Face-Recognition Software Fingers Suspects

eldavojohn writes, "In Holyoke and Northampton, Massachusetts, the police have a new member on the team. It's facial recognition software that will mine the 9.5 million state license images of Massachusetts residents. From the article: 'Police Chief Anthony R. Scott said yesterday he will take advantage of the state's offer to tap into a computer system that can identify suspects through the Registry of Motor Vehicle's Facial Recognition System.' The kicker is that this system been in use since May and has been successful." An article from Iowa a few weeks back mentions that software from the same company (Digimark) is in use to catch potential fraud in applying for driver's licenses in Alabama, Colorado, Kansas, Massachusetts, Oregon, and Texas. But offering the software and photo database as a resource to police departments raises the stakes considerably. I wonder what the false positive rate is.

Comments

finger suspects ?

[Monsieur_F]

$ finger suspect
finger: suspect: no such user.

$ finger suspects
finger: suspects: no such user.

Re:finger suspects ?

[smittyoneeach]

You, for one, have clearly ventured into Soviet Rootkit territory: your new PCI overlord welcomes you!

False positive rate?

[Anonymous Brave Guy]

I wonder what the false positive rate is.

Speaking as someone with (a) some common sense and (b) a formal CS background including image processing work, I think it's fair to say that it won't be zero.

I hope they have good procedures in place to immediately drop any proceedings against those who are misidentified, and that any automatic identification using this system is not somehow considered 100% reliable in court.

Re:False positive rate?

[Adam Zweimiller]

Yeah, but futhermore wouldn't it be safe to say that they don't just go indict someone with charges without a gumshoe comparing the photographs themselves? I mean, theres go to be some sort of human involvement. Lets say the have a CCTV image of a buglary suspect and they use this software to scan the DMV photos for a match, and the software returns 1 or more matches. They don't just throw the match(es) in jail right then. I think its a safe bet that law enforcement would use their own peepers to compare the DMV photographs with the CCTV to see if its close, and then go about questioning the match(es) for their whereabouts..etc..looking for other evidence before going ahead with prosecution. It's obvious that this system is meant to give leads rather than 100% solve cases. Sure there are going to be false positives, it's a computer look for matches. It's more than likely that it's designed to be liberal with its matches simply to give detectives a list of a dozen possible suspects rather than the entire population of a city/town etc. Regardless, I can't say I'm entirely surprised that a slashdot editor took this chance to stir the pot on something that for the most part is cool, useful, and manages to assist law enforcement without trampling our privacy.

Re:False positive rate?

[PieSquared]

I'd hope this would be used as much to rule out suspects as to convict a specific one. And even if it were a not un-realistic .1%... I don't think facial recognition alone would do someone in. But this could certainly help, even if alone it wouldn't convict.

Re:False positive rate?

[snark23]

It won't be zero, but it also can't be very high or else it wouldn't be cost effective for the police. Assuming that it takes a non-trivial amount of human time to process each positive, a high false-to-true positive ratio would be a show-stopper.

Re:False positive rate?

[trianglman]

The issue is more than just the false positive rate. The problem is that they are going through the entire DMV records. As it stands right now, most places can only go through previously arrested people for things like fingerprint and facial matches, which is something that comes with having a record. I, as a law abiding citizen on the other hand, should not be immediately thrown under suspicion just because my face is somewhat similar to a blurry CCTV image, which is what the false positive rate could cause. I have a job that requires me to be in a certain place at a certain time, thats not exactly possible if I am being held for questioning because of something someone I have never met did something on the other side of town. If I could trust our government to use new technologies judiciously and with restraint, it wouldn't be a problem, but this hasn't ever been the case and, short of some utopia suddenly appearing, probably never will.

Re:False positive rate?

[theeddie55]

People seem to have the wrong idea here, I highly doubt that a system like this would examine a photo and say this is your criminal, it's more likely to be used to narrow down searches from... well... everyone, to maybe giving the top 50 matches or something to work with. Sounds like a great idea to me.

Re:False positive rate?

[nbauman]

>I wonder what the false positive rate is.

It will be like the Do Not Fly list.

Years ago, Scientific American had a story about a prototype system for facial recognition created of students at Brooklyn College. They had a database of about 1,000 faces, and they showed the 2 most similar and the 2 most different. The 2 most different were very different. The 2 most similar were so similar, I couldn't tell them apart. So back-of-the-envelope, I'd say about 2 faces in 1,000 will be so similar you can't tell them apart.

(Surely on Slashdot somebody must know the current research.)

So if there are 300 million people in the U.S., and you have a common-looking face, you'll have a close match to 300,000 people. Or 8,000 people in New York City.

(In New York a popular Catholic priest was arrested, based on a victim's identification, and charged with rape. His parisioners couln't believe it. Finally the cops found another guy, and charged him with the rape. Finally, they found a *third* guy, and he seemed to be the one. The newspapers published the 3 pictures. They really looked alike. Funny thing was, they were different racial types, too. One guy was hispanic, another guy was Italian.)

Re:False positive rate?

[bob65]

Well getting lies from random people on the streets would be analogous to getting a false positive from this system (except that the system arguably isn't out to frame anyone on purpose), and since police have no trouble dealing with such false positives from "tips", they should have even less trouble dealing with false positives from this system.

Re:False positive rate?

[Duhavid]

I think you are assuming that the police are not having a problem
dealing with false positives from "tips". I suspect that is not
proven.

Re:False positive rate?

[kaufmanmoore]

Police seem to be having a problem with "tips" from that stripper in the Duke Lacrosse case

Re:False positive rate?

[Brandybuck]

It's funny how people are crying over this invasion of their rights. Fine. The police are horrible people. I'll accept that for the sake of the discussion.

But consider the opposite side of the equation... if you are able. What if the system was only 50% sucessful? Isn't that at least a high enough success rate to send out a cop to personally ID the guy? What's the difference between this system, and some old lady down the street calling in to the cops saying she recognized the guy from the photo in the post office?

Or do we just cease trying to catch criminals, just because we might make a mistake in identification and cause someone an afternoon of grief? Do we really need to jump of the far left deep end in order to be good people?

Re:False positive rate?

[KKlaus]

I, as a law abiding citizen on the other hand, should not be immediately thrown under suspicion just because my face is somewhat similar to a blurry CCTV image, which is what the false positive rate could cause.

That's not really a valid complaint in and of itself. The system already works with you being a suspect for looking like whoever committed the crime. That's what wanted posters are about, what "have you seen this man" questions are about, etc. It's not like criminals pose for cameras, so using imperfect pictures of them to find them will always involve false positives, even if its only humans involved. So the fact that said system is receiving some computer aid to help LEO's isn't bad in and of itself.

What would make it bad would be if it were abusive and wildly inaccurate. A "calling all cars, be on the lookout for a black male between 4'9" and 6'5", ages 18-35, weighing 100 to 350 pounds." If the system is used fairly, and police understand that people the system fingers are fairly likely to be law abiding citizens and should therefore be treated with courtesy and respect, I think it will work fine.

Re:False positive rate?

[NormalVisual]

If the system is used fairly, and police understand that people the system fingers are fairly likely to be law abiding citizens and should therefore be treated with courtesy and respect, I think it will work fine.

