Well, you need to look no further than the existing systems like Echelon to know wether or not it will be effective or abused.
Effective? They must be covering that up pretty well then. Abused? Well, several non-US companies and countries are a bit angry that Echelon has been used to gain information for US corporations.
Its rather hard to make an exact judgement on how many false positives there would be, because it simply depends on the conditions in which the scans are done and the setup on the software. If you check Visionics homepage and read the data on their software it has about.68 equal error rate. If I read the data correctly, that means that.68% are false positives, and.68% will be false negatives at that, probably optimal for this type of use, threshold of recognition for that certain database test. Its hard to say how that translates into comparing a large database under uncertain conditions with tens of thousands of people, but it does not sound like false hits will be a rare occurance.
The problem is difficult either way, and in my opinion pretty hopeless. You can manage to use security to catch the absolutely most wanted people, and in that case face recog software will perhaps be useful due to a very small database. But try everyone everywhere who has ever been suspected, or even worse, the loads of people with no record at all who may do bad things... it just isnt viable.
Um, the false positive rate in face recognition systems is huge when you're talking about any reasonably large database. You'd have hundreds, if not thousands, of false positives at every airport every day. And the problem is that even if you do have a human, the resemblance would likely be enough that a human would trust the computer unless there is some obvious reason that it is incorrect (wrong skin color, eye color, or something other the computer face recog software would be worse than a human at deciding).
Face recognition software is useful for one-to-many applications like indentifying a certain face against a large database where you can tune it to less tight matches just to find possible faces, or security ID checks where you can tune it to be really tight about the match so people have to try a few times at worst. But since you cant get rid of the errors, only make them more or less likely to be false positives or false negatives, it absolutely stinks at a many-to-many comparison.
Perhaps you've missed this, but it appears that Echelon is mostly used to spy on non-US corporations to the advantage of US corporations. The kickbacks are better for the snoopers that way. It sure has proven good for getting those airplane construction contracts, but how good has it proven for preventing terrorism?
Re:Some perspective on the causualties
on
More WTC News
·
· Score: 2
Want some more perspective?
200000 people die every day.
While they may die of starvation, war, accidents, not being able to afford medicines, or just old age, the pain felt by their close ones is just as real.
Most of that death and pain doesnt look as good on TV or sell commercial spots tho.
Perhaps we should take some time to reflect upon perspective, grief and our frailty as human beings.
Re:What the hell do you expect?
on
More WTC News
·
· Score: 2
You know, possibly, just possibly, if the people operating Echelon had been a bit more on the watch for possible terrorist activity rather than spending their time spying on other countries corporations to further US corporations...
Echelon may have been built to monitor terrorist activity, but it seems it's used mostly to profit from kickbacks or favors, rather than to prevent terrorist activity.
Certainly think twice? Or what? The US will... kill them? Perhaps it escaped your notice, but some of these people actually flew planes right into buildings. They dont strike me as someone who will certainly think twice just because you threaten to kill them.
The US, or preferably the entire international community, because in my opinion this is an attack on humanity, will likely find any surviving responsible people and deal with those, and with all right.
But I dont think for a second that that will deter any further future attacks. They have very likely read the history books. They know the US will retaliate. They just dont care. At all.
The clerk, however, maybe checks you against a few really essential photos. A facial ID database could easily contain thousands or hundreds of thousands of faces (especially if you add people with bad credit, known scammers, shoplifters and other petty criminals). You might not get detained for looking like someone with bad credit, but will they take your credit card?
The identity twin is sortof different. While your fathers friend might be looked at by people seeing his ID, like you say, he didnt actually look like the guy, which was a fairly good giveaway that he actually wasnt the wanted guy. But if you get a facial misidentification, how do you easily prove that you are not the person they're looking for? ID card? Dont think so. You'd have to go for fingerprints or DNA test. Good luck explaining that to a boss when you get stopped going to work... "Sorry, they thought I was a wanted criminal so I had to go to the police station again.". For the 10th time that month. It wont quite go down well after a while.
What makes you think they'll catch the criminal pretty soon? You may magnify the chances that they'll catch really really really stupid criminals, but what's to say the rest dont figure it out and, you know, like, stay out of monitored areas?
