A study by... erm... PBS, I think, found that middle school aged children trust advertising more than news for exactly that reason. The advertisers have a clear agenda, so you can correct for that, whilst double talking adults (at least that's my recollection) had so many agendas that it was impossible to separate spin from fact. I suspect that's why so few people watch the news or read newspapers these days
It is misleading to say that this is a case about content. While the kid did sue about content, it was not because it was offensive, but rather because it was posted under his name.
We all (I hope) agree that my ISP isn't resposible if I call you a lazy ass bastard in an email (not that I ever would mind you:-P), I think the kid has a very good point that it is my ISPs responsibilty to make sure that I'm not able to open an account in your name.
I think it does border on negligence not to verify the identity of the person opening an account -- at least if we assume that all accounts need identities (I'd love to have a totally anonymous one, but I doubt it would be feasible -- so few ISPs take cash).
So saying this is a case about content isn't wrong, but not 100 % correct either.
talking out of my ass (a great trick at parties) isn't this a different type of noise? Ok, so we see oscillations in the digital pulses, but since we're in the digital world, that's ok, no? I mean as long as the noise is below the theshold of the signal it does no harm?
The only analog part of the entire design is chip outs 5 and 7 from the DAC and those go directly (well, via some resistors, but still fairly directly) to the headphones. (NB: volume is apparently performed on the digital data stream, and not as a post processing step by the DAC -- although I could be misinterpreting the pins on the dac)
Mind you, he did mention that he had high THD, which might be due using the veroboard.
Or have I misunderstood everything (IANAEE (electrical engineer)).
And of course we all know that apple computer was severely limited in what sort of musical abilities it was able to equip its computers with because of this.
IIRC, when the macs were equiped with real sound (you had to buy a card to get anything but bleeps out of the Apple II -- just like a PC) they had to get apple record's permission.
How this pertains to secure communication is that by the sender sending his conclusions (in the clear!) about the prior state of the particle to the receiver, she can deduce the secret message he sent in.
Now the trick is that if a man in the middle is trying to spoof this, it will be impossible for him to relay the message forward to the receiver. (he cannot manufacture particles wiht the desired properties to send to the recipient)
So a simple authentication phase afterwards will identify any eaves dropping.
The problem is that you need to know what was done to the "sending" particle in order to decode the "receiving" one. Also, if you look at the reciever too soon, then it becomes the sender.
Basically, it is like XORing with an unknown bit. The sender knows what he sent in, so looking at the result, he can deduce the original state of the random bit.
The spooky part is that the corresponding random bit on the other end changes instantaneaously. unfortunately, the receiver knows only the result of the XOR, and this is not enough to send a message. She also needs the information the sender deduced about the unkown state to decode the information. NB She has to perform an XOR as well to read the information, so if she tries to read too soon, she'll have sent rather than received.
cunningly mislabeled "volume"? I had a printer driver like that for my old mac. It had a "do not fuck up" checkbox that had to be selected, but it was labeled something weird, like "download fonts once", or something equally hard to guess.
Typically, art has value outside its direct function (f.ex. haute couture), while commodities have no identity, appart from the quality it was bought for (f.ex jeans -- any brand of denim is pretty much the same).
what you are willing to pay for it is rather irrelevant.
So songs by metallica might be art to a fan, but since I'm not, it's a commodity -- that commodity being noise generation.
And regardless of the legal realities, it's a sleazy thing to do.
I don't know that I agree. (but then I'm biased)
The whole point (as I had understood it) of free software is that I be allowed to use it as I see fit. Now as the customer in the above scenario, I'd be very happy that I be able to improve the software I bought by including readline. I think that is perfectly ethical.
Is it unethical for me to copy a good but undistributable implemenation's API to give customers the freedom to improve the application? I can't see that it is.
Wasn't this what the entire Motif/Lesstif thing was about? Surely lesstif isn't sleazy?
I love libreadline, but it is GPLed. Say I want to write a non GPLed program, but one which will be released in some linkable form.
Now for the gedanken question: Is it legal for me to write some crappy readline replacement -- one that provides the same API -- and provide that along with my program, and tell the buyer that yes I know my line reader sucks, and if they want to, they can just relink to readline?
My program doesn't link against readline, so I'm not distributing a derived work. Any changes that the purchaser makes are not my responsability.