Well, there's your problem right there. Sometimes it's not about justice, it's about the police arresting someone and the DA getting a conviction for political gain, or just so that they don't look like a bunch of fools. The justice system in the US is supposed to operate on a presumption of innocence, but it doesn't always work that way in real life.

Re:False positive rate?

[KKlaus]

Sure I guess my point was that if the police want to throw you in jail for looking like someone who committed a crime, they don't need some database of license photos to do it. They already have that capability.

Re:False positive rate?

[mdfst13]

"any automatic identification using this system is not somehow considered 100% reliable in court."

They already have systems like this for fingerprints: local police send fingerprints to the FBI; the FBI puts them in computer; computer spits out a possible match or matches; a real person then looks at the submitted fingerprint and the stored fingerprint and makes a decision. If it goes to court, a real person who is locally available will testify as to the match (well first, you can see both have a tented arch here...).

It's also worth noting that this is probably aimed more at finding suspects than at convictions. I.e. a match starts an investigation. If you're already investigating someone, you don't need to compare security photos to their license photo--you can just compare to the actual person. I would be very surprised to see the results of this used in court.

Re:False positive rate?

[dekkerdreyer]

"Round up twice the usual suspects!" - Louie Renault

Re:False positive rate?

[Znork]

"It won't be zero, but it also can't be very high or else it wouldn't be cost effective for the police."

You're assuming they want to nail _the_ perpetrator, not _a_ suspect. Consider the number of times that US prosecutors have actually opposed conceivable exonerating DNA tests even for convicts on death row, and you might not think a high false positive rate would be a showstopper at all.

From what I've seen of facial recognition software, the error rates are horrible. Set it to sensitive and you get error rates in the percentages, set it less sensitive and you fail the matches all the time. That's ok for building entry systems where you have a small sample to compare against and the person wanting to enter can try repeatedly with varying angles and light, but it's absolutely crap when you have a large database matched against images taken under dubious conditions.

Re:False positive rate?

[cheater512]

Police dont walk around blindfolded. I'm sure they make their own decision.
This would just help them narrow down the possibilities.

Re:False positive rate?

[shmlco]

Yes, but what if you work in some area, like DC, where they only way to get to work every day is to drive past a battery of these cameras? And you're stopped, daily, because the system says so?

Which is the difference between then and now. Then, someone COULD have done it true, but statistically, the odds of it happening were slight. (Cop being there, you being there, cop noticing you and running a check, dozens of other people not being checked because cop was checking someone else, cop not munching a donut, and so on.)

Whereas a system like this checks everyone, everytime.

And if such a system throw false positives, then it WILL have an impact. Especially if the "match" is against, say, an escaped felon, "armed and presumed dangerous."

Re:False positive rate?

[shmlco]

"Isn't that at least a high enough success rate to send out a cop to personally ID the guy?"

If the "match" is against, say, an escaped felon, "armed and presumed dangerous," I strongly suspect the cops are going to do just a bit more than saunter up and ask for your id...

Re:False positive rate?

[Troed]

Exactly, and this happened only a few weeks ago.

Innocent Girl Held A Week In North Platte Jail

Click the link - look at the photographs. This error was made by humans - the actual witness himself. Now imagine what image processing software would think.

http://nebraska.statepaper.com/vnews/display.v/ART /2006/11/02/454a32dc2d927

Re:False positive rate?

[AndroidCat]

Simple fact dooming this program: Who looks like their licence photo?

Re:False positive rate?

[Anonymous Brave Guy]

Blockquoth the AC:

Then (and this is the "well DUH" moment so listen carefully) a human being can visually check the footage and the licenced photos, and considering it's a police investigation, do the rest of the evidence gathering required to make a case against a suspect.

That's lovely, and as long as this sort of system is used as a way to make the human checkers' job easier and not an excuse to remove human checking from the system that's great. The problem is that systems like these often do get used to remove the human element. That does, after all, make systems more efficient and reduce costs, as long as you don't notice the people who get shafted due to false positives/negatives.

I live in the UK, where automatic number plate recognition (ANPR) technology, which is effectively OCR for car licence plates, is finding a lot of friends among the authorities. It is increasingly used in speed cameras that measure your average speed over a distance. It is used in supermarket car parks to check people aren't staying too long. It is used to monitor those entering the congestion charging zone in London.

The problem is that whatever checks are supposed to be there to prevent a machine screw-up becoming a false charge demonstrably don't work reliably in practice. There is a man who lives near me, quite elderly, who barely drives at all and hasn't been to London in many years. He has received not one but multiple fines for going into the London congestion charging zone without paying. (Yes, he has provided plenty of proof that he really was elsewhere at the time.) Perhaps this is down to some arsehole cloning the guy's number plate, and insufficient checks being made on the make and model of the vehicle to realise this. Perhaps it's down to poor scanning by the OCR software, which tends to read a 5 as an S from some angles, or doesn't understand some foreign plate formats. It doesn't really matter what caused it: the point is that the process has advanced from a flawed automatic diagnosis to sending formal penalty notices to an innocent old man, with insufficient checks and balances in between to spot the mistake.

Exactly the same risk exists with any automatic recognition system, including the facial recognition technology we're discussing here. Would anyone like to bet me that in a couple of years' time, the senior police officers in the area won't be on the news, proudly announcing the way they've both cut costs and reduced the number of licenses wrongly awarded thanks to a new system that immediately rejects anyone whose face matches (while quietly ignoring the increase in people wrongly rejected by the same process)?

This is where we must demand more from our authorities. Mistakes will always be made in real life, though of course we should try to minimise them, but what really counts is how fast you fix them, how easy it is for a wrongly accused party to clear their name, and how little distress is caused to them as a result.

Re:False positive rate?

[MasterMnd]

I agree with this line of argument, but this case may be a bad example. This is actually a case where Facial recognition software would've been able to better identify the differences between the 2 faces. Look carefully at the 2 faces themselves, their heads are shaped completely differently, different noses, birthmarks, freckles, different ears, different eye spacing, even eye color looks quite a bit different. I expect these things would all be noticed by facial recognition software. I imagine that humans use different cues to recognize similarity then the computer does. We see similar eye shape, hair color ans style, skin color, and jewelery, and our brains fill in the rest.

Unfortunately facial recognition software wouldn't have helped in this case anyway since there was no photo of the original suspect.

Does all this mean that the false-positive rate of the software is acceptable? No, it just means that in some cases it could be less then human comparison alone. Or, even better, facial recognition used in addition to human comparison could probably reduce false positives. But that ignores any issues with automatically comparing photos with a large database.

Re:False positive rate?

[P3NIS_CLEAVER]

You guys have totally missed the point. This is a good thing for privacy.

They are doing this to discourage driver licence fraud. In many states we have had to fight to get SSN taken off of our DL... so how does is the DMV supposed to discourage fraud? If DL fraud is rampant, this gives ammunition to proponents of a national ID, with the argument that the states are unable to identify people corretly.

Another side to the issue are the people that have been a victim of identity theft, I am sure they will be happy that the DMV is doing this.
What does this have to do with being considered a criminal? If your licence is flagged they are going to give the application more scrutiny. Wouldn't you be happy if it was another face trying to get a licence in your name?

Re:False positive rate?

