I understand your opinion, but I don't agree with you forcing it on other people. Cultural standards for acceptable language vary the world over. In many places the OP's language would be considered quite civil. Please don't confuse a global forum like slashdot with a quiet local meeting - people vary a lot. The issue that I took with you was that by ignoring his actual point, focusing on his language rather than what he said, you added to the noise in this place, and detracted from the signal. In some sense that my own personal opinion that I'm forcing on you, but I judged that your actions had decreased the overall utility of the site.
That's interesting. I stand (partially) corrected. Although it's not really quite the same thing, codes that input confidence estimates of the bits sound a little similar to LPN (Learning Parity with Noise) which is currently being explored as a hard problem for use in cryptographic protocols such as HB+. The underlying general problem (sorry I'm too lazy to Google and the relevant proceedings are not to hand) is believed to be NP-hard, although I don't think a reduction for LPN has been proven.
Of course if there is a systematic bias in the errors then it becomes a lot easier...
I wouldn't be so sure that he was. He tackled the major flaws in the marketing material quite well.
There is no such thing as a low resolution bit. Even with ECC the resolution of a bit is still 2. Think about that for a while - I am not aware of a single error code that allows you to drop the number of discrete states lower than two and can still reconstruct the data.
Low quality is a little less well defined. In the analogy that the AC was busting apart the low-quality was the actual degradation of the contrast of an analogue print. I can see where you are going with low quality as an increase in the noise floor for the bit which can be corrected by a code. But in his analogy the noise floor is being raised across the entire storage array. So for your code to be able to recover data you need a basic trade of capacity for redundancy, which is covered in detail elsewhere in the comments.
His basic point appeared to be that while ECC can tolerate a known, limited increase in noise. The "proven track-record" referred to by the company seems to involve an unknown, unlimited decrease in quality. Not really something to trust for archival purposes.
I was not comparing the iPhone to the the Surface. Seriously, reread my comment because I'm not sure how you could get it that wrong. The clue is the subtle descriptive language like table sized input device. Now you may think that the iPhone is a brick, but surely table-sized is going too far?
April 2006 predates the Surface by a long way. It was not invented by microsoft.
Gee. If you measure your social ceiling in relation to slashdot then you have your own problems. You appear to make the same cultural assumptions about the use of language on this site that the GP did. Perhaps you are incapable of recognising your lack of insight
Seriously... this is a public site so get your fucking head out of your fucking ass and quit complaining about the gp's language. Retarded is a very mild form of abuse in europe, and in comparison to the US market it is a perfectly accurate assessment. What on earth makes your think that he should aim his language at your particular definition of clean?
Some interesting comments. I've used linux for years but recently switched to a mac. Lots of reasons provoked it but the main issue was that I wanted laptop with decent battery life and powersaving that ran some variant of unix. After looking around the macbook-pro worked out cheaper than any similar spec machine (mainly because I get a hefty academic discount and finding something as pokey as the nvidia mobile was difficult).
From the UI perspective I loved it instantly. Some things are just designed right from the ground up, after fighting modal dialog boxes in windows/linux for years I find nonmodal preferences etc a breath of fresh air. Other things confused the crap out of me - the idea that the mac is magically more "usuable" is a nice ideal, but it is as hit and miss as other platforms. Single application instances work for some type of applications, but for other applications (or just processes) I do want to spawn multiple copies (i.e mplayer).
You can get the virtual filesystems that you want as macfuse works pretty well and I use it to mount remote ssh filesystems. Given the focus of the article I've got to say that multitouch scrolling is the best UI improvement that I've seen since the scrollwheel. It is so natural and intuitive. I didn't get the refresh of the macbook-pro with the other multi-touch gestures but I think that I got the one that I would use day in and out. Zoom looks useful but a little gimmicky.
I've suprised that a USB hard drive wouldn't work. Are they eve configurable? Most just expose a FAT filesystem through a standard interface when you plug them in. They even work on my linux desktop at work...:) And as for the idea that everthing just magically works... it only took two system updates for Apple to get the bloody keyboard working properly under Leopard. Even mickysoft can do better than that...
It's strange that you say that because the first multi-touch demo that I remember seeing was a table sized input device that they used to control C&C and google earth. It strikes me as strange because it is the exact same technology that Microsoft have cloned and claimed to have invented with the Surface... Small world when you're competing with Microsoft I guess.
