India Is Planning a Huge China-Style Facial Recognition Program (bloomberg.com) 51
India is planning to set up one of the world's largest facial recognition systems, potentially a lucrative opportunity for surveillance companies and a nightmare for privacy advocates who fear it will lead to a Chinese-style Orwellian state. Bloomberg reports: Prime Minister Narendra Modi's government will open bids next month to build a system to centralize facial recognition data captured through surveillance cameras across India. It would link up with databases containing records for everything from passports to fingerprints to help India's depleted police force identify criminals, missing persons and dead bodies. The government says the move is designed to help one of the world's most understaffed police forces, which has one officer for every 724 citizens -- well below global norms. It also could be a boon for companies: TechSci Research estimates India's facial recognition market will grow sixfold by 2024 to $4.3 billion, nearly on par with China.
But the project is also ringing alarm bells in a nation with no data privacy laws and a government that just shut down the internet for the last seven weeks in the key state of Kashmir to prevent unrest. While India is still far from implementing a system that matches China's ability to use technology to control the population, the lack of proper safeguards opens the door for abuses.
But the project is also ringing alarm bells in a nation with no data privacy laws and a government that just shut down the internet for the last seven weeks in the key state of Kashmir to prevent unrest. While India is still far from implementing a system that matches China's ability to use technology to control the population, the lack of proper safeguards opens the door for abuses.
Fuck India (Score:1)
Re:Fuck India (Score:5, Insightful)
Fuck India. Leave the people of Kashmir alone you motherfuckers!
I realise that you used the word 'motherfucker' as an incest themed insult (in fact you English speakers have a disturbing tendency to work sexual intercourse into every sentence that leaves your lips) but please be advised that it a poor insult. You see, from the point of view of set theory, guys who fuck their own mother are only a tiny edge case of the set of all 'motherfuckers'. For example, all you have to do to become a 'motherfucker' is to sleep with a woman who is a mother, any woman who is a mother. Furthermore, any of you guys who are married, who have at least one child, and are still having sex with the mother of your child are by definition 'motherfuckers'. So you may want to consider another and less ambiguous choice of insult.
Re: (Score:2)
Says someone defending a country passing China as the most populous.
And this has what? ... to do with most native English speakers being the potty mouthed hillbillies of linguistics?
Re: (Score:2)
Sophomoric pedantry. As with many, many insults, the meaning is context-specific. All you have to do to be a pussy is be a cat. All you have to do to be an ass is be a donkey. All you have to do to be a bitch is have a litter of puppies. All you have to do to be a bastard is be born out of wedlock. All you have to do to be a faggot is be a bundle of sticks. Etc.
Re: (Score:2)
Sophomoric pedantry...
No, you are overthinking it, that war garden variety trolling.
Re: Fuck India (Score:1)
We need a global redrawing of borders. (Score:1)
Based on actual cultures.
Not colonialist and imperialist assholes, posing as countries and unions.
At the very least, Africa, South America, India, Afghanistan, the Midlle East, the USA and the EU need to be completely redrawn. Based on each individual's decision who he associates himself with. And nothing else.
Obviously, settling disputes with regard to who gets to use the resources between their areas will be the hard part.
I suggest that 1., you need to be adjacent to it, 2. you don't get to just move in a
Re: (Score:1)
Based on each individual's decision who he associates himself with. And nothing else.
I predict a lot of people would associate themselves with Luxembourg. [wikipedia.org]
Though they will drop down from first place on the list after the influx of new citizens...(Socialists don't think that far ahead though)
Re: (Score:1)
Re: (Score:2)
Fuck India. Leave the people of Kashmir alone you motherfuckers!
Spotted the jihadist! Go blow up another wedding or something.
No Running Water (Score:4, Insightful)
Crushing poverty and a lack of simple services like running water in much of the country and this is their priority....
Re:No Running Water (Score:4, Informative)
Re: (Score:3)
That's basically what all religions tell you. Accept what you have, don't you dare rebel against those that made this shit up to rule over you.
Re: (Score:2)
I'm free not to go to church exactly because we're not a theocracy. You think I could have in the 1600s? Or if you want a more contemporary example, not believe in Allah in today's Iran?
The same can be said for the USA though. (Score:1, Insightful)
From my standpoint, the rural, and especially southern USA might aswell be a post-apocalyptic wasteland. As if they were decades behind.
I get Fallout vibes whenever I get to experience it.
[[The people are nicer than we're told though. Stereotypical rednecks are a bit like stereotypical Russians: They know how to have fun. Except with less education and more Jesus (sadly apparently without having read the bible). ;)
But I digress.]]