Well, you do have a problem there. IMO, the best thing you can hope for is that people like me will find your program useful at work, in which case I'll take the code and do whatever porting is necessary to get it running on HP-UX, AIX and perhaps True64, and submit it back to you.
Ask the people behind the software porting archives for the various platforms, they might be able to hook you up with either hardware or people. Or ask for porting/testing help on newsgroups or mailing lists. There are a lot of people out there with loads of junk hardware. You could try calling the companies in question, and maybe they can put you in touch with someone inside who might be able to arrange either access or testing, but finding the right person would be difficult. You'd probably have better luck finding someone who works for the companies by going over opensource development mailing lists. Oh, and VA/Sourceforge has a compilefarm with several platforms too.
Of course, those options are only valid if you're doing opensource work and your application is both interesting and reasonably well written. Do use autoconf and automake, it makes porting a bit less painful.
If you do proprietary software... well, good luck. There's a reason that software is more or less supported on the various platforms...
You miss the part where the security guard decides that the computer is always right, and the suspect does look somewhat like the criminal, and in any case wether the computer got the ID right or wrong it isnt the security guards problem since it's not his fault, or on his conscience, if someone innocent is misidentified.
It isnt even close to the same thing as with finger print software. Finger prints are several orders of magnitude more reliable than a facial ID. And fingerprints get run on suspects, not on every single person in an area every minute every day of the year. Imagine instead that you have 50 genetically identical twins, of which several were criminals, and the government wanted to put up fingerprint ID terminals all over the place where you would be required to ID yourself. You'd better get used to getting detained all the time...
Well, that's step two of the process. First you use massive facial recognition in public places to arrest everyone a few times per year due to false positives, and maybe rough them up a bit. Then you offer people the possibility of tatooing a barcode on the forehead (or, more likely a id chip implant or something), which could actually be accurately read, saving them from the mistaken identity problem and subsequent beatings.
Ximian is charging now too, I believe, as per last weeks announcements. Eazel... well, this is just a hunch, but you might actually need more than a week between when you try to charge at product release and your company going under. Eazel didnt have the capital to follow the idea through.
I pretty much agree. The competition can deal with MS in the market as long as MS isnt allowed to violate the law to the left and right. But the courts have to take a very strong stand here, and the remedies have to be enforced the second MS is at it again. MS with serious limits on their behaviour is better than two really nasty half-size MS with no limits. They'd just go on cooperating like today anyway because together they'd be as strong or stronger, and neither would gain by changing attitude towards anyone else just because they're different companies...
If anything should be broken up into small itty bitty pieces, or at least thrown in jail (or in mental hospital, which some might merit), it's the execs and board at MS. Teaching our dear beloved high level corporate execs they arent above the law and that they have a responsibility to stay within it, apart from their responsibility to the share holders, is something that is long overdue, and not just at MS.
Huh. Not in my experience. Most applications are bottlenecked because they're written by 3-week VB class graduates it seems. Well, maybe not that bad, but it appears the current popular paradigm in software optimization is 'buy more hardware', and you cant get the application programmers to fix their code until the hardware companies refuse to sell you more hardware because you're making their machines performance look bad (that actually happened to me once...).
Well, actually, from what that article said (I read the same one), the actual original clippy was a quite sane idea and possibly actually useful. The problem was that it didnt pop up unless the user actually needed/could use assistance. Which meant it triggered quite rarely.
Of course, you cant really use such a thing in marketing. Take a salesman trying to play(?) extremely clueless with Word for half an hour before the thing triggers in front of a fidgeting audience who wants to see the latest "impressive" tech... dont think so.
So they made it trigger more often. Good for the demo, absolute mindnumbing idiotic for anyone subjected to the thing in ordinary use...
