Beta
×

Welcome to the Slashdot Beta site -- learn more here. Use the link in the footer or click here to return to the Classic version of Slashdot.

Thank you!

Before you choose to head back to the Classic look of the site, we'd appreciate it if you share your thoughts on the Beta; your feedback is what drives our ongoing development.

Beta is different and we value you taking the time to try it out. Please take a look at the changes we've made in Beta and  learn more about it. Thanks for reading, and for making the site better!

Facial Recognition Gone Wrong

samzenpus posted more than 3 years ago | from the doppleganger-trouble dept.

Government 375

An anonymous reader writes "John H. Gass hadn't had a traffic ticket in years, so the Natick resident was surprised this spring when he received a letter from the Massachusetts Registry of Motor Vehicles informing him to cease driving because his license had been revoked. It turned out Gass was flagged because he looks like another driver, not because his image was being used to create a fake identity. His driving privileges were returned but, he alleges in a lawsuit, only after 10 days of bureaucratic wrangling to prove he is who he says he is. And apparently, he has company. Last year, the facial recognition system picked out more than 1,000 cases that resulted in State Police investigations, officials say. And some of those people are guilty of nothing more than looking like someone else. Not all go through the long process that Gass says he endured, but each must visit the Registry with proof of their identity. Massachusetts began using the software after receiving a $1.5 million grant from the US Department of Homeland Security as part of an effort to prevent terrorism, reduce fraud, and improve the reliability and accuracy of personal identification documents that states issue."

cancel ×

375 comments

Sorry! There are no comments related to the filter you selected.

It's a growing list (1)

retroworks (652802) | more than 3 years ago | (#36798528)

As Whitey Bulger proved, it's not who you are, in Massachusetts, it's who you know. And now, who you look like.

Re:It's a growing list (2)

dkleinsc (563838) | more than 3 years ago | (#36798560)

in Massachusetts, it's who you know

And here I was thinking the key question is whether your last name was "Kennedy".

Re:It's a growing list (1)

tacktick (1866274) | more than 3 years ago | (#36798720)

This is a good thing. Especially if some criminal stole your identity and picture.
They should work on communicating with people better though.

Re:It's a growing list (3, Insightful)

maxume (22995) | more than 3 years ago | (#36799146)

Please stop saying that identity can be stolen. If someone makes themselves look like you it is impersonation. If they use information associated with you to bamboozle the weak-ass authentication used by financial institutions, it is fraud.

In neither case has your identity been stolen. A man's wife would not sleep with a different man simply because the second man had a bank account in her husband's name, and so on.

Re:It's a growing list (1)

kelemvor4 (1980226) | more than 3 years ago | (#36799178)

This is a good thing. Especially if some criminal stole your identity and picture. They should work on communicating with people better though.

I think you'd have a hard time convincing John H. Gass

Nice work. (5, Insightful)

geminidomino (614729) | more than 3 years ago | (#36798532)

Massachusetts began using the software... to prevent terrorism, reduce fraud, and improve the reliability and accuracy of personal identification documents that states issue."

Came up snake-eyes on that role, dincha?

Urg. (2, Informative)

geminidomino (614729) | more than 3 years ago | (#36798582)

And I cut myself with my own rapier wit by messing up a quote tag and using the wrong homophone.

Coffee needs to brew faster...

Re:Urg. (1, Interesting)

rbrausse (1319883) | more than 3 years ago | (#36798620)

Coffee needs to brew faster...

great stuff [ineedcoffee.com] :)

Re:Nice work. (1)

fuzzyfuzzyfungus (1223518) | more than 3 years ago | (#36798924)

Barring some research on the reliability of the humans previously used(or, if no checks were done, estimates of likely false negatives), it's sort of hard to say.

It is well known that machine-vision is still rather dodgy outside of well-controlled applications(pick-and-place? the puny humans might as well surrender. general purpose object recognition in cluttered environments? not quite so good.); but it is also known that humans are quite fallible, and generally more fallible than they think.

An anecdote, however Kafkaesque, can't really tell you much about "improve[d] reliability and accuracy". It can tell you that customer service sucks, or that a department is taking a guilty-until-proven-innocent approach; but it can't tell you whether they were doing more or less of that than before...

Re:Nice work. (1)

Bob the Super Hamste (1152367) | more than 3 years ago | (#36798944)

Hell they completely struck out 0 of 3. The only point that might be in question would be the fraud but I would argue that fraud was committed when the purchased the product since it doesn't deliver what it promises. If it can't correctly match faces how could it be expected to prevent terrorism, reduce fraud, and increase accuracy of identification. I will believe that facial recognition will work once we can get OCT that works on typed (not even hand written) text that has an accuracy of 100% until then software like this should be considered if junk if the false positives are greater than the true negatives, or true positives depending on how the matrix is defined.

Re:Nice work. (2)

Bob the Super Hamste (1152367) | more than 3 years ago | (#36799018)

I would like to correct my previous post as I am not fully with it yet this morning and got my validation matrix confused.

Hell they completely struck out 0 of 3. The only point that might be in question would be the fraud but I would argue that fraud was committed when the purchased the product since it doesn't deliver what it promises. If it can't correctly match faces how could it be expected to prevent terrorism, reduce fraud, and increase accuracy of identification. I will believe that facial recognition will work once we can get OCT that works on typed (not even hand written) text that has an accuracy of 100% until then software like this should be considered if junk if the false positives are greater than the true positives.

