn't that.
Linux is 99% C and 1% assembler. Today it would be foolish to write the entire OS in assembler - at least because the compiler often optimizes the machine code better than you do, unless you really spend time on it. I wrote assembly code years ago, and it is minimally maintainable or reviewable. Imagine how far Linux 1.00 would go if it was all in assembly... and you'd immediately lose portability.
Yes, all of it. Every single journal or proceedings containing papers relevant to global warming research is now debunked ... that was sarcasm.
Despite the sarcasm, it is unclear what you are trying to say. Of course, not everything is automatically tainted; the OP was careful to say "much", not "every single *".
I think that indeed, much of the published results needs a review now. If a publication was based on tainted data, or used improper methods, then indeed it is in doubt. But if a work was done honestly and is based on reasonably good data, there isn't anything wrong with it. I have no idea what the percentage is, but that will become known soon. I'm sure there will be many honest scientists who just were given "adjusted" data, simply because CRU wasn't giving their source data to anyone.
It's clear from the emails that the science community now views any skeptic, no matter how reputable, as 100% hostile
That's where the problem is, and the emails only confirm what was suspected all along. Nobody cares any more what CRU people think - they are done for. Not for their opinions - for what they did. Emails only explain why they concealed data and unethically suppressed critics.
I would NEVER use a word like "Hide" in context of normalizing a dataset. That smacks way too much of fraudulent data manipulation.
That's nothing. You should see their source code... what do you call this?
You don't even need to plot this.
I would be required to leave a backdoor into my machine, so that the police (or whatever government agency) could RDP in if my kid pushes this panic button?
I don't mean to advocate for or against such a thing in principle, I just don't know enough. However from the technical POV you do not need to leave a backdoor into your computer. The IM software only needs to have a plugin that implements the dolphin button, and that plugin, when activated, gives remote access to *that IM session only*. The remote user does not need to access anything above and beyond the IM software. That software may be augmented to provide additional, technical details about the session, so that the other person can be located. The plugin will not be a backdoor; it is even possible to release it as open source, and you can compile one yourself. After all, it's a very simple piece of code. I don't know how many LOCs a plugin wrapper may need, but the TCP session to a certain host won't take more than 100 lines. When the remote session is open the incoming data is sent there too, and the remote input is injected into the IM's outgoing stream. Or you may block the local I/O entirely, if you don't want the child to see what the police is talking about.
Huh?
s/migrate/mitigate/g, as it seems...
The point is the cops told him to do something and he didn't do it...
The order must be lawful. In this case I believe the order to ask the crowd to disperse was lawful, and it doesn't matter if the message is shouted through the bullhorn or sent through Twitter.
The police could themselves tell the crowd to go home, but from every POV it is better if the organizers do that - they have better contact with the crowd. That's why he was approached and asked to cooperate.
her camera takes photos that are ~ 4.5MB each
Nobody should foist 4.5 MB images onto anyone, this is not a format for distribution regardless of how you do it. The problem needs to be fixed where it originates: you need to configure her camera to default to a more reasonable format. I can understand taking raw images if you are a professional, but then you'd be importing them right into Photoshop and you'd know what to do with them.
But if you or your mother are not professionals that need every pixel then you should change the settings on the camera. You should end up with a JPEG that is anywhere from 200 to 300 kB, and you can't see any loss of quality in that. Also the flash card in the camera will hold more of those.
It would seem that there might be a distinction, and I'm curious where people draw the line.
I think the line is clearly visible:
Case A (a CD): anyone can have a copy for $10-20. The IP owner's business is not in trouble because you bought (or even stole) a music CD. At worst they lost those $10-20 that you didn't pay.
Case B (an IP): nobody can have a copy for any reasonable amount of money. The IP owner's business may be in trouble because the data was stolen, and many people may lose jobs if the business folds.
So to tell those cases apart you only need to ask if the data is ultimately intended for public use or not.
A lot of people still try to share photos by email.
Are they somehow failing at that, or what? Where is the problem?
If you understand the risks of what happens on Facebook well enough to make an informed decision to put your stuff up there, you probably understand it well enough to throw up a quick web server.