Must be fun to work for MS research. Watch the few rare good ideas they turn up get turned into absolute crap that you have to feel embarrased about and never mention you were involved in them for fear of a good solid kick in some sensitive part...
a) If they had experience of it before hp-ux 10.20 I can most certainly understand.
b) Not really. Not sure where you're getting versions 7 and 8 from since they'd be before '95 some time. 10.20, which is largely 32 bit, is discontinued in that no new machines since 2000 supports it. 32 HP-UX 11 is sortof pointless except possibly as a migration step (altho most stuff will work on the 64 bit version either way). Which pretty much leaves 64-bit HP-UX 11 as the only port choice.
c) HP 9000 machines beat the crap price/performance wise out of Sun. In case you hadnt noticed, Sun has had SERIOUS performance problems the last few years, not catching up until the US-III this year. Add to that that Solaris is, as you say, very basic and you have to fork out a fortune for Veritas to have volume management and clustering, and the price gets even worse. IBM is fairly good performancewise since their last line was launched, they've long had rather good prices, but while AIX may be sortof stable, the RS6000 hardware blows up like it's part of the standard operation. And IBM support compared to Sun, Compaq or HP? Ehm. Ugh. Well, they do answer the phone. Im not sure there's anybody actually home on the other side, but they do answer the phone.
HP-UX isnt really peculiar anymore. 10.20 was a vast improvement, and in 11.00 I dont get anywhere near the same porting trouble that there used to be. Still some PITA features like having to relink the kernel for some config changes, and screwy config files for different NIC's, but that's easily outweighed by the fact that the machines dont go kaboom like the others when there's a hiccup on the SAN.
Use opt-in email. Set up procmail filter rules to your main mailbox where you only put people whose email you consider worthwhile. Dump the unknowns in a low priority mailbox (or/dev/null, or autorespond). Mailing lists as usual in their own boxes.
And by picking conception as the beginning of life you are faced with a huge moral dilemma. Since each and every one of your cells can potentially become a twin brother to you, how do you face killing thousands of people when you scratch your head? Can you forgive yourself the billions of brothers you will have killed by the time you reach old age? Not to mention when you kill other peoples potential siblings by shaking hands or something. Mass murder.
A cell is a cell. There is no magic in a newly fertilized egg that cannot be reproduced with technology (altho it would be expensive and impractical). Or are the newly fertilized more worthy of life because they happen to be in the right environment with the right cellular membrane?
Go ahead, choose fertilization as the beginning of life for humans. But if a single cell is a human life, then man are you gonna have a problem with guilt.
Well, the donors could always donate their money to some university in a country where the politicians havent been taking their classes on how to run a country in Iran.
Of course, if there was any demographic info, I wouldnt. However, I would be willing to pay for the actual service of having any music I paid for perpetually available in the format of my choice. No more buying it again to have it in the popular digital format of the day.
I'll bet the 6% 'buying music online', were actually talking about buying CD's online.
Pay-per-play, watermarked, uncopyable, one-computer-only, etc music will never ever work. People will stop buying music entirely before that. It just isnt worth the hassle.
The only way I would buy music online would be if I could buy a permanent license to a piece of music. That is, I get a key, they provide me with the music in any form I wish, any time in the future, in a choice of several of the most popular up-to-date formats available.
Judge Jackson did get rather annoyed with MS, because they delayed, stalled, lied, faked evidence, lied, lied, lied, pretended to be victimized, didnt think that law applied to them, lied, lied, faked evidence, lied and lied again.
That is the reason he had a hard time keeping his temper in check.
Of course, he's right, anything Gates or any other MS employee says is inherently without credibility. They seem fundamentally incapable of uttering a complete sentance without lying, and probably merit a long time of serious therapy and medication.
Last time you checked must've been when we were all using vt100's. Software written for NT may work on 2k. It might not. Software written for NT may even break on a service pack. Or, hell, even if you install some software that does a nasty dll upgrade.
You cant have been using computers much if you havent noticed that.
The commercial software for linux is tested on specific platforms, which means they know it runs on that. It will likely run on most other kernels, glibc versions and distributions too, or if there is a problem you can usually solve it with an LD_PRELOAD for a specific library.
This is configurable by the publisher of the ebook in question, Adobes reader supports printing if the publisher allows it, but probably not. After all, if you could print it that would be like copying, which is bad because then they cant sell the same thing to you twice. Heh.