This is especially true since there are negative consequences for the individual who was flagged as a false positive.

Guilty until proven innocent (4, Informative)

yincrash (854885) | more than 3 years ago | (#36798536)

Kaprielian said the Registry gives drivers enough time to respond to the suspension letters and that it is the individual’s “burden’’ to clear up any confusion. She added that protecting the public far outweighs any inconvenience Gass or anyone else might experience. “A driver’s license is not a matter of civil rights. It’s not a right. It’s a privilege,’’ she said. “Yes, it is an inconvenience [to have to clear your name], but lots of people have their identities stolen, and that’s an inconvenience, too.’’

Re:Guilty until proven innocent (4, Insightful)

Vanderhoth (1582661) | more than 3 years ago | (#36798572)

According to the statement you're guilty until you prove your innocent, so much for innocent until proven guilty. I'm sure using tax dollars and grants to use a system that illegally convicts innocent people without a trial or hearing is considerable different then a criminal stealing someones identity.

Oblig. Star Trek reference (5, Insightful)

TapeCutter (624760) | more than 3 years ago | (#36798624)

you're guilty until you prove your innocent

...because it would be unfair to put an innocent person on trial.

Re:Oblig. Star Trek reference (1)

Vanderhoth (1582661) | more than 3 years ago | (#36798782)

No, The RMV could have a person investigate incidents flagged by the program, which would be presuming the program could be wrong. This is despite of the fact that they know and admit the program isn't 100% accurate. Regardless they seem to be using the assumption that the statistical computer program is infallible and the people it's flagging are guilty and thus the onus is on the innocent party to prove they just happen to look like someone who did something wrong.

Re:Oblig. Star Trek reference (0)

Anonymous Coward | more than 3 years ago | (#36799082)

> the onus is on the innocent party to prove they just happen to look like someone who did something wrong

It's not even that - the onus is on you to prove you don't look like *another innocent bystander*. The article didn't even say that either of the people had done anything wrong, just that they looked similar and the software is intended to make sure that one person only has one license.

Re:Guilty until proven innocent (0)

Anonymous Coward | more than 3 years ago | (#36798662)

Please don't confuse the issue. There are certain situations where innocent until proven guilty does not apply without restrictions. Take for instance, the situation where a cop needs to give someone a ticket for a broken tail-light. The ticket coming from the cop needs to carry a presumption of veracity, otherwise it would be very impractical for cops to ticket anyone. Of course there are issues that arise from that, especially with automated computer systems; but, like I said, the corollary does not does not apply unrestrictedly.

Re:Guilty until proven innocent (2)

Vanderhoth (1582661) | more than 3 years ago | (#36798734)

However, they admit that the system is flawed and frequently picks people who have done nothing wrong. The solution would be to have a physical person investigate each incident the program is reporting. Instead the program is taken as infallible, even though its users know it's not, and innocent people are assumed to be at fault until that person clears their own name.

Re:Guilty until proven innocent (0)

Anonymous Coward | more than 3 years ago | (#36799076)

Maybe the solution would be to heavily sanction the responsible law enforcer (right up to where the buck stops) who wrongly (and automatically, without trial) penalised someone who did nothing wrong. As in, perhaps, monetary fines, paid to the inconvenienced citizen to cover lost time, income, expenses, legal costs, etc. (should be paid out of his personal funds, not out of state funds). Could also be loaded with a predetermined factor/premium to cover aggravation. Else jail time for said individual. Oh, I forgot: automatic computer, so nobody is responsible.... well tough for you, Mr Governor, you should have thought it out before causing a law-abiding citizen to be prosecuted as if guilty.

But I dunno. This creates another layer in the system that can be abused for political and/or monetary gain.

Ahhh stuffit, just nuke the whole place out of existence and start from scratch.

Re:Guilty until proven innocent (1)

Attila Dimedici (1036002) | more than 3 years ago | (#36798754)

Take for instance, the situation where a cop needs to give someone a ticket for a broken tail-light. The ticket coming from the cop needs to carry a presumption of veracity, otherwise it would be very impractical for cops to ticket anyone.

Are you aware that when a cop writes a ticket for a broken tail-light, that is equivalent to when a cop arrests someone for murder? In both cases, one must still choose to either plead "guilty" or "not guilty". In the case of the ticket, the penalty for pleading guilty is pre-determined and all one needs to do is mail in a check along with the ticket signed that you are pleading guilty. If you plead "not guilty", you need to provide some evidence that the police officer is lying about the broken tail-light, just as you would if the cop claimed to have seen you kill someone and you were pleading "not guilty".

Re:Guilty until proven innocent (1)

yincrash (854885) | more than 3 years ago | (#36798816)

Exactly. You have a right to a hearing, and you don't have to pay your fine before the hearing, or any fine at all if you are found innocent.

Re:Guilty until proven innocent (-1, Troll)

pnewhook (788591) | more than 3 years ago | (#36798790)

According to the statement you're guilty until you prove your innocent, so much for innocent until proven guilty.