I think India is too diverse and as anyone knows, should never have been force
P.S.: Obviously, mass-surveillance is wrong and ev (Score:1)
And this big project is nothing to aspire to, but one of the most shameful and evil things since literally Stalin's era.
Forgot to add that key bit.
Re: (Score:3, Insightful)
Well, I am a Christian who lives in the rural southern USA (South Carolina to be more precise) and I can say first hand that it is nothing like your vision. We have manufacturing plants everywhere filled with skilled "stereotypically redneck" workers who do a pretty good job at what they do. You don't need to have a college degree to learn how to do something well. It is not "wrong" to have less education than what you would consider acceptable, nor is it "wrong" to be a follower of Jesus though I agree tha
Re: (Score:2)
The world needs people to do unskilled work, semi-skilled work, skilled work, and professional work.
The semi-skilled work is in the process of going away. It's either being turned into unskilled work with automated tools, or being automated away completely. Once upon a time, a worker had to know how to use a torque wrench correctly (or, as was more common, mistorqued fasteners and caused reliability problems.) Now, we use machine tools to torque a set of bolts at once, and/or in prescribed sequence, and the human assembly line worker's responsibility is to finely position the tool so that it can seat onto
Re: (Score:2)
Part of the problem is that governments get sold a bill of goods; namely that surveillance will keep people safer, lower crime, etc. May or may not be true, but good goals if you want to show that you're "doing something" as a politician, especially if the alternative "something" is redistributing wealth away from major campaign donors.
Re: (Score:3)
[[The people are nicer than we're told though. Stereotypical rednecks are a bit like stereotypical Russians: They know how to have fun.
That's because today they have the jobs that graft and thug violence drove out of your state.
Re: (Score:2)
Re:No Running Water (Score:4, Interesting)
Crushing poverty and a lack of simple services like running water in much of the country and this is their priority....
Remember, for a long time India was ruled by UK, which is ruled by system of aristocracy. This dovetailed very well with the primordial caste system in India, and under UK rule, it morphed into something even more extreme. For instance, only people of a certain caste could get civil service jobs.
If you are an aristocrat, your best strategy is to keep the lower classes in poverty. If you help those people out of poverty and give them simple services, they might get rebellious.
Poverty is an aristocratic tool to control the poor.
Re: (Score:3)
India remains poor because of monstrous levels of corruption. Being a nominal democracy does little to relieve that burden.
Re: (Score:3)
Re: (Score:2)
India remains poor because of monstrous levels of corruption. Being a nominal democracy does little to relieve that burden.
It probably makes it a bit worse. Representative democracy is prone to corruption. Francis Fukuyama argues persuasively in "Political Order and Political Decay" that high levels of corruption is a normal stage that all developing democracies go through. The US has had massive corruption at all levels of government for much of its history, it wasn't until the early 20th century that corruption was brought under control... and corruption has actually been on the rise again for the last 30 years or so. Most
Re: (Score:2)
"Remember, for a long time India was ruled by UK, "
Nations which are CURRENTLY being dominated by another get to use excuses like that when explaining why they aren't doing more for their people. That's no excuse for choosing to spend your budget doing things to people instead of for them.
India needs to put official efforts into cleaning water and restoring forests, and stop depending on inspired citizens to do all of the work of bioremediation. Otherwise they're just going to wind up with a lot of excellen
Re: (Score:2)
That's basically the reason for this move. Crushing poverty leads to a bigger and bigger pool of very angry people. Malcontent people are like fissionable material, you can handle it in small quantities but if the critical mass is reached, the shit blows up.
So you need something to control them. Alternatively you could of course try to reduce the mass, but where's the money in that?
Re: (Score:2)
In arms manufacturing, probably.
Re: (Score:2)
Crushing poverty and a lack of simple services like running water in much of the country and this is their priority....
Which is exactly why their leaders need tools of population control. It is not a question will it blow up or not, but when it will blow up.
I suspect one more thing for them to add to the other stuff in their "Russian Shop" shopping basket in addition to missiles, stealth busting and aircraft carriers. End of the day, they already have the relationship at government/military level and the Russian stuff is presently best in class. Faceapp is only the tip of the gigantic iceberg: https://www.fagain.co.uk/nod [fagain.co.uk]
Better surveillance than the UK? (Score:2)
Re: (Score:1)
Suddenly a lot of people with shared, fake, created ID will get detected in India as illegal migrants.
They will then have to prove to the gov that they are who their "issued" ID say they are.
The easy way is to go over every gov document in a digital search.
Lots of people using the same shared name, location, dates that got reused for a shared ID as they "existed" tend to get detected.
But that sharing and reuse takes some better searches.
Next up: Murica an Europe (Score:1)
I wonder what asshole and his business is (lobbying) behind this.
But he sure will be be on the list to go back in time and shoot. Right next to Hitler and DJ Denniz Pop.