[coaxial]

That brings up an interesting point. You're right to say that cops filter out false tips from people, but these "tips" aren't coming from a person. They're coming from a machine. That's an important distinction. There could be an bias (concious or unconcious) to believe the machine. The machine is "state of the art," "highly complex," and most importantly, "scientific." If it gives you good tips, then one may become inclined to trust it when it gives less obviously good results. Cops already use with scientific methods that they may or may not totally understand how they work. Fingerprints, trace evidence, and now DNA. This is just another tool. However, without the experience of what it can and can not do, one may give it an undeserved biased towards believabilty.

I'm not saying the cops are or would become slaves to the machine as it were, but it's the border cases I'm more interested in.

Re:False positive rate?

[Entrope]

There is no false positive problem. Police said so on TV!

Explanation: I was flipping through channels and landed on one that was following a Los Angeles SWAT team. They were serving a warrant on an apartment based upon a tip from a confidential informant that there was a gun in the apartment (and I think that drugs were sold out of the apartment -- I missed the start). Nobody answered the door, so the SWAT team battered down the front and back doors, broke a window to investigate a room that was locked, and ransacked the entire place. They found no drugs and exactly one "weapon": a pistol-shaped BB gun. The conclusion from the SWAT team leader: that the confidential informant had proved his worth *and* that their destruction of the apartment had shut down a drug distribution center.

I wish I were making this up. Sadly, this is probably typical of cases that Radley Balko has cataloged.

Re:False positive rate?

[Entrope]

You seem to be assuming that the false matching rate will be reasonably low. Unless both your source and target images are extremely high quality, if you want 90+% matching accuracy, you will detect 30+% of other faces as being in your template database. As you add more mug shots to the template database, the false alarm rate will go up.

Think how long you wait in line at the DMV now. Do you want the person behind the counter to spend an extra five, ten, or thirty minutes vetting the past and proof of identity for half the people in front of you? Do you want to be one of the people who are "guilty" of resembling some criminal somewhere in the state and who is harassed for that accident of nature?

Re:False positive rate?

[P3NIS_CLEAVER]

Yes, but the people you are falsely matched with most likely still have their licences and have no reason to try to created one falsely. If you assume that only 1% of people lose their licences, your false 'positive' rate will be much less.

I know that this doesn't help with willful identity theft, but most people apply for false licences because they have lost their licence.

Re:False positive rate?

[JAppi]

Facial recognition software wouldn't help much here since they didn't have any photos to check the rapist against. I don't think the victim should be blamed for picking out the wrong suspect. It's one of those things that you forget and can't remember the face, just because you want to forget the whole thing.

Re:False positive rate?

[cayenne8]

"Ultimately it's up to the police to figure out who the suspects are, and using this tool is sort of analogous to receiving a bunch of "tips" from "witnesses" that point to a group of potential suspects."

Trouble is..to the general populace....the computer is infallible...much more correct than a person could be...people really beleive that in droves out there.

Much like they all believe that Doctors know it all...give them complete 100% trust...not knowing that the Dr's largely are making only very educated guesses in a world of science that is far from exact...

With this in mind...if the computer 'convicts' you...many if not most of the people will believe it!

Re:False positive rate?

[Entrope]

You misunderstand the scale of the problem. I will use Virginia as an example, since it is my home state. Virginia has seven million residents, of which perhaps four million drive. With 1% in the database, that means the template database will contain 40,000 faces. Even with my unusual hair color, I am certain that I closely resemble several people in a random sampling of 40,000 other Virginians.

It is worse than that, though: The database is not populated only from those with suspended or revoked licenses. It includes convicted felons. (Side question: If a conviction is old enough to be sealed under law, how does the ex-felon know all these databases have been updated?) According to at least one estimate, there are 200,000 Virginia residents who have been convicted of a felony at some point in their life. According to the Virginia Criminal Sentencing Commission, there were 22,020 state felony convictions in 2005 alone, so a total of 200,000 sounds reasonable. If you include those living in neighboring states or those convicted in federal cases, the number goes up several times.

Re:False positive rate?

[nbauman]

That's a good one.

The state department of motor vehicles license photo database raises an interesting opportunity.

If I were a defense lawyer, and my client were being charged on the basis of eyewitness testimony, I'd go to the license database and pull out the 12 drivers who most resembled my client. Then I'd line the photos up and ask the witness to pick out my client. Unless my client's face was very unusual, the witness wouldn't be able to do it.

(The Innocence Project at Cardoza College used DNA testing to find people who were convicted of crimes but were really innocent. They said that the most common reason for false convictions was mistaken eyewitness identification. Scientific studies show the limitations to eyewitness testimony, but defense lawyers can't use that in court.)

Re:False positive rate?

[Brandybuck]

So you're saying we shouldn't even try to catch the armed and presumed dangerous felon?

Re:False positive rate?

[nbauman]

You're making a good point which is well known to people who do the statistics for medical tests.

If you have a witness look through a database of people who were arrested for a similar crime in the same neighborhood -- say, 1,000 people -- then the chances of a false match are *relatively* low, because you've only got 1,000 people to choose from.

But if the witness looks through a database of all drivers license holders in the entire state -- say, 10 million people -- then the chances of a false match are almost certain. If 1 person in a thousand looks like me, which is realistic, you'll get 10,000 matches. It depends on how close a match you require, but eyewitness identification is not too close to start with.

Then all the cops have to do is go through the list of 10,000 suspects and find the first one who doesn't have a good alibi and can't afford a good lawyer.

Re:False positive rate?

[Renraku]

They should test these in Britain first. They seem to like being watched.

That way when a camera sees someone flip off someone else, or walk out of a shop a little too quickly, his summons can be printed and mailed within the hour! Saves them the trouble of having to feel a tiny bit guilty for flipping someone off, or wonder if he looked like a thief when walking out of the shop because he was about to be late for work.

Imagine if every stupid thing you done had to be accounted for. Not just to the higher ups in whatever religion you subscribe to, but to the authorities.

"Well, Mr. Anderson, we have you on camera here speeding on a three-mile stretch of wide open road on this sunny day. You were going three over. How do you justify that? The law is the law, and those that break it are criminals. You kissed your girlfriend in public the other day, too. You used tongue. We've got her in the next room interrogating her, so you better tell us everything. Did you two then go on to have sex in public? How is your sex life? Have you been seeing other people? Has this happened before? Also the magazine you bought the other day had an advert fall on the ground when you were walking. We'll add that to your total charges today. Your total will be 400 pounds."

Mr. Anderson sighs, because he knows that it was already deducted from his bank account. The machine prints a receipt and he goes on his way.

Fast forward 10 years, to the civil war.

The photo-recog is used to hunt down people and their families that were suspected of being in the resistance. Those people are tried and shot as being traitors. There's no trial, no chance for appeals, nothing. Just a cigarette, blindfold, and 7.62mm bullet to the back of the head.

Sounds like a good idea, eh? It'll make sure that we're in complete control of the government, at all times. You have nothing to hide, I hope.

Re:False positive rate?

[that this is not und]

Trouble is..to the general populace....the computer is infallible...much more correct than a person could be...people really beleive that in droves out there.