Welcome to the result of 8 years of Bush rule. (Yes that colossal waste of tax dollars and abuse of power that is called Homeland Security was created in the Bush era)

Re:Guilty until proven innocent (3)

Jawnn (445279) | more than 3 years ago | (#36798842)

According to the statement you're guilty until you prove your innocent, so much for innocent until proven guilty.

Welcome to the result of 10 years of Bush rule. (Yes that colossal waste of tax dollars and abuse of power that is called Homeland Security was created in the Bush era)

TFTFY, 'cause, to be fair, the current leaders have done nothing but continue to rubber-stamp all those "patriotic" abuses and outright violations of our civil rights.

Re:Guilty until proven innocent (2, Insightful)

ScentCone (795499) | more than 3 years ago | (#36798900)

was created in the Bush era

It was created by congress. Of course you know that, and are just hoping that nobody else actually understands how things work.

I'm also curious what magical power you think that The Eeeevil Bush had, after leaving office, that allowed him to prevent a complete lefty/Democrat monopoly of power in the legislative and executive branches from altering the policy and funding of the agency you're whining about. If it was Eeeeevil Bush who controlled congress even though ultra-liberal Nancy Pelosi actually ran that institution from 2006 until just recently, how exactly is it that he was getting his Eeeeevil work done? Really. Please provide the details, which should be fascinating.

I just hope Leonardo DeCaprio stays out of trouble (1)

elrous0 (869638) | more than 3 years ago | (#36799034)

Because he could really drag me down with him.

Re:Guilty until proven innocent (5, Insightful)

indyogb (1517319) | more than 3 years ago | (#36798632)

“A driver’s license is not a matter of civil rights. It’s not a right. It’s a privilege..."

So says the government(s). I disagree. Just because something isn't specifically protected by the Constitution doesn't mean it isn't a right. Travel by the standard means of the time (in this case, automobiles), is a natural right. Also, it is nice that a system used to "prevent terrorism" is being used to suspend driver's licenses of ordinary, non-terrorist, citizens.

Government(s) in the US are at flank speed ahead towards power and control. Even the court system is on their side (e.g. imminent domain for increased tax revenues from a few years back, recent rulings about police entering homes w/o warrants in IN, etc., etc.). In the end, it is all about the $$$. Where is it, who has it, and how can we get more of it.

Re:Guilty until proven innocent (-1, Flamebait)

Anonymous Coward | more than 3 years ago | (#36798684)

If it was a right you wouldn't need to have a licence in the first place, would you?

Fucking idiot.

Re:Guilty until proven innocent (1, Informative)

alen (225700) | more than 3 years ago | (#36798714)

someone got up on the wrong side of the keyboard today

Re:Guilty until proven innocent (4, Informative)

Dunbal (464142) | more than 3 years ago | (#36798768)

You don't need a license to drive a car. You need a license to drive a car on a PUBLIC ROAD. You have an inherent right to drive. When you want to drive in a public place, you have to prove that you meet certain standards for the safety of other public road users (thus the driving test). A license is not a "gift" from the government, it is in fact a responsibility that the government has ensured you are a safe vehicle operator. Eventually this has turned into an additional source of revenue and now, apparently, a "privilege". Beware of people who say rights are privileges, they are the ones who would see you in chains.

Re:Guilty until proven innocent (1)

vlm (69642) | more than 3 years ago | (#36798928)

You don't need a license to drive a car. You need a license to drive a car on a PUBLIC ROAD. You have an inherent right to drive. When you want to drive in a public place, you have to prove that you meet certain standards for the safety of other public road users (thus the driving test). A license is not a "gift" from the government, it is in fact a responsibility that the government has ensured you are a safe vehicle operator. Eventually this has turned into

an identification card, which if you think about it is pretty weird.

I have had numerous government licenses over the years, amateur radio license, student pilot license (which at that time was just an instructor signed 3rd class med cert), fishing license, parking permit at govt facility, a security pass at an army base, and none of them had the overhead of being "identification".

Probably time to separate the concept of driver license and id card.

Re:Guilty until proven innocent (1)

c (8461) | more than 3 years ago | (#36798752)

> Travel by the standard means of the time (in this case, automobiles), is a natural right.

There's also the bit in TFA about the guy driving as part of his job. Many western jurisdictions have some concept of a "right to make a living"...

Re:Guilty until proven innocent (-1)

Anonymous Coward | more than 3 years ago | (#36798644)

how do people with such defective brains even find employment? we'd be better off if we chemically digested her and used her as fertilizer. at least she's be useful then.

Re:Guilty until proven innocent (-1)

Anonymous Coward | more than 3 years ago | (#36798922)

worthless mods. you're also useless and would be better used to grow crops.

Re:Guilty until proven innocent (2)

goldspider (445116) | more than 3 years ago | (#36798646)

So by their own admission, their error is comparable to stealing someone's identity. And they don't see this as a problem.

Is there anything bureaucrats can't cock up?

Re:Guilty until proven innocent (3, Insightful)

Vanderhoth (1582661) | more than 3 years ago | (#36798682)

Is there anything bureaucrats can't cock up?

Breathing. Otherwise they wouldn't live long enough to make our lives so difficult.