No that's not true at all. I think you've missed the gp's point (ironically enough). Science is not a competition between scientists - although it looks that way from a funding perspective. Science is a competition between the Scientist and the universe. If there is a way to gain an advantage over the universe, and it doesn't have its own issues of morality then it is the correct thing to do.
The stupid hours that we work (I write this at 3am during my second 18hr stint in the lab this week), the coffee that we drink (mmmm tasty) and anything else we do to try and attain an advantage - is fair and within the game that we are playing.
Now the question of whether or not the specific drugs in the article are morally acceptable for general use is a seperate issue. It's not one that I have a strong opinion on as I've never tried them myself and have no desire to do so. But if somebody else uses these substances, and produces good work from it then fair play to them.
The only way to do Science and respect yourself for doing it is to throw everything that you possibly can into it. I would would ask you how you could do science without that level of commitment and still respect yourself for it?
The problem with Blogs is that they are inevitabley the whining and yapping of dogs that don't know any better. The worthless opinion that you link to fails to explain the original experiment correctly before weighing in. While it doesn't add anything of value, I guess it lets you slur the reputation of Dr Chen which is what you apparently wanted to do.
Chen didn't try to prove that the experiment was definitely flawed - he showed that their own reasoning for why it was correct was not valid. That is there were no conclusions that could be derived from the experiment because their methodology was incorrect.
It's actually quite a nice point that he's made, although some other poster further up pointed out that there are much deeper flaws in the original experiment such as the complete lack of a decent control. If they'd rerun it with an animal that had monochrome vision and they still got a 2/3s result then they would have known that something was amiss.
As far as Chen's point goes, I suspect most, if not all animals with chromatic vision have a subtle ranking, because we have evolved to look for certain foods. This is suspected to be why humans can distinguish red/green combinations much better than others.
Thanks for the reply, that was very informative. The section about what is wrong with NAT that firewalls fix is a bit loose - the points about open ports and machines in the DMZ apply to firewalls equally as much as NATs. The remote admin and pings are red herrings - most NATs that I've seen in the past few years switch these off by default now.
The points about where firewalls are more secure products were really interesting. I hadn't thought of the sequence number bouncing, and yet it's a basic part of many crypto protocols to stop replay attacks. The outbound filtering matches up well with the other reply about compromised machines.
I've heard this claim before that NAT can be overcome (can't remember where) and it is an interesting point. In the typical installation - private class C behind the NAT. Port forwarding switched off. How exactly would you go about overcoming the firewall?
I can see that a carefully crafted packed could fool the system into relaying it - although if the router has been designed properly the PPP to the outside world and the hub behind should be physically separate networks. But how would you fool it into masquerading the connection to get the other half of the socket connected?
I ask out of genuine curiosity as it is a claim that I've come across but not worked out how it could be done.
Did you really go to all the trouble of misunderstanding him just to flog your point?
At no point did the GP claim that a hardware firewall and a NAT router were the same thing. He pointed out that he doesn't need a (software) firewall on his box when it's sitting behind NAT. You know exactly what he meant, and why in the context of crappy windows security he's correct - but you tried desparately to show that you knew more about the subject than he does.
Reread (exactly) what he said. Understand it. Shut the fuck up.
Annoying though it is my bank worked around this awhile ago. Instead of entering my PIN through the keyboard they flash up a java keyboard with randomised key layout on the screen which I have to click with the mouse. It is more annoying than tapping in the code as it takes effort to read the screen and translate my PIN onto it, but it must save quite a few of their customers from keyloggers. If it becomes popular amongst other banks then expect a similar arms race to the one underway between CAPTCHAs and spammers.
Although the OP has already said above that he was quoting from memory and the number is probably wrong - it would be a neat trick. At the height of the cold war "winning" was defined as having more of your population survive than the other side. This was also the criteria that the Indians were using to claim they could "win" a war against Pakistan if Kashmir ever went hot.
Depending on how long you run the stats for it is not impossible that that percentage of the US would have been wiped out in a full-scale exchange with the old USSR. Not all of them in the initial explosions (which would have blanketed every major urban area and several non-urban but military sites) but the country would not have been able to function in the aftermath.
Ignoring the effects of a nuclear winter and just considering the raw effects of massive irratiated zones up and down the country, complete standstill in economy and transportation. Within a few years of the initial exchange those death tolls don't look quite so unreasonable. Even for the "winning side". Scary stuff indeed.