Ground Reality (Score:3, Funny)
Re: (Score:2)
It will likely just fall into disrepair, like everything else in the "developing" world.
Re: Ground Reality (Score:2)
They all look alike (Score:2)
Which is to say, any group of 1.3B faces is going to all look alike.
Or at least, too alike for a computer to distinguish with any reliability.
Great for India (Score:1)
Getting free stuff from tax payers over decades that was to help support citizens.
Birthday problem (Score:4, Insightful)
Moral issues aside (not that they should be put to the side, but there are plenty of others talking about them), I wonder how they think they can make this work. Also, a correct understanding of the problems I'm going to discuss raises a whole new set of moral questions.
There's a central problem in matching systems that is known in computer science and math as the "Birthday Problem". Suppose you're at a party with 40 people and someone asks you if you want to bet him $10 that no two people at the party have the same birthday. Assuming you don't have any knowledge of the birthdates of the people at the party and you're equally certain that the other person doesn't either, do you take that bet? There are 365 possible birthdates (we'll ignore leap years for simplicity) and only 40 people present, so naively it seems like a good bet. It's not. It's awful. The probability of a birthday "collision" among 40 people, assuming birthdates are uniformly distributed, is nearly 90%.
The reason this is so gives you a hint toward why this is a problem for face recognition and other matching systems: 40 people at the party means there are 40*39/2 = 780 pairs of people. The "only 40 people but 365 days" naive calculation your brain might have done is valid, but it's a different problem. If the bet had been about whether there was anyone at the party that had a specific randomly-chosen birthday, the odds would be 40/365 = 11%. But the bet is that some pair of people have the same birthday, and there aren't 40 chances, there are 780.
(The way to calculate the actual probability here is to look at the probability that everyone has a unique birthday. The first person can have any birthday, so 365/365. The second person can't have the same birthday as the first, so there are only 364 possibilities, so the odds of two people not having the same birthday is 365/365 * 364/365. Add a third person and it's 365/365 * 364/365 * 363/365. So for n people, n < 365, the probability of zero collisions is (365!/(365-n-1)!) / (365^n). The probability of one or more collisions -- pairs of people with the same birthday -- is therefore 1 - (365!/(365-n-1)!) / (365^n).)
Any biometric matching system fundamentally converts data about part of a person into a vector, a point in n-dimensional space (for some n). The scanning and conversion processes are fuzzy, though, so it's a slightly different point every time, but the systems are designs such that all the scans and conversions for a given person produce points that are close to one another in n-space. When a pair of points (say, one from a database of known people and one from a camera image of an unknown person) are compared, the system computes the distance between them and compares the result to a threshold. If they're close enough, the system says they're probably the same person.
Any given biometric acquisition and matching system has error rates: it produces both false positives, erroneous matches, and false negatives, erroneous non-matches. These are called the False Accept Rate (FAR) and False Reject Rate (FRR). Obviously, both are dependent on the choice of threshold distance. Looser thresholds increase FAR and decrease FRR, tighter thresholds do the reverse. When the threshold is adjusted so the error rates are equal the resulting rate is called the Equal Error Rate (ERR -- yes, I know the acronym doesn't match). The ERR is a measure of the overall accuracy of the system.
You can think of the ERR as dividing the n-space of biometric scans into roughly 1/ERR "buckets" that are analogous to days in the birthday problem. A common ERR is 1/50,000, 2E-5. Now, 2E-5 is pretty good if you're matching one person against one stored point (called a template). It means that the system is right 99.998% of the time. But if you're matching a person against a database of k other people, it quickly begins to suck as k grows. If k=1000, the probability that a random person not in the database will "match" someone in the database is 2%,
Re: (Score:1)
Way beyond any "issue" as everyone without ID will get an interview to show all their gov documents
Thats the human part of "biometric acquisition" when documents are fake, shared, dont exist, got placed in at a later date.
Re "every face captured on a camera will match tens of thousands of faces in the database" and all their supporting documents
Its not just a "face" its an entire review of all matching
can not prove biometrics are unique enough (Score:1)
Re: (Score:2)
No biometric measure has ever been proved unique amongst a large population.
You're just asserting that the problems I pointed out might hold even in the case of a biometric matching system with perfect accuracy. But that really doesn't matter, because we'll never get anywhere close to perfect accuracy. The fact that there might be a few identical twins, identical per whatever biometric you're measuring, doesn't matter because any real system will not only categorize the actually-identical people as the same, it'll call a bunch of others identical, too.
Even if it did work, if your facial parameters as stored in the system are stolen then you are screwed for life unless you start mutilating your face.
This actually doesn't matter
Re: (Score:1)
97% Indians doesn't care (Score:2)