Re:Guilty until proven innocent (2)

Nofsck Ingcloo (145724) | more than 3 years ago | (#36798674)

“A driver’s license is not a matter of civil rights. It’s not a right. It’s a privilege,’’ she said. Yep, that "logic" is used to shred the Fourth Amendment and now to enable this junk. We've got to fix this bug. The right to travel unmolested by car should, inded, be a civil right.

Re:Guilty until proven innocent (2)

goldspider (445116) | more than 3 years ago | (#36798894)

Therefore you have no rights while you are driving on a public thoroughfare.

And people STILL dismiss "slippery slope" arguments.

Identity Theft? Way to Aim High (3, Insightful)

eldavojohn (898314) | more than 3 years ago | (#36798688)

“Yes, it is an inconvenience [to have to clear your name], but lots of people have their identities stolen, and that’s an inconvenience, too.’’

So their defense is to list crimes that are worse than what they (law enforcement) are doing? I guess if you aim low, there's no chance of failure.

So if I went up to someone and said, "Hey, I know you think I'm a jerk because I call you harmful names but lots of people get raped in a parking lot and that's harmful too." They should thank their lucky stars I'm just calling them names and not raping them in a parking lot? Isn't that more of a threat than an excuse? I don't get it, is the Registry of Motor Vehicles threatening to steal or sell everyone's identity if they don't like being wrongly accused?

Facial recognition is not quite yet where it has to be. I worked on some of this stuff way back in college and the case studies we did on open face databases had abysmal recall rates [wikipedia.org] . Basically it should be concluded that until your chance of a false positive is equivalent with winning the lottery, you shouldn't implement this. I say "winning the lottery" because it is such a terrible violation of rights that you should be prepared to pay out a million dollars to the poor citizen that is wrongly accused of some crime or infraction just based upon the features of their face. It's a high stakes game and if you're going to use it as a short cut, you better be prepared to accept a high amount of risk.

Re:Identity Theft? Way to Aim High (1)

Terrasque (796014) | more than 3 years ago | (#36799142)

So their defense is to list crimes that are worse than what they (law enforcement) are doing?

"Your Honor, it's true that I've stolen millions of dollars, and beaten people to within an inch of their lives, but remember that there are people out there who have KILLED hundreds of people!"

"You know, you make a really compelling argument there. You're free to go."

or, does this only work if you're in the government?

Re:Guilty until proven innocent (0)

Anonymous Coward | more than 3 years ago | (#36798786)

“Yes, it is an inconvenience [to have to clear your name], but lots of people have their identities stolen, and that’s an inconvenience, too.’’

So according to Massachusetts logic, if a police officer beats you with his club for no reason, it's all ok as long as he tells you you're lucky to have not been shot?

Re:Guilty until proven innocent (1)

chomsky68 (1719996) | more than 3 years ago | (#36798788)

I hope that one day Kaprielian gets mixed up the same way and is subjected to a full cavity search at the airport...

Re:Guilty until proven innocent (3, Interesting)

martyros (588782) | more than 3 years ago | (#36799070)

I think that some people need to make masks that look just like Kaprielian, and probably her boss, and some of the local senators, and get "caught" by some of these cameras doing something that results in automatic suspension of a license. Then we'll see how long the "It's just inconvenient" attitude lasts.

In Michigan several years ago they passed a law that allowed a policeman to cut up your driving license right in front of you if you were caught driving drunk. That law was struck down as being unconstitutional, because even though the cop was right there and could smell the alcohol on your breath and hear your slurred speech, a cop is not judge and jury; you still have a right to due process under the law. If that was unconstitutional, I can't see how this isn't as well.

Your tax dollars at work! (1)

Anonymous Coward | more than 3 years ago | (#36798564)

Tell me again why we need to raise taxes to close government deficits when they're spending money on crap like this?

Re:Your tax dollars at work! (1)

JustOK (667959) | more than 3 years ago | (#36798634)

terrorists.

I've got nothing to hide (4, Insightful)

captainpanic (1173915) | more than 3 years ago | (#36798574)

... except my face apparently.

Anyone still wondering why privacy is such an important issue? I never want to hear the "I have nothing to hide" argument again.

Re:I've got nothing to hide (1)

Derf the (610150) | more than 3 years ago | (#36798698)

At last, an argument for the Western, all sex's, Burka [well a niqaab, at least].
You heard it here first.

Re:I've got nothing to hide (4, Interesting)

Gideon Wells (1412675) | more than 3 years ago | (#36798808)

I tried explaining the TSA mess to my sister and mother this weekend. They said it sounded terrific that they were checking elderly people with diapers and making wheel chaired bound people prove they needed a wheelchair.

Mother: "We'll be safer."
Sister: "I have nothing to hide."

Me:"Terrorists have announced plans to start trying to hide bombs surgically implanted in their skin."

Mother: "So... they have to perform surgery mid-flight to blow themselves up? *laugh*"

Me: "Remote detonation. First time they catch someone attempting this the TSA will start requiring medical records if they see an operation scar."

Sister: "So what? I got nothing to hide. In fact, I think it is a good idea for them to have our medical records to make sure our flights are safe."