PS This new enforced preview is a bugger. I missed the date on your pop. estimate and so now I see your point. I thought you merely meant that a percentage that high was unlikey.... oh well, moving on.
True. But it does come up fifth, and number two is wikipedia which has huge pagerank for any subject. So whoever bought it has gained a very high placing for that search query without paying Google any money for a placed search result. Whether or not the money they paid for it compares with the money for paid results on that query is another question entirely.
You seem to be wrong on just about every point that you've tried to make. Take a look at the previous article on multithreading that made the slashdot frontpage (I think about a week ago). Somebody tried the same argument as you and was shot down quite forcefully in the largest thread in the discussion.
None of the people that argued could come up with a single real-world problem that couldn't be hacked into working on multi-core systems. When you say "SMT" I assume you've made a typo and mean SMP. SMP is a very general class of parallel systems and is certainly not "useless for most general computing". Name a single real application that cannot take advantage of multiple cores.
You don't seem to know the difference between SIMD and MIMD. Here's a clue: multi-core systems are MIMD, although they typically have SIMD instructions within each core.
Your game example is weak; it sounds as if your game has no internal logic. Once you add AI and some sort of world simulation there is plenty of scope for using those 16 cores.
Lastly, quoting Savain just makes you sound like a retard. The guy is a crank and has been pushing that bullshit on his webpage for years. It only shows that he doesn't understand language design or computer architecture as well as he thinks, and that he should really start taking some form of medication.
Yeah, that would be my guess as well. I'm sure that it's deliberate that they've left it vague though. Looking at who is behind it I'm sure it's not something that would have escaped them.
The video was too laggy to watch (guess it got busy), but their draft patent policy made interesting reading. It has one set of rules for open-source usage, and one set of rules for commercial usage. Are they assuming that these two sets of users are disjoint?
I wonder which set of licensing conditions / price a commercial open-source project would fall under?
Slashdot Mirror
I understand your opinion, but I don't agree with you forcing it on other people. Cultural standards for acceptable language vary the world over. In many places the OP's language would be considered quite civil. Please don't confuse a global forum like slashdot with a quiet local meeting - people vary a lot. The issue that I took with you was that by ignoring his actual point, focusing on his language rather than what he said, you added to the noise in this place, and detracted from the signal. In some sense that my own personal opinion that I'm forcing on you, but I judged that your actions had decreased the overall utility of the site.
That's interesting. I stand (partially) corrected. Although it's not really quite the same thing, codes that input confidence estimates of the bits sound a little similar to LPN (Learning Parity with Noise) which is currently being explored as a hard problem for use in cryptographic protocols such as HB+. The underlying general problem (sorry I'm too lazy to Google and the relevant proceedings are not to hand) is believed to be NP-hard, although I don't think a reduction for LPN has been proven.
Of course if there is a systematic bias in the errors then it becomes a lot easier...
Very true, I believe that Blair et al. have published a lot in this area.
I wouldn't be so sure that he was. He tackled the major flaws in the marketing material quite well.
There is no such thing as a low resolution bit. Even with ECC the resolution of a bit is still 2. Think about that for a while - I am not aware of a single error code that allows you to drop the number of discrete states lower than two and can still reconstruct the data.
Low quality is a little less well defined. In the analogy that the AC was busting apart the low-quality was the actual degradation of the contrast of an analogue print. I can see where you are going with low quality as an increase in the noise floor for the bit which can be corrected by a code. But in his analogy the noise floor is being raised across the entire storage array. So for your code to be able to recover data you need a basic trade of capacity for redundancy, which is covered in detail elsewhere in the comments.
His basic point appeared to be that while ECC can tolerate a known, limited increase in noise. The "proven track-record" referred to by the company seems to involve an unknown, unlimited decrease in quality. Not really something to trust for archival purposes.
I believe that I may have worked with some of your collaborators
Besides, decades of research have improved those models to the extent that we can accurately predict the weather anywhere up to 20mins in the future.
I was not comparing the iPhone to the the Surface. Seriously, reread my comment because I'm not sure how you could get it that wrong. The clue is the subtle descriptive language like table sized input device. Now you may think that the iPhone is a brick, but surely table-sized is going too far? April 2006 predates the Surface by a long way. It was not invented by microsoft.