Slashdot Mirror
Well, you need to look no further than the existing systems like Echelon to know wether or not it will be effective or abused.
Effective? They must be covering that up pretty well then. Abused? Well, several non-US companies and countries are a bit angry that Echelon has been used to gain information for US corporations.
So, how do you think it will be used?
Its rather hard to make an exact judgement on how many false positives there would be, because it simply depends on the conditions in which the scans are done and the setup on the software. If you check Visionics homepage and read the data on their software it has about .68 equal error rate. If I read the data correctly, that means that .68% are false positives, and .68% will be false negatives at that, probably optimal for this type of use, threshold of recognition for that certain database test. Its hard to say how that translates into comparing a large database under uncertain conditions with tens of thousands of people, but it does not sound like false hits will be a rare occurance.
The problem is difficult either way, and in my opinion pretty hopeless. You can manage to use security to catch the absolutely most wanted people, and in that case face recog software will perhaps be useful due to a very small database. But try everyone everywhere who has ever been suspected, or even worse, the loads of people with no record at all who may do bad things... it just isnt viable.
Um, the false positive rate in face recognition systems is huge when you're talking about any reasonably large database. You'd have hundreds, if not thousands, of false positives at every airport every day. And the problem is that even if you do have a human, the resemblance would likely be enough that a human would trust the computer unless there is some obvious reason that it is incorrect (wrong skin color, eye color, or something other the computer face recog software would be worse than a human at deciding).
Face recognition software is useful for one-to-many applications like indentifying a certain face against a large database where you can tune it to less tight matches just to find possible faces, or security ID checks where you can tune it to be really tight about the match so people have to try a few times at worst. But since you cant get rid of the errors, only make them more or less likely to be false positives or false negatives, it absolutely stinks at a many-to-many comparison.
Perhaps you've missed this, but it appears that Echelon is mostly used to spy on non-US corporations to the advantage of US corporations. The kickbacks are better for the snoopers that way. It sure has proven good for getting those airplane construction contracts, but how good has it proven for preventing terrorism?
Want some more perspective?
200000 people die every day.
While they may die of starvation, war, accidents, not being able to afford medicines, or just old age, the pain felt by their close ones is just as real.
Most of that death and pain doesnt look as good on TV or sell commercial spots tho.
Perhaps we should take some time to reflect upon perspective, grief and our frailty as human beings.
You know, possibly, just possibly, if the people operating Echelon had been a bit more on the watch for possible terrorist activity rather than spending their time spying on other countries corporations to further US corporations...
Echelon may have been built to monitor terrorist activity, but it seems it's used mostly to profit from kickbacks or favors, rather than to prevent terrorist activity.
Certainly think twice? Or what? The US will... kill them? Perhaps it escaped your notice, but some of these people actually flew planes right into buildings. They dont strike me as someone who will certainly think twice just because you threaten to kill them.
The US, or preferably the entire international community, because in my opinion this is an attack on humanity, will likely find any surviving responsible people and deal with those, and with all right.
But I dont think for a second that that will deter any further future attacks. They have very likely read the history books. They know the US will retaliate. They just dont care. At all.
Mmmm... not sure you'd want Lemmings if you're detaining possible suicide bombers.
5...4...3...2...1 OH NO! Kaboom.
The clerk, however, maybe checks you against a few really essential photos. A facial ID database could easily contain thousands or hundreds of thousands of faces (especially if you add people with bad credit, known scammers, shoplifters and other petty criminals). You might not get detained for looking like someone with bad credit, but will they take your credit card?
The identity twin is sortof different. While your fathers friend might be looked at by people seeing his ID, like you say, he didnt actually look like the guy, which was a fairly good giveaway that he actually wasnt the wanted guy. But if you get a facial misidentification, how do you easily prove that you are not the person they're looking for? ID card? Dont think so. You'd have to go for fingerprints or DNA test. Good luck explaining that to a boss when you get stopped going to work... "Sorry, they thought I was a wanted criminal so I had to go to the police station again.". For the 10th time that month. It wont quite go down well after a while.
What makes you think they'll catch the criminal pretty soon? You may magnify the chances that they'll catch really really really stupid criminals, but what's to say the rest dont figure it out and, you know, like, stay out of monitored areas?