The velocity of money is pretty high; I expect that unless every merchant cooperated with you, the std 6-degrees of separation argument would imply that the scanner of money could infer very little.
For example, the bank might know that I withdrew this money a week ago, and now it was deposited in D.C. What does that tell them? Only that there is a flow from Boston to D.C. It tells them nothing of value about me (other than that I was at the head of that flow, ok so maybe somewhat interesting) -- but maybe something very interesting about the flows of money.
Y'know, I'd like to see this implemented. It wouldn't be too hard, and the flow graph would make a great poster.
Does anyone know of any totally passive cooling systems? Maybe a convective cooling system, f.ex, which heats up some freon-substitue and carries the head up to the top of the case where there is a bin fin-encrusted cooling area?
I've decided that silence is really key if I'm going to be able to run my server 24/7/365. (or is that 24/7/52 ?)
The executable/non-executable test is misleading, as it disregards interpreters.
I remember reading about an interpreter for C, and I know there is one for java, and I know I've written a fair number of interpreters for sub-scheme.
Source code is the same as the executable.
So let's gedanken. Most people would claim that a clearly written paper describing an algorithm is free speach. Currently natural language processing isn't up to reading and executing that algorithm, but in a few decades it might very well be. So then the paper becomes an executable as well.
Has anyone picked up the Ghost Dog soundtrack? Apparently all the songs are by the RZA (from the Wu-Tang, I think). I saw the movie, and was really impressed with the background song that plays whilst the dove flies ("this is what it sounds like/ when doves fly":-b) The rest I didn't notice so much, so I'm angling for some feedback from someone who's chumped up the change for it already.
metageeks and geek wannabes in one post. Now that is priceless. Not that I disagree -- quite the opposite -- but those two concepts in one post made me just smile.
Interesting that both bilingualism (in Canadia) and firearms (in the States) are issues that are heatedly debated in that country, legally guaranteed in that country, and non-issues in the other country.
Interesting also that sports fans here in the bloodthirsty (supposedly) USA are lambs compared to those of (again supposedly) more civilised europe.
Not that I think we can draw any conclusions from these points, but they fascinate me, at least
I'm not thrilled about adding all that hard-to-debug software to get only 20% better performance on SPECmark benchmarks.
R U shittin' me? 20% is a kick ass SPEC improvement, esp since it's done totally in software. You can bet that there will be some sort of hardware support soon. Like the article starts out stating, the compiler team get hugeass bonusses based on piddly decimal point improvements (I might be overstating the case somewhat, tho). 20% is like, well, huge. The compiler guys didn't believe the numbers at first.
R U shittin' me mk II? compared to the black voodoo magic of some compiler transformations, this transformation is almost clear as crystal. And it runs in REAL TIME. These days, O(n^3) worst case is almost acceptable for static optimsation algs.
But you are right about your other points, sorta. While hint bits are a cheapass version of this idea, it appears that the real gains come when you start completely removing the branches altogether and propagate assumptions forward (redoing register allocation, that sort of thing).
Even more funner would be to cross kernel boundaries and inline kernel calls. Transmeta can do this, as their optimiser run at the meta level, whilst dynamo cannot, as it is merely a user level program.
But that's perfectly ok, tho, i'n'it? Since the api to the secret stuff is fairly well defined, you just need to recreate it. By overspeccing the HD cache, you should be able to hide the fact that you don't have that good of a disk-placement alg.
Actually, now that I think about it, you should get fairly good performance from just setting your HD block size to something a bit larger -- say about a 1 MB. That's at least a good fraction of a minute, which should be enough to hide seek times.
Slashdot Mirror
A study by ... erm... PBS, I think, found that middle school aged children trust advertising more than news for exactly that reason. The advertisers have a clear agenda, so you can correct for that, whilst double talking adults (at least that's my recollection) had so many agendas that it was impossible to separate spin from fact. I suspect that's why so few people watch the news or read newspapers these days
Johan
damn slashes in tags -- I can never remember the direction they go..
Actually, not really.
:-P), I think the kid has a very good point that it is my ISPs responsibilty to make sure that I'm not able to open an account in your name.
It is misleading to say that this is a case about
content. While the kid did sue about content, it was not because it was offensive, but rather because it was posted under his name.