Gee. If you measure your social ceiling in relation to slashdot then you have your own problems. You appear to make the same cultural assumptions about the use of language on this site that the GP did. Perhaps you are incapable of recognising your lack of insight
Seriously... this is a public site so get your fucking head out of your fucking ass and quit complaining about the gp's language. Retarded is a very mild form of abuse in europe, and in comparison to the US market it is a perfectly accurate assessment. What on earth makes your think that he should aim his language at your particular definition of clean?
Some interesting comments. I've used linux for years but recently switched to a mac. Lots of reasons provoked it but the main issue was that I wanted laptop with decent battery life and powersaving that ran some variant of unix. After looking around the macbook-pro worked out cheaper than any similar spec machine (mainly because I get a hefty academic discount and finding something as pokey as the nvidia mobile was difficult).
:) And as for the idea that everthing just magically works... it only took two system updates for Apple to get the bloody keyboard working properly under Leopard. Even mickysoft can do better than that...
From the UI perspective I loved it instantly. Some things are just designed right from the ground up, after fighting modal dialog boxes in windows/linux for years I find nonmodal preferences etc a breath of fresh air. Other things confused the crap out of me - the idea that the mac is magically more "usuable" is a nice ideal, but it is as hit and miss as other platforms. Single application instances work for some type of applications, but for other applications (or just processes) I do want to spawn multiple copies (i.e mplayer).
You can get the virtual filesystems that you want as macfuse works pretty well and I use it to mount remote ssh filesystems. Given the focus of the article I've got to say that multitouch scrolling is the best UI improvement that I've seen since the scrollwheel. It is so natural and intuitive. I didn't get the refresh of the macbook-pro with the other multi-touch gestures but I think that I got the one that I would use day in and out. Zoom looks useful but a little gimmicky.
I've suprised that a USB hard drive wouldn't work. Are they eve configurable? Most just expose a FAT filesystem through a standard interface when you plug them in. They even work on my linux desktop at work...
It's strange that you say that because the first multi-touch demo that I remember seeing was a table sized input device that they used to control C&C and google earth. It strikes me as strange because it is the exact same technology that Microsoft have cloned and claimed to have invented with the Surface... Small world when you're competing with Microsoft I guess.
No that's not true at all. I think you've missed the gp's point (ironically enough). Science is not a competition between scientists - although it looks that way from a funding perspective. Science is a competition between the Scientist and the universe. If there is a way to gain an advantage over the universe, and it doesn't have its own issues of morality then it is the correct thing to do.
The stupid hours that we work (I write this at 3am during my second 18hr stint in the lab this week), the coffee that we drink (mmmm tasty) and anything else we do to try and attain an advantage - is fair and within the game that we are playing.
Now the question of whether or not the specific drugs in the article are morally acceptable for general use is a seperate issue. It's not one that I have a strong opinion on as I've never tried them myself and have no desire to do so. But if somebody else uses these substances, and produces good work from it then fair play to them.
The only way to do Science and respect yourself for doing it is to throw everything that you possibly can into it. I would would ask you how you could do science without that level of commitment and still respect yourself for it?
The problem with Blogs is that they are inevitabley the whining and yapping of dogs that don't know any better. The worthless opinion that you link to fails to explain the original experiment correctly before weighing in. While it doesn't add anything of value, I guess it lets you slur the reputation of Dr Chen which is what you apparently wanted to do.
Chen didn't try to prove that the experiment was definitely flawed - he showed that their own reasoning for why it was correct was not valid. That is there were no conclusions that could be derived from the experiment because their methodology was incorrect.
It's actually quite a nice point that he's made, although some other poster further up pointed out that there are much deeper flaws in the original experiment such as the complete lack of a decent control. If they'd rerun it with an animal that had monochrome vision and they still got a 2/3s result then they would have known that something was amiss.
As far as Chen's point goes, I suspect most, if not all animals with chromatic vision have a subtle ranking, because we have evolved to look for certain foods. This is suspected to be why humans can distinguish red/green combinations much better than others.
Thanks for the reply, that was very informative. The section about what is wrong with NAT that firewalls fix is a bit loose - the points about open ports and machines in the DMZ apply to firewalls equally as much as NATs. The remote admin and pings are red herrings - most NATs that I've seen in the past few years switch these off by default now.
The points about where firewalls are more secure products were really interesting. I hadn't thought of the sequence number bouncing, and yet it's a basic part of many crypto protocols to stop replay attacks. The outbound filtering matches up well with the other reply about compromised machines.