You have evidence to prove this? Because I can remember, going way back, that there is a common belief, often the root of some humor, in the fallability of computers. Going way back to the late fifties there have been jokes and cartoons about ludicrous computer errors. Almost as long as there have been computers, people have understood the GIGO (garbage in, garbage out) phenomena.

Re:False positive rate?

[that this is not und]

If you are 'stopped, daily' three days in a row, the system will throw a flag and your situation will be corrected. It's too expensive to just let it go and waste a 1/4 man-day of police time stopping you ever day. So your scenario is hypothetical and irrelevant.

Re:False positive rate?

[that this is not und]

Think how long you wait in line at the DMV now.

Last time I went to the DMV, I was in and out in less than 15 minutes. And this was on a Saturday morning, when 'everybody is at the DMV.' I was renewing my D.L. and also getting tabs for three different vehicles. It took almost no time at all.

And I live in a state where the news media is always carping and whining about 'how long a trip to the DMV takes.'

Re:False positive rate?

[that this is not und]

The problem most people have with their ID photo is acknowledging that that is what they look like. People carry around preconceived notions about how they appear to others.

That said, in a poll taken at any random gathering of people, they would all reinforce the idea that ID photos are terrible representations of the people they picture.

Re:False positive rate?

[shmlco]

An assumption, not backed by fact. Unless you're the author of the system?

Re:False positive rate?

[shmlco]

Your orignal comment was, "What if the system was only 50% sucessful?" With the implication that 50% of the time it's going to be unsuccessful.

My example simply served to demonstrate the severity of that potential mistake...

Re:False positive rate?

[that this is not und]

Well, I figure everybody here is flinging assumptions this way and that....

Ain't no facts anywhere I can see in much of the diatribe here.

Ewwwww!

[Anonymous Coward]

This might be software that is a lot more popular with the ladies than the gents!

Re:Ewwwww!

[couchslug]

Depend on the gents, thweetie!

Well...

[Anonymous Coward]

Did the software at least buy them dinner first?

Oh yeah?

[slughead]

Face-Recognition Software Fingers Suspects

And what does it do if they're male?

Re:Oh yeah?

[raehl]

And what does it do if they're male?

It fingers them.

Re:Oh yeah?

[Plutonite]

Yes, and I hope we all agree that the word "face" and the phrase "fingers suspects" are uncomfortable in the same sentence.

Wonderful

[iamdrscience]

As a license-holding Massachusetts resident who lives right near Holyoke and Northampton (Amherst) it's nice to know I can look forward to being a criminal suspect in the near future.

Re:Wonderful

[b0s0z0ku]

Eh, up there you're all (ugly) lesbians, filthy hippies or both anyway so who gives a crap?

Nah, more like limousine liberals. The *real* hippies have all moved out to Vermont or Oregon. Cheaper living, more land, cleaner air, and less intrusive government :-P

-b.

False positives before, too

[Wylfing]

I fear "automatic" matching of criminals and trying to catch them, e.g., when they renew their license. Here is a true false-positive story that happened to me. I went to renew my driver's license, and the nice lady informed me that she could not issue me a license because I had had mine revoked in Maryland due to felony charges. Now, I have never committed a felony and I have never been to Maryland, let alone had a driver's license there. The nice lady was unpersuaded by this information. The database said I was a felon in Maryland, and that was the end of the story.

After much yelling about the problem, it was finally revealed that the real felon's name was exactly like mine except for one letter, and some moron doing data entry had gone ahead and decided we were the same person, based solely on name. Since this data problem was local to the "matching" system they had implemented, and not prevalent in who-knows-how-many databases, it was cleared up with a little investigation. However, if that "match" had been replicated into other systems, I could very well have had a nasty time clearing my name. The lady at the DMV was 100% convinced that I was a felon based on what the computer told her. Quite likely, no one else would have believed I was innocent either.

I can see this system playing havoc with people too. I have met people with no connection to each other but who nevertheless look virtually identical.

Re:False positives before, too

[whm]

I can see this system playing havoc with people too. I have met people with no connection to each other but who nevertheless look virtually identical.

This article is a great example of what you've described,

http://nebraska.statepaper.com/pages/drudged/innoc ent.html

In summary: There are two girls that look nearly identical. One of them committed a crime, and the other was put in jail for a week. There are photos in the article.

Re:False positives before, too

[Jason1729]

Maybe it will be for the best if this story becomes the norm. If it just happens in a few rare cases people (and politicians) will ignore the problem. If it happens a lot the system will be abandoned.

Re:False positives before, too

[CodeMasterPhilzar]

I didn't see any mention in the article that a match from the computer system would be admissible in court. It is just a tool for narrowing the suspect list down from "everyone but me" to maybe this one or more people from the database, to maybe someone not in the database...

Re:False positives before, too

[kaufmanmoore]

Having a DUI can affect you getting a license and in NC people who don't pay child support can have them revoked.

Re:False positives before, too

[Skidge]

I had a similar problem renewing my license in NY state. I have a fairly common name and someone with an identical name, down to the middle initial, and I believe even the same birthday had some charges for running a red light and fleeing the scene of an accident in NY City and some violations in Florida, as well. Problem was, I was living as far from NYC as you can get in the state and had only been to NYC when I was 5 years old. Not to mention that I was only 18 and had rarely driven outside of my county. Luckily, my mother knew someone at the DMV and through many phone calls and faxes, it was cleared up.

While my inconvenience was minor, it's scary to think was trouble someone could land in due to data quality issues.

Re:False positives before, too

[value_added]

After much yelling about the problem, it was finally revealed that the real felon's name was exactly like mine except for one letter

Is your name Tuttle, by chance?

Re:False positives before, too

[P3NIS_CLEAVER]

Aren't you glad that the DMV lady raised the concern instead of when you got pulled over by a cop? You should be glad this is what happened.

Re:False positives before, too

[b0s0z0ku]

The jerk-off cops didn't care that they had the wrong person. They still kept her illegally and lied to a judge to keep her in jail even longer.

If they lied to a judge, shouldn't they go to prison for perjury? Lying to secure an arrest or conviction carries the same amount of prison time (or even execution) as your "victim" served in a lot of states. So shouldn't they sit in jail for at least a week?

Better yet, there's always the option of someone finding them in a dark alley and using a baseball bat to rearrange their faces so that children won't ever be able to look at them in the face without running crying to mommy.

-b.

Re:False positives before, too

[b0s0z0ku]

And when did having a felony conviction in one's past having any effect on whether one can get a driver's license or not?

I don't think this applies to all felons. However, it would make sense if you were (say) driving drunk and killed someone with your car. Manslaughter is a felony, and you'd definitely not want that person driving anytime soon.

-b.

Good old SQL...

[eyeball]

Part of the SQL better include something like "... WHERE OCCUPATION IS NOT 'politician' " otherwise there's be total anarchy.

Re:Good old SQL...

[b0s0z0ku]

Why Democrats getting stopped? Because they would be more likely to want to fight the no fly lists and other anti-terrorist measures.

Not really. Members of both parties, these days, with few exceptions are authoritarians. Remember that Democrats were strongly behind the PATRIOT Act and the Dept of Homeland Security right after 9/11. If you want freedom, vote Libertarian or Independent.

