Ironically, people investing their own money to solve their problems (because there is no one else who offers a cheaper solution) would not sound too much out of place in an Ayn Rand novel...
Well I guess the prior information that there was something that looked like a face under the blurred blob could be used for some kind of statisitcal estimation of how the face might have looked. I think that much less information is needed to extract biometric features from a blurred picture, than to do real un-blurring. Has something like that been used in practice?
It could work somehow like the 'de-pixelization' of license plates where AFAIK the most likely licenseplate number is estimated instead of doing real (impossible) de-pixelization.
Or to put it that way: transistor power amp distortion will just sound as if the amp is broken, while a tube power amp distortion sounds like ***put in your favourite guitarist who was not member of panthera***
Eventually, Google's employees will be as over-managed as most other employees at most other software companies.
Is that a natural law or just the result of bad management?
Valid point, but slightly overrated in my opinion.
I just wrote some fairly low level scientific/numerical stuff in java (tons of array access mixed with floating point calculations, plus some preprocessing stuff with large bit-vectors (class BitVector from the colt library)) that is just as fast as the Fortran/C code it replaced. Of course this is only true for the newest java VM on fairly new intel/amd hardware. Also the algorithm seems to be relatively easy to optimize (drop array access checks etc.).
On combination of IBM VM/Power4 hardware on the other hand the performance was, uhm, suboptimal to put it that way. (though I have not tested the Fortran/C version on that hardware)
Why do I have the feeling that c and c-derived languages like c++, c# and java (maybe not really c-derived, but generally real 'non scripting' languages that somehow look and work similarily to c (but that would have been too long)) will still be around after "the Web" is long dead and gone?
The web has gone through immense changes in the last 10 years, while the basic technologies that keep the web running have stayed remarkably stable. Windows has had some gui changes, UNIX is now called Linux, those fancy new web-servers even run scripts to generate HTML code. But still it is the same old c-ish technology. I don't see how this will become irrelevant even when "the Web" takes over even more...
Banning sex with stupid individuals (they act like kids). You'll have to be this smart to have sex. Loop up "dumm fickt gut" and you will know the common german's oppinion on this one.
I didn't have to pay google for a product (+free advertising) I didn't need, when I bought my laptop. That's the difference. They can advertise for anything they like on their page, as long as no one is forced to use (as in 'pay for') it. End of the story.
Or to introduce a third option: it depends on the position of the change in the gene. The effect can go from completely different to absolutely the same (and I'm only talking about protein coding regions) and anything in between...
It could also be the case that dslr owners (all cameras on the list but the powershot and cybershot are dslrs) upload their photos more often without post processing. I'm not sure how many programs keep the EXIF info (I don't think gimp saves the EXIF info from the original photo). I own a nikon d50 and generally the photos straight out of the cam (unlike the ones from my sony cybershot) look fine without postprocessing.
Right. Keeping track of all the knowledge in natural science is as hard an as important as the science itself. It is no wonder we have groups working on datamining at the bioinformatics department at my university.
I've never seen a LCD with faded backlight. But I'm sure that by the time the public becomes aware of the problem, the LCD manufacturers will have LED backlights up and running (they will switch to LEDs for better color reproduction sooner or later anyhow). I'm not so sure if the main plasma problems, namely burn-in and high price, will be solved as easily to take away LCD's marketshare.
My point is that from the way you quoted the programmer one could get the impression that he said something like "differences imply manipulation". I just heared him say that the exit polls should not be significantly different (which is not very specific, I agree. But at least it is more qualified than a simple implication).
He said that exit polling data should not be significantly different. There is a whole discipline in mathematics called 'statistics' that deal quite a lot with this thing called 'significance'. You can calculate how big the difference can get until it gets extremely unlikely that it occurred by chance.
That makes it even worse. Software like that has been around for years. It should be able to detect when it is overfitting during training. At least I would expect this from a new and especially user friendly program. Most of its users won't even know about special machine learning problems like overfitting.
... who thinks that saving 100m$ over five years sounds rather i ridiculous for a (supposedly) large bank? I can not even imagine that it is worth the bad publicity they get through this. If I were an American I would be extremely pissed off by this Bank of 'America'. What will they do with their next 100m$? Some TV commercials?
Exactly what I was thinking, when I saw the screenshots. Looks like kde from 3 years ago when the (now) uber-ugly keramik theme was state of the art. At the first look it was somehow stunning, but today... that kind of super glossy gui has not aged gracefully. Now it just looks dated.
They actually managed to introduce this bug in two different locations? The programmer even changed the '!=' to a '==' after he copied the code without noticing the missing parentheses. Another reason not to use cut/paste for security relevant code...
Why not NT4? That's a good question given that the win32 api has not fundamentally changed in the last 10 years. So what are those components in XP that w2k has not (and that could be useful in a web browser)?
Well, I said this in the past and I have to say it again: let google do the math before you post any calculations on slashdot:
(100 - 99.9)% of (1 day) = 1.44 minutes
Slashdot Mirror
so windows server 2003, sql serevr 2005 are crappy software?
yes
Ironically, people investing their own money to solve their problems (because there is no one else who offers a cheaper solution) would not sound too much out of place in an Ayn Rand novel ...
