As I’m reading this, this post is #1 on HN with a mere 9 points and zero comments.
This is frustrating me because I cannot conceive how this can possibly rank higher than other articles with such weak data points .
Why does it frustrate you? If it had stayed at zero comments and few points, it would have dropped from the front page quickly, probably within minutes.
From http://hnrankings.info/ it looks like the #2 submission was "Detecting the use of "curl | bash" server-side" (324 points, 10 hours ago, 105 comments)
Now your comments has 35 minutes and the OP has 56 minutes. So with some approximations, lets say that at the time you wrote this the OP has an age of 0.5 hour and the #2 post has 9.5 hours.
Roughly the order of the HN post is: upvotes/age^1.5 * lot-of-penalites
[There are many post that discuss the details of the order in HN that are more accurate, I don't remember if the gravity factor is 1.5 or 1.6 or whatever, and it may change from time to time. It's part of the secret sauce, but let's use this version as an approximation.]
The weight of the articles half a hour ago was:
This post -> 9/(.5)^1.5 = 25
Curl/bash post -> 324/(9.5)^1.5 = 11
So it's expected that this post was in front of the curl/bash post. Not surprising.
As far as I know, the post don't get a better position because they have comments. [It's part of the secret sauce, ...] I personally prefer post with few comments.
There are many penalties, some are automatic, some are added by the mods and some are caused by the flags of the users. [It's part of the secret sauce, ...] One penalty that is surprising for many users is the "flamewar" detector, that applies a penalty when an article has too much comments (something like "comments > upvotes + 40"???) So too many comments is usually bad for the article.
In this case, I think that neither of the articles has activated the "flamewar" detector or other penalty. I think the problem is that if you don't have a rough idea of the order algorithm it is sometimes surprising.
When volunteers presented passport applications containing photographs of other people, the officials did not recognize the discrepancy fourteen per cent of the time. One might attribute this alarming statistic to reckless inattention. Yet the system for hiring passport officers likely rests on the faulty premise that applicants have essentially equal skill in recognizing faces.
The obvious low-tech solution to this problem is disallowing bringing your own photo, and having a photo booth at the passport application site, where you have to take the photo in front of the officer.
Why do many countries still permit bringing your own photo?
The setup you described would be really expensive and cumbersome compared to people submitting their own photos, assuming of course you can apply for a passport by post or online and don't have to go to an office with an appointment anyway, in which case it may make some sense even if it only increases the quality of photos by a few percent -- to be honest I'd be skeptical that the improvement would be more than this, the parameters for self taken photos are rather strict after all, such that it's quite hard to take photo that 'doesn't look like you' because of light, hair, makeup etc and have it be accepted.
Mostly agreed. However, your last bit misses the point:
> the parameters for self taken photos are rather strict after all, such that it's quite hard to take photo that 'doesn't look like you' because of light, hair, makeup etc and have it be accepted.
The point there isn't to control for lighting conditions, but to make it physically impossible for the photo in the passport to be of someone who didn't bring the application in.
This is only partially true: because you cannot wear your glasses and you may need to put your hair in a specific way to clear your face, you definitely can obtain a picture that doesn't look like your casual look (even if the natural facial features are matching).
Lots of places do exactly this. In fact the last time I applied for a passport, it was completely paperless, no paper forms to enter, no stapled-on photos to scan. It's hard to believe this would not be cheaper, in 2018.
Visa applications too -- sometimes there's one kiosk-like machine which wants your fingerprints as well.
Biometric-compatible photos must be done at the site of issuance, because people just cannot stop smiling and doing other nonsense when taking photographs themselves.
A photo booth at the site of issuance is very cheap compared to salaries of employees who work there. And it lowers total cost of issuance when you consider costs on the side of the citizen.
Not true in the UK, most people apply for their biometric passport online (or by post, which costs more). Adults applying for their first passport may be asked to attend an interview.