-b.

fingers suspects

[Placebo Messiah]

sounds like a Realdoll(TM) upgrade

but no stats

[troll -1]

Sounds promising for law agencies but given that no caught suspects have been named and that criticism persists that face recognition technology is inherently unreliable, I wonder how much of this is just (sales) hype. I mean, come on, give us some real data where you can say it's effective because ... here are the names of the criminals we caught and it can all be credited to the system.

Re:but no stats

[colmore]

There's a fundamental, mathematical problem for any system that screens large populations looking for a small number of targets.

Let's say your system is 99% reliable, that is to say, 1% of the time it checks a negative it reports a positive and vice versa.

Now you screen 1,000,000 people looking for one suspect, your system turns up 10,001 positives. Which one is it?

This is a problem that has been well-studied in cancer screenings. For certain rare types of cancers, there are nearly 100% reliable tests that nonetheless when they report a positive, are usually wrong.

Now it's fine to say, in the case of the cancer, that the 1% of the population should be informed and then checked via another procedure or something. But when we're talking about a process that fingers potential criminals, and in modern criminal justice where merely being a suspect hurts your life in a myriad of ways (god help you if the information winds up somewhere accessible to google, or worse yet, the case has anything to do with terrorism).

I have the same objection to large-scale wiretapping operations, if anything, the human factor there greatly increases the problem.

Re:but no stats

[b0s0z0ku]

Now you screen 1,000,000 people looking for one suspect, your system turns up 10,001 positives. Which one is it?

Worse, what if the answer is "none," since the actual criminal is either not in the database or wasn't recognized? Will the pressure to solve the case become great enough to pick the likeliest suspect out of those 10,001, and try to get a conviction.

In Texas, there was a case where two janitors were the suspect in the killing of a girl. One was Black, one was White. There was no strong evidence either way, so the police said to the Black man, "someone's gotta fry for this, you're the n-gg-r so you're elected." Turned out that neither of them was guilty, but the Black janitor lost 10 years of his life on Death Row. Fortunately, they didn't actually execute him.

-b.

Re:but no stats

[old man moss]

Exactly. And these systems are often tested on databases which have a large variety of faces photographed under identical conditions to get their headline 95% accuracy figure.

I did an experiment with a database of only 20 similar looking people under a range of conditions with humans doing the recognition and the best result was just over 80%.

Re:be careful..

[smaddox]

I don't see how facial recognition systems can ever be "near flawless". Most systems I know of use neural networks to match patterns. Neural networks model the brain, and even humans can't always tell apart two people if you only have a picture of their face. Humans use a lot more than a face to determine who a person is.

Jack-booted thugs

[dlaur]

The next time someone tells me that the slippery slope is an invalid argument, I'm going to slap them.

Re:be careful..

[jonathan_the_ninja]

Let's just hope they're more foolproof than ScumSoft's system in Space Quest 3...good memories.

Next stop...

[GreatWurm]

Coming soon to Toll Booths and ATMs, little brother.

Re:Next stop...

[timmarhy]

the booth with slam shut on you, and armed guards shall come and escort you to a holding cell pending questing. all the while telling you it's for your own good.

I, for one, welcome our new CSI overlords!

[`Sean]

Finally! Inept police departments will be able to solve murders and other heinous crimes using awesome computer graphics in 47 minutes or less...just like on TV!

Enhance...enhance...enhance...

internet renewal is your friend

[bxbaser]

Your picture stays the same.
That coupled with the ravages of solar radiation upon your facial skin after a few years you will never be recoginzed.
You will be unstopable.

Suddenly...

[Anonymous Coward]

A whole lot of moustaches and beards become the new fasion...

Now a "Person of Interest"?

[enormouse]

Two potential issues with automated matching, not just facial recognition are: ending up in files as a 'person of interest', 'subject of investigation' or whatever you want to call it, simply because you were in a potential match. This doesn't look good in deep background checks. And, what would be more troublesome, is spending time defending yourself in a preliminary investigation if authorities begin to rely too heavily on fuzzy matches, unless you track your every move.

Oh goodies

[Anonymous Coward]

Great, yet another unpleasant use of interdepartmental government cooperation. One completely unrelated activity (in this case, driving) being used to gather data for another activity (criminal apprehension). I don't know about everyone else, but when I went to get my License I didn't think "oh swell, this information will be culled through every time the police are looking for someone, criminal or not". Perhaps it is my incessant paranoia, but I don't like the Idea of my name/information being put in a database and constantly culled through looking for criminals unless I've been convicted of actually doing something wrong. As far as I am concerned it treats me as a suspect every time, even though I'm probably not within hundreds of miles of the crime, I don't care whether or not a computer is doing it, I'm still being treated as a potential criminal every time. I would not have so much of a problem with the information from convicted criminal mug shots being gone through (though I suppose even in that case there are some issues, but much less prevalent (those people have been "convicted" of crimes)), but innocent, never convicted or suspected citizens, sounds like things are getting scary to me.

As for the level of trust that can be placed in this system......, I would place it as low at best. The inaccuracies of currently understood facial recognition software aside. The fact that swat teams routinely smash into the wrong persons home, because of a misspelled address or faulty descriptions should clue into that this system would probably trouble a lot of innocent people. I have little doubt that there would be many false positives involving people who looked relatively similar to a criminal who made it all the way up to the "arrest" phase of being a suspect before the police finally discovered it was a mistake. And in a environment where, at least as far as police mistakes/abuse are concerned, treated with a light slap on the wrist, paid leave of absence, or a reprimand on their file are about all the punishment that can be expected, I don't think they need a tool as inaccurate and dangerous as this. If they can eventually learn to use their current tools better (like putting heavy/warranted restrictions on access to DMV info, Phone Records, and Credit Card info) and punish/repair mistakes appropriately. Then maybe they should be allowed equally restricted access to a tool as dangerous as this with the affore mentioned criminal mug shots restriction, but not until. //end of rant

Re:Oh goodies

[Anonymous Coward]

I have a short barreled semi-automatic 12 gauge shotgun and other weapons loaded and ready to kill anyone (police or not) who tries to force their way into my home without showing a warrant first. I realize I would die in the process if it were a cop or SWAT team executing a so called "no-knock warrant", but some things are worth dying for, and the sanctity of my own home is one of them.

Re:Oh goodies

[Anonymous Coward]

Eric - now calm down, you've forgotten to take your meds again, haven't you?

WTF!?~

[sc0p3]

This is exactly like finger printing everyone in the state. Privacy has gone out the window. Making use of photos which people allowed for use on their license, to be used to finger them is criminal.

Re:WTF!?~

[naChoZ]

Interesting. That makes me wonder, could someone go and renew their license yet refuse to have their picture taken citing the 5th amendment.

Re:WTF!?~

[Jah-Wren Ryel]

could someone go and renew their license yet refuse to have their picture taken citing the 5th amendment.

Forget the 5th amendment, I think this ought to be considered a direct violation of the 4th amendment: The right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated...

Searching a database of pictures which were collected under the premise of a different use seems like an unreasonable search of my papers or if 'papers' is considered narrowly as actual paper, my effects.

Re:WTF!?~

[RealGrouchy]

No, it's more of a terms of use problem than a freedom of speech.

When I got my DL photo taken, it was under the understanding that it was necesary for me to get a license to do one particular thing--drive. *I* am allowed to provide that ID in siutations not pertaining to driving in order to identify myself, but I did not grant the gov't to do so, and has not done so.