They are the first to Google people, gossip, etc. My sister has been caught looking at homes with binoculars at night. When a family friend joking texted her saying he could see her at night you'd be surprised how fast her translucent curtains were replaced with thick drapes. Some people don't care about privacy until it affects them.

Re:I've got nothing to hide except my face (2)

TaoPhoenix (980487) | more than 3 years ago | (#36798920)

All you IAALs out there, can we get some ThisIsAdviceButNotLegalAdvice?

Motorcycle riders would escape this system, wouldn't they? And they are almost the only group allowed to wear helmets.

I thought I read (but it may have been an unreputable source) that we are allowed to wear helmets while driving our cars. Is that true? Or do the other characteristics of driving cars such as different view ranges negate that legality?

What if a cop pulls you over and actually states that he doesn't like you wearing your helmet?

Re:I've got nothing to hide except my face (1)

rbrausse (1319883) | more than 3 years ago | (#36799028)

maybe I read a different article but the description sounds like a database search; they compared copies of the license pictures and similar pictures were selected.

It happened to me (-1)

Anonymous Coward | more than 3 years ago | (#36798584)

A couple months ago, I picked up a trollop at a bar. She sucked my dick and I gave her a facial. Later, I found out she had dude parts. I was shocked for a couple days, but I can laugh about it now. Oh, and the conventional wisdom that men give better blowjobs than women -- not true, in my (one time) experience.

In other news (4, Funny)

larry bagina (561269) | more than 3 years ago | (#36798600)

The ethnic population in Massachusetts has shrunk to one black and one asian.

Re:In other news (1)

eyebits (649032) | more than 3 years ago | (#36799104)

>...one asian. Lol. I guess you've never been to MIT.

Fight Terrorism, ban cars! (0)

Anonymous Coward | more than 3 years ago | (#36798622)

...from the Massachusetts Registry of Motor Vehicles...
Massachusetts began using the software...as part of an effort to prevent terrorism,...

Someone got their driving license revoked due to a program to combat terrorism?
Oh no! Terrorists are driving cars! Quickly! Let's outlaw cars!

Oh, and incidentally:...to prevent terrorism, reduce fraud, ...
"reduce fraud"? *rofl*

That is just so hilarious in this context :)

Governments Are Always Right (1)

shawnhcorey (1315781) | more than 3 years ago | (#36798638)

I don't see what the complaint is about. For years, the governments of the world have a no-fly list. If your name appears on it, you can't fly. It doesn't matter if you didn't do anything wrong, because Governments Are Always Right. Stop whining. :)

Pastafarianism will solve this (5, Funny)

TenDollarMan (1307733) | more than 3 years ago | (#36798640)

All you need to do is wear a welding mask as your Pastafarian religous headwear.

It works in Austria. G'day mate.

Re:Pastafarianism will solve this (0, Funny)

Anonymous Coward | more than 3 years ago | (#36798664)

There is so much wrong with this post I don't know where to begin. It's like you picked a handful a meme and got everything about it wrong.

Re:Pastafarianism will solve this (-1)

Anonymous Coward | more than 3 years ago | (#36798748)

Hey ho buster......Please spell check ....wait for it.....next time you're going to complain about someone elses errors!!

Re:Pastafarianism will solve this (1)

jeffmeden (135043) | more than 3 years ago | (#36799052)

There is so much wrong with this post I don't know where to begin. It's like you picked a handful a meme and got everything about it wrong.

Or, he was going for a "Dumb and Dumber" reference and got everything about it *right*...

you know who else didn't know where to begin? (1)

decora (1710862) | more than 3 years ago | (#36799094)

thats right. adolph hitler.

"I don't know where to begin"

-- Mein Kampf, 1987.

That was right before he killed all the Muslims.

Re:Pastafarianism will solve this (0)

Inda (580031) | more than 3 years ago | (#36798774)

Austria and Australia are on the same map.

You are all forgiven.

Re:Pastafarianism will solve this (1, Funny)

chomsky68 (1719996) | more than 3 years ago | (#36798862)

It works in Austria. Grüß Gott Kumpel.

FTFY

Re:Pastafarianism will solve this (1)

Jason Levine (196982) | more than 3 years ago | (#36799030)

So then one Pastafarian bad driver will cause all of the others to lose their licenses.

Sounds Like a Good System (1)

Gr33nJ3ll0 (1367543) | more than 3 years ago | (#36798650)

After RTFA it appears that they're sending out maybe 1,500 of these every year, of which there are now 2 known problems, and his complaint is that he went without his license for a week or so while he cleared his name. Now he's sueing the state, for undisclosed sums, but even he would be hard pressed to really say he was out much other than a weeks wages. Given the 0.0013% failure rate, which results in a minor inconvenience this sounds like a VERY good system.

Re:Sounds Like a Good System (1, Insightful)

RazzleFrog (537054) | more than 3 years ago | (#36798706)

âoeWe send out 1,500 suspension letters every day,â(TM)â(TM)

"Neither the Registry nor State Police keep tabs on the number of people wrongly tagged by the system. But Gass estimates in his lawsuit that hundreds might have received revocation notices in error since the system was installed."

That's a DAY. Do you really believe that only 2 of them are problems? Most are probably just fixed without people making a big stink because they are used to the government screwing up.