Just last week I applied Indian Passport because of some changes in my address & othet particulars, the process was to fill an online form, pay fees online, book appointment online, show up with a confirmation SMS & old Original Passport alongwith one more ID Original. So basically no photocopies, no paper. Went in at pre-opted time, entered a number from my SMS in a kiosk machine, confirmed all particulars filled in online as correct, stood for photo, put fingers on scanner when asked, got another token number, where when called, a single person asked me to show him my original existing passport & second ID, all ok, received a SMS & Email with all the details, & a case number to follow the progress. All paperless & interacted with only one human. All this was managed by Tata Technology & Indian Government.
Fun Fact, Machine didn't show me my photo till the very last step, & I thought of changing & having one more try to pose with looking bit good, got an error that Photo will he taken only once, & only them can initiate the re-click of photo, not me.
We also have similar machines in Portugal for ID card and passport issuance, but luckily they will ask if you like the picture or if you want another taken...
I think yes, as Tata is a reputed brand & group of companies (auto, tech, steel, infra & many) & also because whole passport office & operation is always supervised by government officials. Before this, it was always a big bundle of original & photocopied documents & forms.
With a quick look there seem to be 100+ unique TCS domain logins on github. I’m sure there are some smart people there, but this seems like a seriously troubled organization.
Putting trust in TCS just because it is Tata is very worrisome. Tata products are 'reputed' only in local Indian market bubble, propped up by government protectionist measures.
It sounds odd because typically in the UK you need to get your photos and application signed by some 'trusted member of society', e.g. someone who does a job on an approved list of professions and isn't supposed to have known you for more than 2 years.
I suppose that's easily bypassed though. I had a much harder time validating my photo for my business account - the only verification that mattered then was one I had to pay for at the post office. It had to have a paper trail.
The article is interesting and raises interesting points. For instance about the "face-recognition" impairment in police officers, which is a serious problem.
I was however surprised that this ability was not well known, I remember clearly reading stuff about it a LONG time ago. Using it in a systematic way & in a police unit may be new however.
And, of course, as EVERY TIME the topic is raised, I was aghast by the complete off-hand dismissal of the human right issues of the CCTV network. It is a HUGE mistake to underestimate the abuse potential, that's why it must be addressed NOW, because when surveillance will be 100% widespread and accepted as "normal" by people, it will be too late to go back. And we are going there very fast as the example of the UK has already shown us.
Some of the worse things are created out of very good intentions, here the police officer is clearly very honest & sincere about his use of the CCTV footage to catch criminals, but he forgets that he will NOT be the only one with access to it. I'd like to ask him: okay, so can you vouch personally for ALL the people who will use that system in the future, most of whom you have never met (and never will)? Do you REALLY believe that there will not be criminals, corrupt public servants or authoritarian officials who will access it (for nefarious purposes)? Note that this is already a HUGE issue with the secret services in all countries: there IS abuse, a lot of it is publicly known, here we'd be giving "the eye in the sky" enabling even worse abuses (more information available to more -potentially bad- people)!
From the article:
> “There is a friendly eye in the sky,” a Home Office minister proclaimed in 1994. “There is nothing sinister about it, and the innocent have nothing to fear.”
Stupid quote. The simple fact that he feels the need to say that there is "nothing sinister" is already raising a huge red flag. The truth is that history showed us that Nobody is EVER "innocent", there will always be someone to blame you for something: too thin, too fat, too short, too tall, too religious, not religious enough, etc. You know what I mean, it's a constant in humanity's history: the "need" to find an enemy, a scapegoat, someone who is "different" to blame for whatever issues are currently happening.
It is unfortunate (and worrying) that that stupid belief is so widespread (I hear and read it very often)...