Now that these gov'ts are starting to use this *driving* ID to identify peopel for other reasons, then it is a violation of the terms of use, and either they will have to get permission from the drivers to use the ID in this way, or they have to give people an opportunity to opt out. (For example, I haven't driven a car in years, and if my DL were turned into an identity card, I would seriously consider cancelling it).

- RG>

That is him officer!

[TavisJohn]

That's the one that programmed me for evil!!!!!!

license photograph archive

[JimBobJoe]

You'd be surprised how many state legislatures never bothered authorizing their respective DMVs to archive the photographs (which is a huge change from the days of the original photo licenses, where only negative was produced and no photograph maintained.)

I just took a look at the MA code and couldn't find anything allowing the photographs to be archived by the registry of motor vehicles. Maybe someone else with a better knoweledge of MA law can find such a law.

This is not an insignificant issue...the archival of the photographs and sharing them to law enforcement, basically without limit and without warrant to access the database, is the practical equivalent of requiring every citizen above the age of 16 to show up at the local police station and be photographed.

I consider the photograph archival of US license pictures to be one of the biggest and least known/understood privacy invasions in the last 10-15 years.

Related to a story on /. yesterday

[isotope23]

http://politics.slashdot.org/article.pl?sid=06/11/ 17/1630255 ' 'But rather than work out these dilemmas in partnership with their elected leaders, they were encouraged to regard all politicians as corrupt or mendacious by the media, which he described as "a conspiracy to maintain the population in a perpetual state of self-righteous rage." Whether media was left wing or right wing, the message was always that 'leaders are out there to shaft you.'"

Obviously they are. A License to drive has just been turned into a permanent mug shot.

Heh

[Anonymous Coward]

"Face-Recognition Software Fingers Suspects"

Oh, that's just great. First face-regonition violates our privacy, and now it's violating our orifices!

Re:be careful..

[NormalVisual]

Humans use a lot more than a face to determine who a person is

Big time. I and lots of other people can often identify someone without seeing them just from hearing them walk. Video would be a big step forward, as people tend to be very distinctive in their movements, but I doubt there will ever be a computer than can people-watch as well as people themselves can.

The amount of comments here, endorsing ...

[quax]

... such a system is truely scary. What's next? How about 24/7 machine assisted surveilance of all telepone calls just because it may help catch a terrorist? Oh, wait a sec =:-0

Once a criminal, always a suspect?

[Web Goddess]

I had one criminal conviction when I was 18. It has dogged me my entire life. It is so upsetting to hear people say, oh well, as long as it's only *convicted criminals* who go into these insane database searches.

It's wrong to mass-search drivers' license pictures. It's also wrong to mass-search pictures of anyone who has ever been convicted of a crime. Many, many people have a regrettable misdeed in their past. It's wrong to continue to punish people who have, as was once said, "paid their debt to society."

Penitent and paranoid in California.

Re:Once a criminal, always a suspect?

[cheater512]

That argument is awfully similar to GMail scanning your email.

Its a freaking computer! It doesnt give a damn about you.

Re:Once a criminal, always a suspect?

[skahshah]

Hangman to convict : "It's a freaking cord ! It doesn't give a damn about you."

Re:Once a criminal, always a suspect?

[ectotherm]

I understand your point, but by the same token would you want a convicted burglar as a house-sitter or a convicted child molester babysitting your children? Trust, unfortunately, is like fine crystal- once broken, it will never be the same. It can be repaired, but the initial breakage will always be there as a reminder. It is unfortunate that you made a bad choice in your youth, and that the results of your choice appears in any background checks done, but you cannot expect to engender 100% trustworthiness with a criminal record. People, while generally forgiving, always think twice about such things. Is this right? Probably not. It this reality? Definitely. I work at a financial institution, and they do a background check that is more thorough than a prostate exam. The slightest blemish on your record will deny employment. They are simply covering their asses in the event of another "Enron." (You mean you hired this person with a known criminal history, and now they embezzled/misled investors!!??)

Re:Once a criminal, always a suspect?

[kannibal_klown]

Exactly how is it "wrong" to mass-search drivers' license pictures? So long as they don't ONLY rely on the computer then it's alright.

• Person A gets caught on tape shooting a grocery store clerk. Unfortunately, that is the only lead.
• While persuing the ordinary investigation (involving the public, looking for witnesses, looking for speeding vehicles), etc...
• the police run the face through the recognition DB and get a number of hits
• They then use ordinary police work to check out the potential suspects: geography, past records, maybe a few house calls, etc.

Now they shouldn't place TOO much faith in the system, after all when searching through a vast DB there are going to be multiple hits depending on how low you set the threshold. But as a tool it can be quite useful to find the guy.

As for your problems, yeh it sucks. But the article isn't saying that only convicted criminals go in: everybody goes in. So how is that punishing you? If you got nailed for involuntary manslaughter during a drunk driving hit-and-run a few years ago, and then a couple of years later it happens again in your neighborhood, you can be sure your name will come up again (even if only for a few seconds) -- with or without the system.

It's just an investigative tool to help find a list of suspects. The software is not good enough to treat it like "fingerprints" or DNA, it's just saying "these guys kind of look like the guy in the grainy black-and-white video tape. They'd need a lot more to charge you.

Re:Once a criminal, always a suspect?

[Anonymous Coward]

Those "thorough" background checks scan data from a multitude of companies and are a result of court runners who are human and can be incompetent. The last background check I had ran on me shows me as a multi-state offender in California, which scares the living bejebus out of me. Unlike a credit report where I can file a dispute, there is no easy way to go and clear this up.

People have misdemeanors show up as felonies, expunged records show up, completely false records, warrants listed that are invalid, and this is just the tip of the iceberg.

If we are to rely on them then they need to be 10000000% accurate, and if an inaccuracy shows up we need to be able to quickly dispute it (resolutions of 3 days because a job or even freedom may hang in the balance).

And to quote my brother who was a guard at an institution in Tennessee.... The only difference between us and them is they got caught.

Either way I'm getting fed up living in a near police state.

Re:Once a criminal, always a suspect?

[b0s0z0ku]

I understand your point, but by the same token would you want a convicted burglar as a house-sitter or a convicted child molester babysitting your children?

The convicted burglar? Depending on circumstances. If he broke into a store because he was starving on the street at age 18, and he's now 40, sure, why not? The child molester should never be let out onto the streets again. If someone's enough of a bastard to rape a child, they should get a mandatory death penalty.

Trust, unfortunately, is like fine crystal- once broken, it will never be the same. It can be repaired, but the initial breakage will always be there as a reminder. It is unfortunate that you made a bad choice in your youth, and that the results of your choice appears in any background checks done, but you cannot expect to engender 100% trustworthiness with a criminal record.

What about things like drug convictions? Some employers refuse to hire anyone with *any* criminal record. Personally, I think that unless someone has proven themselves to be a complete psychopath - i.e. multiple murders, child rape, etc, they can usually be reformed. The brain isn't the same at age 21 as it is at age 30 or even age 25. Also, people do have the ability to least from their mistakes.

