In Shenzhen, China, a port city on the Pearl River, a bit north of Hong Kong, is going to be the home of an experiment that might just curl your hair.
According to an article in the NYTimes the authorities are taking identity cards to the next step. The card will contain not merely the name and address of the holder, but also work history, education, religion, ethnicity, police record, medical insurance and the landlord’s phone number. The reason for this? Fighting Crime, of course.
This is a little easier in China than, oh, let’s say the US, or Canada, as the government in China can do whatever they want. Ostensibly, we have controls here.
This was illustrated by a comment quoted by the author, Keith Bradsher, in the NYTimes story: "If they do not get the permanent card, they cannot live here, they cannot get government benefits and that is a way for the government to control the population in the future," said Michael Lin, the VP for investor relations at China Public Security Technology, the company providing the technology. Plans are afoot to include credit histories, subway travel payments and even small purchases to the functionality of the ID Card.
Short form? China is going to tag their citizens like cattle, put more cameras everywhere and use face recognition software to track down miscreants: The whole system is built for it. Interestingly enough, the whole system is GPS-based, powered by big North America companies who provide the software, hardware and skills to put the whole thing together.
China Public Security Technology, the company providing the technology, is incorporated in Florida, China Public Security has raised much of the money to develop its technology from two investment funds in Plano, Tex., Pinnacle Fund and Pinnacle China Fund. Three investment banks — Roth Capital Partners in Newport Beach, Calif.; Oppenheimer & Company in New York; and First Asia Finance Group of Hong Kong — helped raise the money.
Check out their website at http://www.chinacpby.com if you want to eyeball their SEC filings, as they are an over the counter stock.
As it stands now, Britain has something like one CCTV camera per 14 citizens. Chicago has promised the Loop will have a complete 2,000 camera remote surveillance system shortly. They already have ‘gunshot detector’ cameras mounted in certain ‘high crime’ areas. Toll-Road EZ-Pass data is obtainable in divorce court proceedings in New Jersey. New York City already has 1,000 cameras in the subways, with another 3,100 monitoring city housing projects.
COMPASS, the traffic monitoring system in Ontario has 104 cameras watching over us. Right now, I’m watching the intersection of the Airport Parkway and Hunt Club Road in Ottawa. Traffic is moving fine, by the way.
There are no statistics about how many private CCTV systems exist, like the ones used in stores to stop shoplifters, or watch over the cash register. For that matter, there are also no legal controls over what the store can do with the images they collect.
The Face Recognition Vendor Test for 2006 asked for participants and got a lot of applicants. The results were not as wonderful as the tinfoil hat brigade would suggest, in that false positive identifications were a little too high still. Vendors in the casino industry security space, however, claim that their software can spot a cheat with little more than a blurry shot of someone’s’ left eyebrow. The truth is somewhere between the two ends of the spectrum.
Which brings us back to the Identity Card system in Shenzen. How easy would it be to implement in Canada? Too easy is the accurate answer.
If you check your drivers’ license or health card, you’ll see a big, fat, magnetic stripe and probably a bar-code readable area. That’s all it takes to find you in the database in the sky. The question becomes what data can be incorporated into your record. Shenzhen wants to have everything, including, if the face recognition software works out, your whereabouts at any given time based on surveillance CCTV cameras.
It wouldn’t surprise me in the least if The Department of Homeland Paranoia is looking longingly at Shenzen and the Peoples’s Republic of China, using them as a beta test. A beta test? Dave, that’s a bit over the top, even for you? Not really, the companies providing the technology are all American companies.
I’ve worked on systems (older ones) that did license plate recognition, using infrared cameras to read your car tag, front, back or both, that could sort and match things in under two seconds. This was before the high speed systems we have now, in the mists of time known as i386 and 486 days. Face recognition software is another step in matching a unique metric to a person.
Faces are almost all unique. Take four or five measurements and you can probably come up with a unique number for the face attached to the body. These are things you can do at home, if you want to. Measure your intraocular distance, that distance between the irises of both eyes, like your optometrist does, with his or her little steel ruler. That’s one unique number.
Measure from the notch on your brow where your nose joins your forehand, to the tip of your nose. Another number. Measure from tip of nose to tip of your jawbone, then measure a triangle from ear hole to jaw tip, to ear hole, as well as the angles of that triangle. Don’t bother noting that you have a four-colour snake tattoo on your face and enough piercings to sound like an ocarina on a windy day. Then measure the little patch of skin between the bottom of your nose and the top of your upper lip.
You now have enough data to do a really cheap, probably effective, numerical representation of the landmarks of your face, or any face.