Re:Sounds Like a Good System (1)

Gr33nJ3ll0 (1367543) | more than 3 years ago | (#36798802)

Right above that sentence it says

Last year, the facial recognition system picked out more than 1,000 cases that resulted in State Police investigations, officials say.

I sincerely doubt they're sending out 1,500 a day, it's probably either a misquote, or the total for ALL reasons. If it was true, that would mean that 390K people would get these every year, or about 10% of all Massachusetts drivers, not including all the people who have their license revoked for other reasons, like DUI, etc. Doesn't seem like a valid statistic to me. Further the guy filling the lawsuit is estimating (to his benefit) hundreds over a period of 5 years, so maybe tens per year. (In state with 4.5 million licensed drivers) So yeah, still sounds like a pretty darned good system to me.

Re:Sounds Like a Good System (2)

Smallpond (221300) | more than 3 years ago | (#36798772)

They don't give numbers for Mass but they do for NY.

New York detected roughly 3,500 instances of possible fraud, resulting in 600 arrests since a system was adopted in 2010.

Looks like about 80% fail rate to me.

Re:Sounds Like a Good System (1)

Gr33nJ3ll0 (1367543) | more than 3 years ago | (#36798830)

Depends on whether or not those people were actually run through the system or the computer flagged it, and the operator said "No". Also, just because they weren't arrested doesn't mean that those people weren't perpetrating fraud, those were just the people they were able to track down, and arrest.

Re:Sounds Like a Good System (1)

Smallpond (221300) | more than 3 years ago | (#36798848)

Sure. And it also depends on whether "detected" is the ones left after throwing out all of the obviously wrong results. The numbers could be off either way.

not to me (1)

decora (1710862) | more than 3 years ago | (#36799056)

they spent millions of taxpayer dollars for a system that has no proven benefit, all it does is further erode civil liberties which 'our brave men and women in uniform' are said to have died to defend.

in defense of intrusive bungling bureaucracy (3, Funny)

circletimessquare (444983) | more than 3 years ago | (#36798666)

Massholes do all drive in the same aggressive manner

(I keed, I keed!)

Re:in defense of intrusive bungling bureaucracy (2)

dkleinsc (563838) | more than 3 years ago | (#36798884)

The difference between a Boston driver and a New York driver: The New York driver takes a right turn from the left lane at 45 mph honking and giving you the finger. The Boston driver does the same thing, but is also drinking coffee, reading the paper, and talking on his cell phone.

Pictures or it didn't happen! (1)

Henriok (6762) | more than 3 years ago | (#36798690)

Nuff said!

Problem with face recognition (3, Informative)

sifi (170630) | more than 3 years ago | (#36798696)

I was looking at facial recognition algorithms some while back, the problem is you get too many false positives.

The problem with all of these algorithms is that it doesn't matter how accurate they are, they are only ever going to be a way to reduce the search space - you should never base a decision solely on the algorithm telling you "this is person X".

For example, some sales person says "Hey I've got this great facial recognition software it is 99.99% accurate!" (that's better than most facial algorithms out there) sounds pretty good right! - Wrong!. Suppose you set it up to look for one terrorist at Heathrow airport. The system is likely to flag up 650 000 people a year (based on 65 million passengers a year); of course it gets even worse if you start looking for more people.

Re:Problem with face recognition (2)

data2 (1382587) | more than 3 years ago | (#36798882)

Except that it is only 6500 people a year, you are right.

Re:Problem with face recognition (1)

sifi (170630) | more than 3 years ago | (#36798984)

Maybe I'll try a calculator next time.... :-)

Re:Problem with face recognition (1)

JasterBobaMereel (1102861) | more than 3 years ago | (#36798886)

Give it 10,000 people to look for (Which the US No Fly list will soon have) and it could flag everybody as someone on the list ...

Re:Problem with face recognition (1)

Attila Dimedici (1036002) | more than 3 years ago | (#36798936)

Almost all of these systems are set up on the opposite principal to the basis for our justice system. The U.S. justice system was established on the principal that it better to fail to identify the guy you are looking for than to identify someone else as that guy. These systems are set up on the assumption that it is better to get the wrong person occasionally than to miss the correct person. This is part of why security systems (not just the computerized ones) are tending to get out of hand. It is complicated by the fact that they do not want to profile for fear of being accused of bias.

Re:Problem with face recognition (1)

MindStalker (22827) | more than 3 years ago | (#36798988)

"Suppose you set it up to look for one terrorist at Heathrow airport. The system is likely to flag up 650 000 people a year (based on 65 million passengers a year)"

Eh, probably not, as many of those people are repeat visitors, and the guys who you arrested because they looked like your terrorist are unlikely to come back and try again :) //And of course if your terrorist is non-white you can reduce your search space significantly on that as well.

This Is Why We Need RFID (0)

Anonymous Coward | more than 3 years ago | (#36798700)

And now you see how mandatory RFID on every person is going to protect citizen rights, avoid bureaucracy, and generally help the peace of mind.

/ runs off giggling in a zig-zag pattern

Bayesian statistics (5, Interesting)

denoir (960304) | more than 3 years ago | (#36798738)

I work for a company that develops neural network software which is used for face recognition on a number of airports. The problem we've had over and over again is that government officials and airport security personnel have great difficulty understanding some elementary statistics.