We all (I hope) agree that my ISP isn't resposible if I call you a lazy ass bastard in an email (not that I ever would mind you
I think it does border on negligence not to verify the identity of the person opening an account -- at least if we assume that all accounts need identities (I'd love to have a totally anonymous one, but I doubt it would be feasible -- so few ISPs take cash).
So saying this is a case about content isn't wrong, but not 100 % correct either.
talking out of my ass (a great trick at parties) isn't this a different type of noise? Ok, so we see oscillations in the digital pulses, but since we're in the digital world, that's ok, no? I mean as long as the noise is below the theshold of the signal it does no harm?
The only analog part of the entire design is chip outs 5 and 7 from the DAC and those go directly (well, via some resistors, but still fairly directly) to the headphones. (NB: volume is apparently performed on the digital data stream, and not as a post processing step by the DAC -- although I could be misinterpreting the pins on the dac)
Mind you, he did mention that he had high THD, which might be due using the veroboard.
Or have I misunderstood everything (IANAEE (electrical engineer)).
And of course we all know that apple computer was severely limited in what sort of musical abilities it was able to equip its computers with because of this.
IIRC, when the macs were equiped with real sound (you had to buy a card to get anything but bleeps out of the Apple II -- just like a PC) they had to get apple record's permission.
Sorry, I hit send too soon.
How this pertains to secure communication is that by the sender sending his conclusions (in the clear!) about the prior state of the particle to the receiver, she can deduce the secret message he sent in.
Now the trick is that if a man in the middle is trying to spoof this, it will be impossible for him to relay the message forward to the receiver. (he cannot manufacture particles wiht the desired properties to send to the recipient)
So a simple authentication phase afterwards will identify any eaves dropping.
"entanglement" creates the paradox of FTL
I can't think of any.
The problem is that you need to know what was done to the "sending" particle in order to decode the "receiving" one. Also, if you look at the reciever too soon, then it becomes the sender.
Basically, it is like XORing with an unknown bit. The sender knows what he sent in, so looking at the result, he can deduce the original state of the random bit.
The spooky part is that the corresponding random bit on the other end changes instantaneaously. unfortunately, the receiver knows only the result of the XOR, and this is not enough to send a message. She also needs the information the sender deduced about the unkown state to decode the information. NB She has to perform an XOR as well to read the information, so if she tries to read too soon, she'll have sent rather than received.
Johan
there is an ``inverse realism slider''
cunningly mislabeled "volume"? I had a printer driver like that for my old mac. It had a "do not fuck up" checkbox that had to be selected, but it was labeled something weird, like "download fonts once", or something equally hard to guess.
Typically, art has value outside its direct function (f.ex. haute couture), while commodities have no identity, appart from the quality it was bought for (f.ex jeans -- any brand of denim is pretty much the same).
what you are willing to pay for it is rather irrelevant.
So songs by metallica might be art to a fan, but since I'm not, it's a commodity -- that commodity being noise generation.
So LCD power use is a function of area, not resolution?
Cool!
Hey laptop guys! Make mine a strongarm w/ a 1280x1024 12" display. And make it light and thin too. I'll even pay through the nose for the pleasure.
I don't know that I agree. (but then I'm biased)
The whole point (as I had understood it) of free software is that I be allowed to use it as I see fit. Now as the customer in the above scenario, I'd be very happy that I be able to improve the software I bought by including readline. I think that is perfectly ethical.
Is it unethical for me to copy a good but undistributable implemenation's API to give customers the freedom to improve the application?
I can't see that it is.
Wasn't this what the entire Motif/Lesstif thing was about? Surely lesstif isn't sleazy?
Johan
To make this concrete:
I love libreadline, but it is GPLed. Say I want to write a non GPLed program, but one which will be released in some linkable form.
Now for the gedanken question:
Is it legal for me to write some crappy readline replacement -- one that provides the same API -- and provide that along with my program, and tell the buyer that yes I know my line reader sucks, and if they want to, they can just relink to readline?
My program doesn't link against readline, so I'm not distributing a derived work. Any changes that the purchaser makes are not my responsability.
Eh?
Johan
Hrm.
ISPs that carry NNTP feeds are normally held not to be responsible for the contents of those newsgroups, since they don't apply any editorial control.
I imagine that a freenet node would be the same thing.
On that topic, how exactly is freenet different from a pull version of NNTP?
Idunno
The velocity of money is pretty high; I expect that unless every merchant cooperated with you, the std 6-degrees of separation argument would imply that the scanner of money could infer very little.