And not all crimes are what they seem anyway. When I was 22, I was arrested (fortunately not convicted and even the arrest record was expunged eventually, thanks to a good lawyer I knew) for felony riot. The reason for the arrest? I had told off - but used no physical violence towards - a cop who had broken into a party on a noise complaint and punched my friend in the face for no reason. The cop wanted to "get me back" so he charged me with a felony - even the judge later agreed that that wasn't a reasonable charge.

-b.

Re:Once a criminal, always a suspect?

[b0s0z0ku]

If you've behaved and it's been long enough (rules vary by state ~3-7 years after finishing your sentence including any parole) you might be able to get an annulment.

Vastly depends on the state and the will of the county judge. Plus you need money for various court fees and possibly for an attorney, but if you aren't working you may not have the money.

-b.

Re:Once a criminal, always a suspect?

[RealGrouchy]

Many, many people have a regrettable misdeed in their past.

This is only half the story. The important bit is that only some of them are caught, charged, convicted, and penalized for it.

If *everyone* were punished according to the law for *every* crime/offence they were to commit, our laws, politicians, and voters would be a whole lot different.

- RG>

Re:Once a criminal, always a suspect?

[that this is not und]

If you're a Democrat, you can be an alcoholic without seeking any 'treatment' at all. The police will cover for you. Especially if you're a Kennedy.

The main thing is, all politicians are crooks. The political process promotes crookedness because we give it too much power.

Re:Once a criminal, always a suspect?

[cheater512]

Very different. This is police searching for possible suspects out of millions of photos.
They would prefer *not* to see you (if your law abiding).

Re:Once a criminal, always a suspect?

[b0s0z0ku]

But given the choice between hiring two equally-qualified candidates, one with a criminal record and one without, I'll take the "clean" candidate.

Provided that the record is for non-violent and/or "victimless" crimes, I'd take the candidate with the record. Why? (a) I don't necessarily want blind followers working for me (b) as a "fuck you" to the government (c) because I'd feel like I was doing something good for someone who needed a job.

-b.

Re:Once a criminal, always a suspect?

[Sarisar]

I always thought it was interesting. So you commit a crime and serve time. The justice system considers that you have been punished for this. So assuming everyone is happy with the justice system (no laughing now, please) then it shouldn't matter, because you have been punished.

As everyone seems to think it matters then this means the justice system doesn't work surely...

(just a thought... wonder what other people think)

Re:Once a criminal, always a suspect?

[PopeRatzo]

"WE" give it way too much power? Do you have a mouse in your pocket?

Speak for yourself.

Re:Once a criminal, always a suspect?

[that this is not und]

I would say the same, except people like you are always saying I need to contribute more tax dollars to 'our' mutual interest and empower politicians to do even more.

Re:Once a criminal, always a suspect?

[b0s0z0ku]

There are no "victimless" crimes. That statement is just an accountability dodge.

Growing pot in your basement and then smoking it yourself (or with friends who are of age) is illegal. Who's being victimized?

Laws against sex toys in some states (sale is a felony).

Illegal gun ownership for self defense even if you don't shoot anyone.

Need I go on?

-b.

Twins

[ms1234]

How does it handle identical twins?

Re:Twins

[Penguinoflight]

Send them both to jail and let them sort it out amongst themselves.

Re:Twins

[KillerCow]

How does it handle identical twins?

Innocent Girl Held A Week In North Platte Jail

Turning the law upside down..

[Anonymous Coward]

I'm intrigued that there's so little objection to the gradual reversal of the first principle of law: innocent until proven guilty. In this case it starts on the wholly false premise that the images used on driver licenses are of a sufficient quality to make a proper match without a high false positives rate (you can find the EU specs for biometric passports here - I know that it has turned making a passport picture into something like an art form, and out of reach of your average 'picture me' box). Result: based on very bad source data you now have to prove your innocence if you somehow resemble a criminal. Nobody has heard of such facilities to be 'tools' (i.e. ASSISTING in the decision process, not actually replacing it). Groan..

Other examples of having to prove your innocence are any broad RIAA 'john doe' suit, being labelled a terrorist (in some cases you can't even prove your innocence there because you're quickly shipped outside the internationally agreed legal framework at Guantanamo Bay), Microsofts' WGA, oh, and forgetting your college card which apparently is good enough to allow police officers to submit you to unwarranted violence which in other nations would lead to such officers facing jail. On that topic, I vaguely recall that the other argument for Iraq was the police brutality the citizens were subjected to. Well, it appears a few lessons were learned there - just not the right ones..

Let me ask you something: does Washington actually have any politicians left with a spine or have they all been bought? Does anyone actually CARE about human rights there other than to harass other nations with and as a pretext to start the odd war when it's politically convenient?

The US is not 'on' the slippery slope - it's damn well sliding fast if citizens don't start making Washingtom behave like most citizens want (I'm making the distinction here because most Americans I know don't seem to agree with what's happening in Washington - proven by the latest election results). It'll be interesting to see if that 'bloody nose' Bush received in the elections will make a difference.

Given the amount of money involved, I somehow don't think so.

/rant..

Masks

[breakitdown]

That's why I always wear a mask when I commit my crimes.

Too many?

[AgentFade2Black]

It will mine the 9.5 million state licenses (snip)
Wait a minute...aren't there only 6.3 million residents or so in Massachusetts?

Re:Too many?

[slughead]

It will mine the 9.5 million state licenses (snip)

Wait a minute...aren't there only 6.3 million residents or so in Massachusetts?

6.3 million residents, 9.5 million registered voters.. are you against giving the dead the right to vote?

Racist.

Re:Too many?

[AgentFade2Black]

No, I'm not. And your attempt at humor falls rather short. Please refrain from calling someone a racist when it has nothing to do with race at all.

Re:Too many?

[Mad_Rain]

They could be counting the out-of-state college students, and/or other people who have been unable to update their licenses to the state they have recently moved to (like me). Of course, this provides me with an additional incentive to get my license changed. :)

false positives

[v1]

I wonder what the false positive rate is.

Even if the rate is unacceptably high for automated use, it could prove very helpful.

Imagine someone coming in to get a license, gets his picture taken, and while he waits for the license to be printed and laminated, the system searches its database. It produces the 10 closest matches it can find, and presents them to the DMV worker. The worker then visually compares the ten images with the actual person sitting in the waiting area. It's not necessary for the system to make the decision, only to make it easier for the people to make the call. You can't expect the DMV people to remember 10,000 faces, but it's perfectly reasonable to ask them to try to identify one face in a group of 10. Then the false positive rate count is handed to the DMV staff.

A process like this should have a very high hit rate, a very low false positive rate, and assuming the sofware is reasonalby fast, would not impact service. It would be almost totally transparent to the public. Sounds like a very good idea.

Re:false positives

[bzipitidoo]

I worked on image recognition software on my last job. The project was a failure, for many reasons. Our stuff was worse than average for such software, but the biggest problem is that facial recognition is very hard. The best stuff out there can maybe hit 90% accuracy. With lots of time consuming help from people, the accuracy rate can be pushed up to 98% or 99%. For 90% accuracy to even be possible, the subjects have to be photographed in a very controlled environment. Everyone's face must be in exactly the same orientation and position, everyone must have the same facial expression, the lighting must never vary, and the background should be blank. Superficial changes like adding or removing glasses, facial hair, makeup, pimples, and so on add to the difficulty. And there's changes from aging. Whittling 10000 pictures down to 10 would take, what, 99.9% accuracy? Just not there, so your plan won't work.