Let me give you an example. One version of the software offers 99.99% accuracy (symmetrical true positive and true negative), a number that always seems very impressive to various officials.

What they don't understand and what we have to remind them all the time is that they need to take into account the large number of faces that are scanned by the software and that a 0.01% false positive rate isn't something you can ignore.

For instance in a large airport that has say a million people getting scanned yearly it means that 100 people will be incorrectly flagged by the system. The prior probability that a traveler is a 'person of interest' is less than 1/100,000. Plugging the number into Bayes' theorem you get that when the system flags a passenger, the probability that the passenger was actually a person of interest is around 9%.

The officials typically only listen to the 99.99% figure and ignore the reality of the relatively large numbers of false positives when dealing with huge numbers of people. Subsequently they treat the people the systems flag much worse than they would if they realized that the probability of a 'catch' being correct was less than 10%. We've done our best to try to educate them but usually they don't want to listen as it's an uncomfortable truth and it's much more convenient to say that the system has an accuracy of 99.99%.

Re:Bayesian statistics (5, Insightful)

captainpanic (1173915) | more than 3 years ago | (#36798758)

If the 99.99% figure is apparently misleading, and if the 99.99% figure is apparently the only one that the politicians look at, stop presenting the 99.99% figure!!!

Re:Bayesian statistics (0)

Anonymous Coward | more than 3 years ago | (#36798810)

But then they wouldn't sell any software.

Re:Bayesian statistics (4, Insightful)

djmurdoch (306849) | more than 3 years ago | (#36798822)

The 99.99% figure is the only one that is reliable. The 9% figure depends on things that vary over time outside the control of the company selling the software, i.e. the proportion of true terrorists (or other true targets) in the passenger stream.

The problem is the education system, that doesn't teach even basic numerical reasoning to people who need to use it all the time.

Re:Bayesian statistics (1)

martyros (588782) | more than 3 years ago | (#36798954)

The problem is the education system, that doesn't teach even basic numerical reasoning to people who need to use it all the time.

I wouldn't call the Base rate fallacy [wikipedia.org] basic numerical reasoning. I didn't learn about baysean probability until my senior year, and I didn't hear about the base rate fallacy until I was a researcher.

Maybe a better technique would be to say there's an "effective true positive accuracy", which varies depending on the circumstances. Maybe have it as a feature of the software, that you have to enter in certain settings (i.e., population of US 300m, number of suspects searching for: 1000), and then when you get a positive, under the match flag up the effective accuracy. (E.g., "Match! Effective accuracy: 2.3%, based on population of 300m and search space of 10000). That would make it clear to everyone operating it that even though you have a match, the person you're talking to is *potentially* a terrorist, but *probably not*.

Re:Bayesian statistics (0)

Anonymous Coward | more than 3 years ago | (#36799068)

The problem is the education system, that doesn't teach even basic numerical reasoning to people who need to use it all the time.

I wouldn't call the Base rate fallacy [wikipedia.org] basic numerical reasoning. I didn't learn about baysean probability until my senior year, and I didn't hear about the base rate fallacy until I was a researcher.

Basic math should still be able to tell you something - if you have a .9999 accuracy, and scan 50,000 people a day, you'll see ~5 errors (either false positives or false negatives) per day. Simple multiplication right there.

Re:Bayesian statistics (1)

captainpanic (1173915) | more than 3 years ago | (#36799080)

The 99.99% figure is the only one that is reliable. The 9% figure depends on things that vary over time outside the control of the company selling the software, i.e. the proportion of true terrorists (or other true targets) in the passenger stream.

You're a typical scientist. You have 1 figure which is always reliable and true, but it is misleading. The other one depends on too many variables but is a lot more insightful. The scientist will choose the first misleading figure, all other people would go for the second. Adapt to your audience: present a simple case in which you can calculate the useful figure.

The problem is the education system, that doesn't teach even basic numerical reasoning to people who need to use it all the time.

Changing the educational system will only give results in 20-25 years, if not even more. That doesn't mean we shouldn't change it... but it does mean it cannot be used as a part of the solution to our immediate problem. You have to find another solution.

Re:Bayesian statistics (1)

Jason Levine (196982) | more than 3 years ago | (#36799152)

In defense of the politicians, they aren't as bad at math as it might first appear. Remember, they're looking at the systems and factoring in the money that will be added into their political campaigns. Approving the systems might also multiply their opportunity to land cushy lobbying jobs after they leave office. As a bonus, approving this subtracts any chance that their opponents could call them weak on terrorism. Thus, these systems get their undivided support.

Re:Bayesian statistics (0)

Anonymous Coward | more than 3 years ago | (#36798846)

Are you saying that figures based on statistics and percentages are relative?

Re:Bayesian statistics (2)

TheTyrannyOfForcedRe (1186313) | more than 3 years ago | (#36798946)

I work for a company that develops neural network software which is used for face recognition on a number of airports. The problem we've had over and over again is that government officials and airport security personnel have great difficulty understanding some elementary statistics.

Let me give you an example. One version of the software offers 99.99% accuracy (symmetrical true positive and true negative), a number that always seems very impressive to various officials.