Well I guess the prior information that there was something that looked like a face under the blurred blob could be used for some kind of statisitcal estimation of how the face might have looked. I think that much less information is needed to extract biometric features from a blurred picture, than to do real un-blurring. Has something like that been used in practice?
It could work somehow like the 'de-pixelization' of license plates where AFAIK the most likely licenseplate number is estimated instead of doing real (impossible) de-pixelization.
Or to put it that way: transistor power amp distortion will just sound as if the amp is broken, while a tube power amp distortion sounds like ***put in your favourite guitarist who was not member of panthera***
Is that a natural law or just the result of bad management?
Valid point, but slightly overrated in my opinion.
I just wrote some fairly low level scientific/numerical stuff in java (tons of array access mixed with floating point calculations, plus some preprocessing stuff with large bit-vectors (class BitVector from the colt library)) that is just as fast as the Fortran/C code it replaced. Of course this is only true for the newest java VM on fairly new intel/amd hardware. Also the algorithm seems to be relatively easy to optimize (drop array access checks etc.).
On combination of IBM VM/Power4 hardware on the other hand the performance was, uhm, suboptimal to put it that way. (though I have not tested the Fortran/C version on that hardware)
Why do I have the feeling that c and c-derived languages like c++, c# and java (maybe not really c-derived, but generally real 'non scripting' languages that somehow look and work similarily to c (but that would have been too long)) will still be around after "the Web" is long dead and gone?
...
The web has gone through immense changes in the last 10 years, while the basic technologies that keep the web running have stayed remarkably stable. Windows has had some gui changes, UNIX is now called Linux, those fancy new web-servers even run scripts to generate HTML code. But still it is the same old c-ish technology. I don't see how this will become irrelevant even when "the Web" takes over even more
and your evidence is from msdn, Microsofts very own propaganda outlet. Now that's what I call convincing...
I didn't have to pay google for a product (+free advertising) I didn't need, when I bought my laptop. That's the difference. They can advertise for anything they like on their page, as long as no one is forced to use (as in 'pay for') it.
End of the story.
Or to introduce a third option: it depends on the position of the change in the gene. The effect can go from completely different to absolutely the same (and I'm only talking about protein coding regions) and anything in between...
The parent was bashing Europe in general. So it's only fair to use the Netherlands as a counterexample.
It could also be the case that dslr owners (all cameras on the list but the powershot and cybershot are dslrs) upload their photos more often without post processing. I'm not sure how many programs keep the EXIF info (I don't think gimp saves the EXIF info from the original photo). I own a nikon d50 and generally the photos straight out of the cam (unlike the ones from my sony cybershot) look fine without postprocessing.
Right. Keeping track of all the knowledge in natural science is as hard an as important as the science itself.
It is no wonder we have groups working on datamining at the bioinformatics department at my university.
I've never seen a LCD with faded backlight. But I'm sure that by the time the public becomes aware of the problem, the LCD manufacturers will have LED backlights up and running (they will switch to LEDs for better color reproduction sooner or later anyhow). I'm not so sure if the main plasma problems, namely burn-in and high price, will be solved as easily to take away LCD's marketshare.
My point is that from the way you quoted the programmer one could get the impression that he said something like "differences imply manipulation". I just heared him say that the exit polls should not be significantly different (which is not very specific, I agree. But at least it is more qualified than a simple implication).
He said that exit polling data should not be significantly different. There is a whole discipline in mathematics called 'statistics' that deal quite a lot with this thing called 'significance'. You can calculate how big the difference can get until it gets extremely unlikely that it occurred by chance.
That makes it even worse. Software like that has been around for years. It should be able to detect when it is overfitting during training. At least I would expect this from a new and especially user friendly program. Most of its users won't even know about special machine learning problems like overfitting.
... who thinks that saving 100m$ over five years sounds rather i ridiculous for a (supposedly) large bank? I can not even imagine that it is worth the bad publicity they get through this.
If I were an American I would be extremely pissed off by this Bank of 'America'. What will they do with their next 100m$? Some TV commercials?
Exactly what I was thinking, when I saw the screenshots. Looks like kde from 3 years ago when the (now) uber-ugly keramik theme was state of the art. At the first look it was somehow stunning, but today ...
that kind of super glossy gui has not aged gracefully. Now it just looks dated.
They actually managed to introduce this bug in two different locations? The programmer even changed the '!=' to a '==' after he copied the code without noticing the missing parentheses. Another reason not to use cut/paste for security relevant code...
Why not NT4? That's a good question given that the win32 api has not fundamentally changed in the last 10 years. So what are those components in XP that w2k has not (and that could be useful in a web browser)?
there seems to be a little mistake in your calculations:
Energy to melt 1 mt of steel: 377kWh = 377 * 3.6Mjoules = 1.4 Gjoules
1.4 Gjoules per mt / 131 Mjoules per gallon = 10.4 gallons / mt
I think you calculated kWh per gallon: 131 / 3.6 = 36 kWh, not mt of steel per gallon.
they are running the webserver as a pixel-shader on one of their geforcen.
Well, I said this in the past and I have to say it again: let google do the math before you post any calculations on slashdot:
(100 - 99.9)% of (1 day) = 1.44 minutes