Computers can kick human butt at number crunching, chess, and such like. But in recognition, animals (including people) are still the masters. There's this vague idea floating around that computers will just naturally be better at facial recognition when scientists figure out how to properly apply their vast processing powers to the problem. Add to that the horror stories of eyewitnesses getting it wrong, of bad disguises actually working especially in the movies (Clark Kent for instance), and one could easily be lead to think we all suck at recognition. Not true-- we're great at it. The above characterizations greatly misrepresent the supposed powers of machine and man. It's the sort of thinking, underestimation, and oversimplification that has lead AI down so many blind alleys, with many fantastically wrong predictions over the years of what computers would soon be able to do. Beat the best human chess players? Just about there, at last, 40 years after the optimists of the 1950s thought it would happen, and the method is a disappointing brute force cruncher that does not employ "real" thinking involving making plans and such like reasoning. Translate languages? Still working on that one, with Babelfish good enough to be occasionally useful. OCR is decent but still not a match for people, or we wouldn't have this. Surely OCR is a much simpler problem of the same sort as facial recognition, and look how hard that still is for computers.

Law enforcement organizations don't understand or appreciate the difficulties and limitations, and aren't being helped by the many businesses that shade the truth and obscure the finer points for the sake of potential big sales. And if all the problems I mentioned above aren't enough, scaling up to millions presents another hard problem. Researchers aren't giving scale much attention at the moment because without something that works at 99% plus, there's no point. That leaves the ignorant dreaming that soon we'll have facial recognition software that can be applied to millions. They will be disappointed.

Re:false positives

[nevinera]

that's really not the point.
It might assist the police in catching criminals. occasionally.
But it's illegal, and will cause a lot more problems for the law-abiding than for the criminal.

even if we assume reasonable practices by the operators (a truly terrible assumption),
those photos were not AUTHORIZED to be used for this purpose.
I read all the fine print when i sign crap at the dmv, and there is nothing in there about the image being used for anything except to print on your driver's license.
It may be ok to use a database of mugshots, ianal, but i do know that gathering this type of information about citizens is both wrong and illegal. The excuse would be even flimsier than the one they use to tap arbitrary phone lines.

I don't know why the US thinks it can do these things, but i think the constitution needs an amendment more specifically giving a right to privacy, and detailing what steps the state may and may not take.

Yes, but that's the whole problem..

[cheros]

In your scenario, the matching tool is used correctly because it ASSISTS in the decision making process. But how often have you heard "it's on the computer so it must be right"?

The whole problem starts with someone considering the computer to be authoritative instead of yet another fraud detection tool - usually followed by downskilling the frontline workers which makes the whole matter worse.

I apologize in advance.

[EinZweiDrei]

"Face-recognition software fingers suspects"?

Great, now they're out to violate not only our privacy, but our personal space?

Re:I apologize in advance.

[monkeyman_67156]

"Face-recognition software fingers suspects"? That just sound's dirty.

"Once done, always more likely"

[mcrbids]

It's wrong to mass-search drivers' license pictures. It's also wrong to mass-search pictures of anyone who has ever been convicted of a crime. Many, many people have a regrettable misdeed in their past. It's wrong to continue to punish people who have, as was once said, "paid their debt to society."

I truly understand what you mean. Although I have not been convicted of any crime (other than a few traffic issues and a failure to appear) my closest friend was so accused. What really sucks is that, in his case, he was very well justified in commiting the "crime".

However, there is a clear and definite truth: once people have manifest a particular type of behavior, they are more likely to do that again. Pedophiles have trouble stopping. Alcoholics tend to either become teetotallers or alcoholics. Shoplifters tend to steal.

So, where does the "debt to society gets paid" mentallity stop and the "society must protect itself" begin?

Being an ex-con is not a crime, and even this facial recognition system recognizes this. But having warrants out for your arrest is something that police might be very, very interested in - and matching such searches done against past conviction records is, IMHO, very justifiable.

Re:You keep voting for powerful governments in MA

[b0s0z0ku]

Vote in a government that openly wants a lot of power, and this is what you get. Oh, but it's necessary for government to have that power to raise taxes, set rules. Because that government is needed to do some good. :-P

And Taxachusetts has been known as a viper's nest of do-gooders and Puritans since the time of the Pilgrims. Keep in mind that in the 1600s, you could be hanged for fathering a child out of wedlock among other things, something that didn't normally happen even in Britain at that time. If you want a *free* society, move to New Hampshire or Vermont, depending on whether your leanings are Libertarian-Conservative or Libertarian-Socialist.

Or come down to NYC where we have a lot of laws but few people willing to actually obey the lot of them.

-b.

False is "once done, always more likely".

[Web Goddess]

One knows more about honesty if one stole a little as a child. A sense of guilt and growth are strong deterrents to never stealing as an adult. A mistake in the teens or early twenties, is by an immature brain. Our brains literally ahven't yet stopped growing and gathering data. We are childish in our views. But once your 'record' is blemished (remember high school's dreaded 'permanent record'?) suspicious, cynical police are scrutinizing you way more frequently. Acquaintances view your character in a worse light, and interpret accordingly. It feeds on itself.

There used to be a time limitation after which your record was automatically expunged. That was probably a function of our limited ability to keep records. Nowadays, these records can be kept indefinitely, and *are* used against you.

Have to admit, I'm curious how "repeat offenders" fits into my viewpoint. I speculate people go 100% straight the first time around, or else they iterate through the criminal system until they tangent off onto a straight path.

Hey that's a nice analogy...tangent off onto a straight path.

Re:You keep voting for powerful governments in MA

[iamdrscience]

Why do you assume everybody in Massachusetts votes and thinks the same way? I'm a libertarian.

False positive rate > 100% might be good

[benhocking]

If the false positive rate is around 200%, and if the false negative rate is 50%, then any given search might tend to yield 2-4 suspects (depending on the variance of the false positives, of course) and about half the time the actual culprit will be included. Of the 2-4 suspects that are not the culprit, most of the time it will be quite easy to eliminate them, either visually by the user, or from simple detective work (i.e., a rock-solid alibi or possibly even common sense in some cases). So, for a fairly low cost, you've got an additional lead. Having a high false positive rate means a lower false negative rate with the added benefit of having the user get used to the fact that the computer is quite fallible.

I'm not arguing for or against the ethics, just the efficiency.

Hmm...oh..I get it .. ?

[the_REAL_sam]

So it would have fingered the suspect in the case of the wendy's chilli bowl..with..the..human.finger..? =\

Unlicensed criminals?

[Arkaaito]

Why driver's license photos? Given that it's going to be called upon when surveillance footage is available, it will be used mostly cases of shoplifting and some robbery. Of all types of criminals, those are probably the least likely to have driver's license photos on record, because they're the youngest and poorest demographics respectively.

Expunged?

[Slashdot Parent]

Sometimes you have have petty stuff that you did as a young adult expunged. If it's really bothering you that much, it might be worth looking into.

All depends on what you did. If you stole a shirt or got into a barfight, you might stand a chance. If you murdered your ex... well... yeah, good luck.