What they don't understand and what we have to remind them all the time is that they need to take into account the large number of faces that are scanned by the software and that a 0.01% false positive rate isn't something you can ignore.

For instance in a large airport that has say a million people getting scanned yearly it means that 100 people will be incorrectly flagged by the system. The prior probability that a traveler is a 'person of interest' is less than 1/100,000. Plugging the number into Bayes' theorem you get that when the system flags a passenger, the probability that the passenger was actually a person of interest is around 9%.

The officials typically only listen to the 99.99% figure and ignore the reality of the relatively large numbers of false positives when dealing with huge numbers of people. Subsequently they treat the people the systems flag much worse than they would if they realized that the probability of a 'catch' being correct was less than 10%. We've done our best to try to educate them but usually they don't want to listen as it's an uncomfortable truth and it's much more convenient to say that the system has an accuracy of 99.99%.

Stop selling them you shit software!

Think a minute before you serve up the tools needed for a authoritarian police state on a silver platter. Do you want to be responsible for the destruction of the last remaining bits of freedom in America? No? Then why are do doing it?

draper? batelle? (1)

decora (1710862) | more than 3 years ago | (#36798968)

you work on the Future Attribute stuff?

Re:Bayesian statistics (1)

Timmmm (636430) | more than 3 years ago | (#36798974)

You should just have your software flash a big sign saying "Probability of match: 9%". I don't see how they could ignore that.

Re:Bayesian statistics (1)

vlm (69642) | more than 3 years ago | (#36799024)

The prior probability that a traveler is a 'person of interest' is less than 1/100,000.

That high? Maybe in the general population 1 in 100,000 are being oppressed by the govt. Think about it, if you know that you'll be oppressed, you wouldn't go to the airport...

So either most of the people being oppressed are innocent thus not expecting to be oppressed, or are too stupid to accomplish anything evil, in which case they're also irrelevant.

Is it even possible to accomplish anything with a system like that?

The other problem with 1 in 100,000 being oppressed is that my local "big" airport processes about that many people per day. And there's a lot of days per decade. But much less than one 9/11 per decade. So, almost all of the people being oppressed basically almost never do anything wrong. So why bother offending and angering everyone?

Re:Bayesian statistics (1)

Bob the Super Hamste (1152367) | more than 3 years ago | (#36799102)

What is needed is the matrix that shows the number of true positives, true negatives, false positives, and false negatives. Additionally I would like to see some numbers not just percentages. I would start with an overestimate of the number of terrorists in a population a good overestimate would probably be that 0.001% (reality is it is probably a couple of orders of magnitude lower but lets over hype the threat) are terrorists and then pick a reasonable population size that is in the same ball park as the application they are using it for. Then show the number of innocent people detained, questions, and treated like dirt so that there is a statical chance (lets be reasonable and say 25% instead of some value approaching 100% or even 50%) of it catching just one terrorist.

Good. (0)

Anonymous Coward | more than 3 years ago | (#36798794)

Maybe it'll end up costing the state many, many millions more for this "security".

That is what scares me about working in this field (0)

Anonymous Coward | more than 3 years ago | (#36798868)

The algorithms are not perfect and there is no doubt that there will be false positives. I would not expect anybody to believe these algorithms 100% but use them to pull up red flags and to start an investigation. How stupid could someone be to not make sure that this was accurate? Oh right these are cops, that explains everything.

tear him for his bad verses (1, Interesting)

Anonymous Coward | more than 3 years ago | (#36798952)

Third Citizen

        Your name, sir, truly.

CINNA THE POET

        Truly, my name is Cinna.

First Citizen

        Tear him to pieces; he's a conspirator.

CINNA THE POET

        I am Cinna the poet, I am Cinna the poet.

Fourth Citizen

        Tear him for his bad verses, tear him for his bad verses.

CINNA THE POET

        I am not Cinna the conspirator.

Fourth Citizen

        It is no matter, his name's Cinna; pluck but his
        name out of his heart, and turn him going.

Third Citizen

        Tear him, tear him! Come, brands ho! fire-brands:
        to Brutus', to Cassius'; burn all: some to Decius'
        house, and some to Casca's; some to Ligarius': away, go!

        Exeunt

- from Julius Ceasar

mod 30wn (-1)

Anonymous Coward | more than 3 years ago | (#36798986)

We strongly urge good maaners ME! It's official deeper 1nto the hot on the heels of and the bottom

Brought to you by (1)

Reverand Dave (1959652) | more than 3 years ago | (#36798990)

The leading repliblican presidential candidate Governor Mitt Romney. Oh the Irony.

first they grope us, now they give us facials (0)

decora (1710862) | more than 3 years ago | (#36799006)

the TSA has gone too far this time.

usa pussies (0)

Anonymous Coward | more than 3 years ago | (#36799122)

Land of the free home of the brave HA HA

Load More Comments
Slashdot Login

Need an Account?

Forgot your password?

Submission Text Formatting Tips

We support a small subset of HTML, namely these tags:

  • b
  • i
  • p
  • br
  • a
  • ol
  • ul
  • li
  • dl
  • dt
  • dd
  • em
  • strong
  • tt
  • blockquote
  • div
  • quote
  • ecode

"ecode" can be used for code snippets, for example:

<ecode>    while(1) { do_something(); } </ecode>