Well, you do have a problem there. IMO, the best thing you can hope for is that people like me will find your program useful at work, in which case I'll take the code and do whatever porting is necessary to get it running on HP-UX, AIX and perhaps True64, and submit it back to you.
Ask the people behind the software porting archives for the various platforms, they might be able to hook you up with either hardware or people. Or ask for porting/testing help on newsgroups or mailing lists. There are a lot of people out there with loads of junk hardware. You could try calling the companies in question, and maybe they can put you in touch with someone inside who might be able to arrange either access or testing, but finding the right person would be difficult. You'd probably have better luck finding someone who works for the companies by going over opensource development mailing lists. Oh, and VA/Sourceforge has a compilefarm with several platforms too.
Of course, those options are only valid if you're doing opensource work and your application is both interesting and reasonably well written. Do use autoconf and automake, it makes porting a bit less painful.
If you do proprietary software... well, good luck. There's a reason that software is more or less supported on the various platforms...
You miss the part where the security guard decides that the computer is always right, and the suspect does look somewhat like the criminal, and in any case wether the computer got the ID right or wrong it isnt the security guards problem since it's not his fault, or on his conscience, if someone innocent is misidentified.
It isnt even close to the same thing as with finger print software. Finger prints are several orders of magnitude more reliable than a facial ID. And fingerprints get run on suspects, not on every single person in an area every minute every day of the year. Imagine instead that you have 50 genetically identical twins, of which several were criminals, and the government wanted to put up fingerprint ID terminals all over the place where you would be required to ID yourself. You'd better get used to getting detained all the time...
Well, that's step two of the process. First you use massive facial recognition in public places to arrest everyone a few times per year due to false positives, and maybe rough them up a bit. Then you offer people the possibility of tatooing a barcode on the forehead (or, more likely a id chip implant or something), which could actually be accurately read, saving them from the mistaken identity problem and subsequent beatings.
Ximian is charging now too, I believe, as per last weeks announcements. Eazel... well, this is just a hunch, but you might actually need more than a week between when you try to charge at product release and your company going under. Eazel didnt have the capital to follow the idea through.
I pretty much agree. The competition can deal with MS in the market as long as MS isnt allowed to violate the law to the left and right. But the courts have to take a very strong stand here, and the remedies have to be enforced the second MS is at it again. MS with serious limits on their behaviour is better than two really nasty half-size MS with no limits. They'd just go on cooperating like today anyway because together they'd be as strong or stronger, and neither would gain by changing attitude towards anyone else just because they're different companies...
If anything should be broken up into small itty bitty pieces, or at least thrown in jail (or in mental hospital, which some might merit), it's the execs and board at MS. Teaching our dear beloved high level corporate execs they arent above the law and that they have a responsibility to stay within it, apart from their responsibility to the share holders, is something that is long overdue, and not just at MS.
Not like that will ever happen tho.
Huh. Not in my experience. Most applications are bottlenecked because they're written by 3-week VB class graduates it seems. Well, maybe not that bad, but it appears the current popular paradigm in software optimization is 'buy more hardware', and you cant get the application programmers to fix their code until the hardware companies refuse to sell you more hardware because you're making their machines performance look bad (that actually happened to me once...).
Well, actually, from what that article said (I read the same one), the actual original clippy was a quite sane idea and possibly actually useful. The problem was that it didnt pop up unless the user actually needed/could use assistance. Which meant it triggered quite rarely.
Of course, you cant really use such a thing in marketing. Take a salesman trying to play(?) extremely clueless with Word for half an hour before the thing triggers in front of a fidgeting audience who wants to see the latest "impressive" tech... dont think so.
So they made it trigger more often. Good for the demo, absolute mindnumbing idiotic for anyone subjected to the thing in ordinary use...