Add’ em all up and then search for faces in your library that have a value near your value. You now have come very close to recognizing a face. Where do you get a library of faces? Mug shots from the cops will do nicely, thanks. Drivers’ Licenses, Health Cards, Costco Membership Cards and Company ID will also supplement the database.
The relationships between those numbers, (your intraocular distance being .987 of the length of your nose, while nose to jaw is 1.244 of your nose length and 1.863 of your intraocular distance) is what you really want. If you have the relationship between landmarks, then you can search by the relationships: Find me everyone with a nose this long and at least an ear-hole to jaw distance of this number on either left or right, as most faces are close to symmetric from ear-hole to jaw, left side or right side.
I am simplifying beyond absurdity here, plugging in some fanciful numbers. Face recognition software uses several dozen landmarks, the more landmarks; the more accuracy, but the core concept is the same. Reduce a face to landmarks and derive a big number from it, out to a few dozen decimal places.
Big Math says if you can do a measurement remotely, you can narrow your search down to a few faces that come close to matching. A database is just Big Math for comparing and searching. Face Recognition Software is a way to automate finding and measuring landmarks, then assigning a value to it with the least amount of human intervention.
All you have to do is say that Jerimiah Fleegus has a face of this value, a credit score of this value, a health card of this value and a drivers’ license of this value. Add in his address, organ donor status and a few other impertinent facts, burn it onto the back of his ID card and voila! Jerimiah Fleegus is now searchable, even in crowds of people.
Tie in some other data, like Jerimiah Fleegus’ car was noted coming downtown a half-hour earlier on a toll-road, then he bought a sub (Chicken Teriyaki, 6-inch, whole wheat) using his debt card one block away, not twenty minutes ago. A public CCTV camera spotted someone close to Jerimiah Fleegus’ face coming out of the restaurant and heading to the ball park.
Then a camera at the ball park pedestrian ramp noted a face, sort of like Jerimiah’s, climbing to a seat in the high up seats. We already know he has season tickets to the Jays and the seats are in the nosebleeds, from his credit card records.
So where is Jerimiah Fleegus? Probably at the ball game, watching the Jays beat the Anaheim Angles like a cheap gong, 4-1 in front of 31,978 fans. We checked, using the cameras in the stands, and yep, there was Jerimiah, so far above the third base line that he is closer to the CN Tower Observation Deck than field level.
He’ll check out of the car park around 10:30 PM and be on the toll road near 10:55 PM. If the car gets off on the Hurontario exist of the toll-road, then he’s probably going home. If we check his Rogers PVR later tonight, we’ll see that he watched an old episode of Mythbusters around midnight and skipped over the commercials. Then he turned it off at 0040 hrs and went to bed.
So then this story, U.S. to Expand Domestic Use Of Spy Satellites off the front page of WSJ online, by ROBERT BLOCK August 15, 2007; Page A1 would point to more societal management coming our way:
"The U.S.’s top intelligence official has greatly expanded the range of federal and local authorities who can get access to information from the nation’s vast network of spy satellites in the U.S.
The decision, made three months ago by Director of National Intelligence Michael McConnell, places for the first time some of the U.S.’s most powerful intelligence-gathering tools at the disposal of domestic security officials. The move was authorized in a May 25 memo sent to Homeland Security Secretary Michael Chertoff asking his department to facilitate access to the spy network on behalf of civilian agencies and law enforcement."
The capabilities of American spy satellites are subject to much speculation, most of it ill-informed or wildly exaggerated. They don’t need to see if Vladimir Putin shakes it with his left hand or his right hand.
For the record, Putin is right-handed, and he doesn’t strike me as the kind of person who would take piss outside, unless he was at his dacha at Solovyovka. Solovyovka is just off the M56, but is well wooded, so he could take a leak in the woods, if so inclined. I’m reasonably certain somewhere in the depths of Homeland Paranoia or the DIA Mapping Branch, there exists pictures that prove or disprove Putin’s handedness, as well as his predilection for taking a piss in the forest. I digress.
We know Google Earth and MS Maps Live have the resolution needed to see license plates. There are enough anecdotal stories about people being photographed from great height, sunbathing nude in the backyard,or coming out of peeler bars, for it to be funny anymore.
We know that the Toronto police are using airborne Infrared cameras to find houses that have too many lights on at 0300: This usually indicates a grow operation and as often as not, means a group of unfriendly people are coming in the door. Unfriendly to the tennant at least.
Data mining we’ve talked about before, but this is almost into the realm of speculative fiction until you stop and think about it. The ability to have this level of data about everyone in a few seconds short of real time, is perfectly doable right now. There is no magic involved, or special computers with a black hole suspended in space plasma at their center.
The issue is not the doing. The issue is the why we’re doing it.
RoadDave 