To add on to your point, we have no idea what unjust laws will be passed in the future. Building the infrastructure for opression, even if not being currently used for that, is dangerous considering the uncertainty of how it will be used in the future
Completely agreeing on the dangers of CCTV. They even say in the article that the efficiency of the system is quite disputed: most crime footage will rot in a database. In general i would always be wary of an approach to security which tries to capture exhaustive information about something:
* there can and will be some form of misuse, even indirect: anti-terrorist laws in france being used to harass activists--ecologists and other; implies making things centralized which defeats security (like TLS PKI)
* costs a lot of money to get a pile of mostly useless data
* someone will throw some big data/machine learning at it and this gets dangerous because we start to base our police (and justice) on opaque statistical tools
Yet seeing that they use people instead of machines to do this job makes me a bit less angry.
This tells us that automated face recognition can get much better. It's now approaching average human level, but there's no reason it has to stop there.
I'm fascinated by the Cambridge Face Memory Test (direct link: http://www.bbk.ac.uk/psychology/psychologyexperiments/experi... ). I got 99%, which seems remarkably and unrealistically high even though people say I have a good memory for faces. Of course, I can't just go take the test again, because now I know what the faces look like.
I've had moments like what the other super-recognizer officers did, though. Before I got married and I was on a dating site I shall not name, I immediately recognized one of the secretaries on there from the C-suite, someone I interacted with a couple times but only that. I quickly broke the match, hopefully before she noticed, since that might have been a little weird.
More often I know faces but not names, which makes things awfully awkward at professional meetings. "Hey, ... guy!"
Heh. I got 97%. I like to think I have a good face memory, but some of the noisy images felt like guessing. Also: weird that the test consists entirely of remembering the faces of white guys.
Yeah, that was weird too. I don't know how reliable this would be for someone who hasn't primarily been around "white guys." At minimum I was expecting some pictures of female faces, but there wasn't even that.
32 comments
[ 3.8 ms ] story [ 73.2 ms ] threadCan anyone here shed some light on this ?
Now your comments has 35 minutes and the OP has 56 minutes. So with some approximations, lets say that at the time you wrote this the OP has an age of 0.5 hour and the #2 post has 9.5 hours.
Roughly the order of the HN post is: upvotes/age^1.5 * lot-of-penalites
[There are many post that discuss the details of the order in HN that are more accurate, I don't remember if the gravity factor is 1.5 or 1.6 or whatever, and it may change from time to time. It's part of the secret sauce, but let's use this version as an approximation.]
The weight of the articles half a hour ago was:
This post -> 9/(.5)^1.5 = 25
Curl/bash post -> 324/(9.5)^1.5 = 11
So it's expected that this post was in front of the curl/bash post. Not surprising.
As far as I know, the post don't get a better position because they have comments. [It's part of the secret sauce, ...] I personally prefer post with few comments.
There are many penalties, some are automatic, some are added by the mods and some are caused by the flags of the users. [It's part of the secret sauce, ...] One penalty that is surprising for many users is the "flamewar" detector, that applies a penalty when an article has too much comments (something like "comments > upvotes + 40"???) So too many comments is usually bad for the article.
In this case, I think that neither of the articles has activated the "flamewar" detector or other penalty. I think the problem is that if you don't have a rough idea of the order algorithm it is sometimes surprising.
The obvious low-tech solution to this problem is disallowing bringing your own photo, and having a photo booth at the passport application site, where you have to take the photo in front of the officer. Why do many countries still permit bringing your own photo?
The setup you described would be really expensive and cumbersome compared to people submitting their own photos, assuming of course you can apply for a passport by post or online and don't have to go to an office with an appointment anyway, in which case it may make some sense even if it only increases the quality of photos by a few percent -- to be honest I'd be skeptical that the improvement would be more than this, the parameters for self taken photos are rather strict after all, such that it's quite hard to take photo that 'doesn't look like you' because of light, hair, makeup etc and have it be accepted.
> the parameters for self taken photos are rather strict after all, such that it's quite hard to take photo that 'doesn't look like you' because of light, hair, makeup etc and have it be accepted.