Yes. But that means that 99% of people will use Facebook and 1% of people will NOT use Facebook. Out of that 1% maybe 10% will want to share a file using Unity. So Opera apparently made a business decision to go for the 0.1% of the market. I wish them luck, they will need it.
Considering that most Americans are in the driver's seat for their own retirement these days, that should be basically everyone.
Among my acquaintances and myself, very few actively manage their 401k. In practice it is easier and more profitable to buy a few mutual funds. Many 401k's, like Fidelity, are not set up for active trading; for example, your order will be fulfilled within 24 hours. This is because 401k is intended for long term investments. If you do switch securities, or change your allocations, it is done after careful research online - not based on a random sample from a random issue of WSJ. Mutual funds also offer greater stability (relative to the market) because they buy large number of different securities and they pay attention to what happens to those securities. It's a full time job.
But of course if you want to trade actively then the same Fidelity offers you an account for that too, and it has all the necessary features for trading (such as conditions for the sale.) Whether or not an active investor needs WSJ is a matter of opinion, but a mere 401k contributor most definitely don't need it, IMO. The 401k manager most likely limits his investment choices to a handful of mutual funds anyway. You only can pick the fund, one or many, that buys what you are interested in. That's how things are done at Fidelity 401k at least.
Says the guy with the billion dollar portfolio...
Could be. Perhaps not his own $1B, but how do you know that mspohr is not involved with managing a mutual fund, for example? Who else but an investor would be reading WSJ for any reason?
So why the fuck would anyone want to be on facebook?
People who can think more than 5 minutes ahead don't want to be on facebook, and they aren't. However the majority of people only care of "now", even if that much. They post a message or a photo because they can't be bothered to think what it can cause in the future.
To make things worse, it generally hurts your privacy to have a social life these days. Even if *you* didn't take the photo and didn't post it for all to see, someone else - out of tens of random people you met - can do so and attach your name to the photo, all that without you knowing. And that's how the data trail grows.
Even on /., formerly a bastion of near-paranoid sysadmins, posts appeared (a few years ago, IIRC) saying "the war for privacy is lost, so abandon all hope, drop all defenses and have fun!" And there is indeed a serious push to abandon all hope because it takes more and more effort with every passing day to keep your privacy, and it certainly puts constraints on how you deal with your friends and how you participate in social events. I do not expect majority of the population, especially teenagers, to sacrifice that much at cost of what they value most [at the moment] - their friends and their social life.
What are they going to do next? Tax preschool?
They are going to tax any group that can't mount an organized defence against such a tax.
Seriously though, how complete and accurate a picture of you and your activities could I create with complete and total access to your [*] ?
You can create a fairly decent picture of a law-abiding person. However a terrorist would present the same profile to you, and you can't tell the difference. He'd be paying cash for his nefarious goods; he will be walking or taking a taxi to meetings of conspirators; he would not use his home computer to contact his terorist superiors, or he'd use a method that is untraceable. If he must use a webmail he will pick a foreign one, not Google. He may even drive to a random residential neighborhood, connect to a non-encrypted AP and send his message, then leave - all within minutes.
This means that a total surveillance society is not any safer than a zero surveillance society. Criminals simply need to work around the surveillance that is there; it may take more work to do it, but they are patient: "The guard may forget that he is watching a prisoner, but the prisoner cannot forget that he is being watched."
In the U.S. at least, using genetic information to deny insurance coverage is illegal.
But it is incredibly profitable, and so it will be done anyway. The insurance company only needs to cite some other factors while rejecting the application.
A business can already, for example, refuse to hire black people (which is illegal) by simply not explaining why they were rejected. And if there is a lawsuit the business owner can always say "it's because some other applicant did better at the interview" and that's that. It would take a smoking gun, like an HR memo spelling out the "no blacks" policy, to lose the court case.
anyone here know where I could get one (or at least, a 30x30x7 (mm))?
Don't know about x7, but here is 30mm L x 30mm H x 6mm W fan. This is a 5V part without tachometer. There is also 259-1327-ND which produces higher airflow (and is noisier, I'd guess.)