For example, the bank might know that I withdrew this money a week ago, and now it was deposited in D.C. What does that tell them? Only that there is a flow from Boston to D.C. It tells them nothing of value about me (other than that I was at the head of that flow, ok so maybe somewhat interesting) -- but maybe something very interesting about the flows of money.
Y'know, I'd like to see this implemented. It wouldn't be too hard, and the flow graph would make a great poster.
Johan
Does anyone know of any totally passive cooling systems? Maybe a convective cooling system, f.ex, which heats up some freon-substitue and carries the head up to the top of the case where there is a bin fin-encrusted cooling area?
I've decided that silence is really key if I'm going to be able to run my server 24/7/365. (or is that 24/7/52 ?)
Johan
Cash can be scanned, sure. But it is still anonymous -- you can trace the bill, but you have no idea who gave it to you.
The executable/non-executable test is misleading, as it disregards interpreters.
I remember reading about an interpreter for C, and I know there is one for java, and I know I've written a fair number of interpreters for sub-scheme.
Source code is the same as the executable.
So let's gedanken. Most people would claim that a clearly written paper describing an algorithm is free speach. Currently natural language processing isn't up to reading and executing that algorithm, but in a few decades it might very well be. So then the paper becomes an executable as well.
Oh bother.
Has anyone picked up the Ghost Dog soundtrack? Apparently all the songs are by the RZA (from the Wu-Tang, I think). I saw the movie, and was really impressed with the background song that plays whilst the dove flies ("this is what it sounds like/ when doves fly" :-b) The rest I didn't notice so much, so I'm angling for some feedback from someone who's chumped up the change for it already.
metageeks and geek wannabes in one post. Now that is priceless. Not that I disagree -- quite the opposite -- but those two concepts in one post made me just smile.
I dunno why.
The boston herald is usually considered to have journalism slightly better than the sun (of "page 3 girl" fame), but not by a significant margin.
Judging from past front pages, they specialise in "man bites dog" and "hardworking family done wrong by state" stories.
You'd do better to read the weekly world news, which is at least is a known quantity.
I understand that it (the herald, that is) has an extensive (and by what I overhear on the T, fairly good) sports section, though.
Interesting that both bilingualism (in Canadia) and firearms (in the States) are issues that are heatedly debated in that country, legally guaranteed in that country, and non-issues in the other country.
Interesting also that sports fans here in the bloodthirsty (supposedly) USA are lambs compared to those of (again supposedly) more civilised europe.
Not that I think we can draw any conclusions from these points, but they fascinate me, at least
also, they won't serve the page unless you accept cookies. grr.
R U shittin' me? 20% is a kick ass SPEC improvement, esp since it's done totally in software. You can bet that there will be some sort of hardware support soon. Like the article starts out stating, the compiler team get hugeass bonusses based on piddly decimal point improvements (I might be overstating the case somewhat, tho). 20% is like, well, huge. The compiler guys didn't believe the numbers at first.
R U shittin' me mk II? compared to the black voodoo magic of some compiler transformations, this transformation is almost clear as crystal. And it runs in REAL TIME. These days, O(n^3) worst case is almost acceptable for static optimsation algs.
But you are right about your other points, sorta. While hint bits are a cheapass version of this idea, it appears that the real gains come when you start completely removing the branches altogether and propagate assumptions forward (redoing register allocation, that sort of thing).
Even more funner would be to cross kernel boundaries and inline kernel calls. Transmeta can do this, as their optimiser run at the meta level, whilst dynamo cannot, as it is merely a user level program.
As a previous poster has already pointed out,
It's a question of mean vs. median.
Imagine the following 5 scores (not necessarily plausible numbers):
70 80 90 100 160
if my off-the-cuff math isn't wrong, the median of those scores is 90, but the mean is 100.
So 4/5 of the people you'd meet are at or below average intelligence. (!)
But that's perfectly ok, tho, i'n'it? Since the api to the secret stuff is fairly well defined, you just need to recreate it. By overspeccing the HD cache, you should be able to hide the fact that you don't have that good of a disk-placement alg.
Actually, now that I think about it, you should get fairly good performance from just setting your HD block size to something a bit larger -- say about a 1 MB. That's at least a good fraction of a minute, which should be enough to hide seek times.
johan