The point there isn't to control for lighting conditions, but to make it physically impossible for the photo in the passport to be of someone who didn't bring the application in.
Visa applications too -- sometimes there's one kiosk-like machine which wants your fingerprints as well.
A photo booth at the site of issuance is very cheap compared to salaries of employees who work there. And it lowers total cost of issuance when you consider costs on the side of the citizen.
And that itself is a relatively new change, and there are still only a few offices around the country.
Fun Fact, Machine didn't show me my photo till the very last step, & I thought of changing & having one more try to pose with looking bit good, got an error that Photo will he taken only once, & only them can initiate the re-click of photo, not me.
Searching something like “tcs.com password” or “proxy.tcs.com” finds vast amounts of internal tata stuff like this https://github.com/nabind/prism_test/blob/9b42b56cd3c108e8d4...
or this https://github.com/manojkarthick/Real-Time-Twitter-Analysis/...
With a quick look there seem to be 100+ unique TCS domain logins on github. I’m sure there are some smart people there, but this seems like a seriously troubled organization.
https://techcrunch.com/2010/02/07/indian-it-giant-tata-consu...
I suppose that's easily bypassed though. I had a much harder time validating my photo for my business account - the only verification that mattered then was one I had to pay for at the post office. It had to have a paper trail.
I was however surprised that this ability was not well known, I remember clearly reading stuff about it a LONG time ago. Using it in a systematic way & in a police unit may be new however.
And, of course, as EVERY TIME the topic is raised, I was aghast by the complete off-hand dismissal of the human right issues of the CCTV network. It is a HUGE mistake to underestimate the abuse potential, that's why it must be addressed NOW, because when surveillance will be 100% widespread and accepted as "normal" by people, it will be too late to go back. And we are going there very fast as the example of the UK has already shown us.
Some of the worse things are created out of very good intentions, here the police officer is clearly very honest & sincere about his use of the CCTV footage to catch criminals, but he forgets that he will NOT be the only one with access to it. I'd like to ask him: okay, so can you vouch personally for ALL the people who will use that system in the future, most of whom you have never met (and never will)? Do you REALLY believe that there will not be criminals, corrupt public servants or authoritarian officials who will access it (for nefarious purposes)? Note that this is already a HUGE issue with the secret services in all countries: there IS abuse, a lot of it is publicly known, here we'd be giving "the eye in the sky" enabling even worse abuses (more information available to more -potentially bad- people)!
From the article:
> “There is a friendly eye in the sky,” a Home Office minister proclaimed in 1994. “There is nothing sinister about it, and the innocent have nothing to fear.”
Stupid quote. The simple fact that he feels the need to say that there is "nothing sinister" is already raising a huge red flag. The truth is that history showed us that Nobody is EVER "innocent", there will always be someone to blame you for something: too thin, too fat, too short, too tall, too religious, not religious enough, etc. You know what I mean, it's a constant in humanity's history: the "need" to find an enemy, a scapegoat, someone who is "different" to blame for whatever issues are currently happening.
It is unfortunate (and worrying) that that stupid belief is so widespread (I hear and read it very often)...
* there can and will be some form of misuse, even indirect: anti-terrorist laws in france being used to harass activists--ecologists and other; implies making things centralized which defeats security (like TLS PKI) * costs a lot of money to get a pile of mostly useless data * someone will throw some big data/machine learning at it and this gets dangerous because we start to base our police (and justice) on opaque statistical tools
Yet seeing that they use people instead of machines to do this job makes me a bit less angry.
I've had moments like what the other super-recognizer officers did, though. Before I got married and I was on a dating site I shall not name, I immediately recognized one of the secretaries on there from the C-suite, someone I interacted with a couple times but only that. I quickly broke the match, hopefully before she noticed, since that might have been a little weird.
More often I know faces but not names, which makes things awfully awkward at professional meetings. "Hey, ... guy!"