Good points, cheers for the info.
Interesting. The bank is Natwest so they fall under UK law. I'll have to see if they have a less annoying alternative buried away somewhere.
I've heard this claim before that NAT can be overcome (can't remember where) and it is an interesting point. In the typical installation - private class C behind the NAT. Port forwarding switched off. How exactly would you go about overcoming the firewall?
I can see that a carefully crafted packed could fool the system into relaying it - although if the router has been designed properly the PPP to the outside world and the hub behind should be physically separate networks. But how would you fool it into masquerading the connection to get the other half of the socket connected?
I ask out of genuine curiosity as it is a claim that I've come across but not worked out how it could be done.
Did you really go to all the trouble of misunderstanding him just to flog your point?
At no point did the GP claim that a hardware firewall and a NAT router were the same thing. He pointed out that he doesn't need a (software) firewall on his box when it's sitting behind NAT. You know exactly what he meant, and why in the context of crappy windows security he's correct - but you tried desparately to show that you knew more about the subject than he does.
Reread (exactly) what he said. Understand it. Shut the fuck up.
Annoying though it is my bank worked around this awhile ago. Instead of entering my PIN through the keyboard they flash up a java keyboard with randomised key layout on the screen which I have to click with the mouse. It is more annoying than tapping in the code as it takes effort to read the screen and translate my PIN onto it, but it must save quite a few of their customers from keyloggers. If it becomes popular amongst other banks then expect a similar arms race to the one underway between CAPTCHAs and spammers.
Although the OP has already said above that he was quoting from memory and the number is probably wrong - it would be a neat trick. At the height of the cold war "winning" was defined as having more of your population survive than the other side. This was also the criteria that the Indians were using to claim they could "win" a war against Pakistan if Kashmir ever went hot.
Depending on how long you run the stats for it is not impossible that that percentage of the US would have been wiped out in a full-scale exchange with the old USSR. Not all of them in the initial explosions (which would have blanketed every major urban area and several non-urban but military sites) but the country would not have been able to function in the aftermath.
Ignoring the effects of a nuclear winter and just considering the raw effects of massive irratiated zones up and down the country, complete standstill in economy and transportation. Within a few years of the initial exchange those death tolls don't look quite so unreasonable. Even for the "winning side". Scary stuff indeed.
PS This new enforced preview is a bugger. I missed the date on your pop. estimate and so now I see your point. I thought you merely meant that a percentage that high was unlikey.... oh well, moving on.
True. But it does come up fifth, and number two is wikipedia which has huge pagerank for any subject. So whoever bought it has gained a very high placing for that search query without paying Google any money for a placed search result. Whether or not the money they paid for it compares with the money for paid results on that query is another question entirely.
Why would you have to? Do you know how much importance Google puts on keywords that are in the URL...
I can't wait for that Web12.0 bubble. That's gonna be some good bubble...
You seem to be wrong on just about every point that you've tried to make. Take a look at the previous article on multithreading that made the slashdot frontpage (I think about a week ago). Somebody tried the same argument as you and was shot down quite forcefully in the largest thread in the discussion.
None of the people that argued could come up with a single real-world problem that couldn't be hacked into working on multi-core systems. When you say "SMT" I assume you've made a typo and mean SMP. SMP is a very general class of parallel systems and is certainly not "useless for most general computing". Name a single real application that cannot take advantage of multiple cores.
You don't seem to know the difference between SIMD and MIMD. Here's a clue: multi-core systems are MIMD, although they typically have SIMD instructions within each core.
Your game example is weak; it sounds as if your game has no internal logic. Once you add AI and some sort of world simulation there is plenty of scope for using those 16 cores.
Lastly, quoting Savain just makes you sound like a retard. The guy is a crank and has been pushing that bullshit on his webpage for years. It only shows that he doesn't understand language design or computer architecture as well as he thinks, and that he should really start taking some form of medication.
Yeah, that would be my guess as well. I'm sure that it's deliberate that they've left it vague though. Looking at who is behind it I'm sure it's not something that would have escaped them.
The video was too laggy to watch (guess it got busy), but their draft patent policy made interesting reading. It has one set of rules for open-source usage, and one set of rules for commercial usage. Are they assuming that these two sets of users are disjoint?
I wonder which set of licensing conditions / price a commercial open-source project would fall under?