Must be fun to work for MS research. Watch the few rare good ideas they turn up get turned into absolute crap that you have to feel embarrased about and never mention you were involved in them for fear of a good solid kick in some sensitive part...
a) If they had experience of it before hp-ux 10.20 I can most certainly understand.
b) Not really. Not sure where you're getting versions 7 and 8 from since they'd be before '95 some time. 10.20, which is largely 32 bit, is discontinued in that no new machines since 2000 supports it. 32 HP-UX 11 is sortof pointless except possibly as a migration step (altho most stuff will work on the 64 bit version either way). Which pretty much leaves 64-bit HP-UX 11 as the only port choice.
c) HP 9000 machines beat the crap price/performance wise out of Sun. In case you hadnt noticed, Sun has had SERIOUS performance problems the last few years, not catching up until the US-III this year. Add to that that Solaris is, as you say, very basic and you have to fork out a fortune for Veritas to have volume management and clustering, and the price gets even worse. IBM is fairly good performancewise since their last line was launched, they've long had rather good prices, but while AIX may be sortof stable, the RS6000 hardware blows up like it's part of the standard operation. And IBM support compared to Sun, Compaq or HP? Ehm. Ugh. Well, they do answer the phone. Im not sure there's anybody actually home on the other side, but they do answer the phone.
HP-UX isnt really peculiar anymore. 10.20 was a vast improvement, and in 11.00 I dont get anywhere near the same porting trouble that there used to be. Still some PITA features like having to relink the kernel for some config changes, and screwy config files for different NIC's, but that's easily outweighed by the fact that the machines dont go kaboom like the others when there's a hiccup on the SAN.
Use opt-in email. Set up procmail filter rules to your main mailbox where you only put people whose email you consider worthwhile. Dump the unknowns in a low priority mailbox (or /dev/null, or autorespond). Mailing lists as usual in their own boxes.
And by picking conception as the beginning of life you are faced with a huge moral dilemma. Since each and every one of your cells can potentially become a twin brother to you, how do you face killing thousands of people when you scratch your head? Can you forgive yourself the billions of brothers you will have killed by the time you reach old age? Not to mention when you kill other peoples potential siblings by shaking hands or something. Mass murder.
A cell is a cell. There is no magic in a newly fertilized egg that cannot be reproduced with technology (altho it would be expensive and impractical). Or are the newly fertilized more worthy of life because they happen to be in the right environment with the right cellular membrane?
Go ahead, choose fertilization as the beginning of life for humans. But if a single cell is a human life, then man are you gonna have a problem with guilt.
Well, the donors could always donate their money to some university in a country where the politicians havent been taking their classes on how to run a country in Iran.
Of course, if there was any demographic info, I wouldnt. However, I would be willing to pay for the actual service of having any music I paid for perpetually available in the format of my choice. No more buying it again to have it in the popular digital format of the day.
If the price was right, that is.
I'll bet the 6% 'buying music online', were actually talking about buying CD's online.
Pay-per-play, watermarked, uncopyable, one-computer-only, etc music will never ever work. People will stop buying music entirely before that. It just isnt worth the hassle.
The only way I would buy music online would be if I could buy a permanent license to a piece of music. That is, I get a key, they provide me with the music in any form I wish, any time in the future, in a choice of several of the most popular up-to-date formats available.
Judge Jackson did get rather annoyed with MS, because they delayed, stalled, lied, faked evidence, lied, lied, lied, pretended to be victimized, didnt think that law applied to them, lied, lied, faked evidence, lied and lied again.
That is the reason he had a hard time keeping his temper in check.
Of course, he's right, anything Gates or any other MS employee says is inherently without credibility. They seem fundamentally incapable of uttering a complete sentance without lying, and probably merit a long time of serious therapy and medication.
Last time you checked must've been when we were all using vt100's. Software written for NT may work on 2k. It might not. Software written for NT may even break on a service pack. Or, hell, even if you install some software that does a nasty dll upgrade.
You cant have been using computers much if you havent noticed that.
The commercial software for linux is tested on specific platforms, which means they know it runs on that. It will likely run on most other kernels, glibc versions and distributions too, or if there is a problem you can usually solve it with an LD_PRELOAD for a specific library.
This is configurable by the publisher of the ebook in question, Adobes reader supports printing if the publisher allows it, but probably not. After all, if you could print it that would be like copying, which is bad because then they cant sell the same thing to you twice. Heh.