Hacker News Reader

Comment by blitzar | Hacker News Reader

16 deployments for 6 months at rail stations - they must have been off 99% of the time to only scan 500k faces.

For scale 500k people is 2 days of Liverpool Street Stations daily average

A lot of those are going to be the same people day in and day out, the article says 500k faces, not people. You also have to filter out those people where the cameras aren't getting an image good enough to scan.

The article doesn't say, and I haven't seen the cameras, but it's not like you're getting every single person parsing through those stations.

But, yes, the numbers seems low.

They're quite tall an positioned just after the barriers so I would believe they're quite effective on capturing faces. https://images.ctfassets.net/pjshm78m9jt4/29ok6Erlnhk6WIqP3w...

BTP say [0] that "the images and biometric data of those who don’t cause an alert are automatically and immediately deleted." - if that's true, they shouldn't be able to tell that it's the same faces being scanned multiple times so it has to be 500k total scans rather than, say, 500k unique faces across 2 million scans

[0] https://www.btp.police.uk/police-forces/british-transport-po...

You could still delete while maintaining the ability to tell if a face has been seen before, by using some fingerprinting + one way hash function

that’s not deleting the data then. making people play Only Connect to parse out weasel phrases is akin to lying by omission. they say the same thing at US airports which is why I still opt out of the face scan at security. of course they keep hashes of the image and its features otherwise what’s the point

It is deleting data, but you are right that it isn't all the data

The sentence says "the [...] data [...] are automatically and immediately deleted". It does not say "some of the data are automatically deleted". It would be a staight lie to delete some, but not all, of the data and then issue the statement that "the data are deleted". This is not an ambiguous case or weasel-wording: it is an unambiguous falsehood.

If you have two coins and you spend one, you cannot truthfully say "I spent the coins". The definite article "the" means all of something you have in this usage, not a single member of a class of something you have. It would be true to say "I spent a coin" or "I spent some of the coins".

The sentence says "...the images and biometric data...are automatically and immediately deleted". It doesn't say that the information they learned by using the images and biometric data that they collected, such as your name, location, the timestamp etc is being deleted. Who you are and where you were at what time isn't an image and isn't your biometric data.

The data necessary to determine a face had previously been seen are biometric data.

If the name/address is already in the “already seen” list, wouldn’t that be enough to determine the face had already been seen? Or are you saying the name/address should be considered biometric data?

We have no reason to believe a given face is linked to a person's name by this system.

It isn’t exactly like coins, because data is both singular and plural.

If we're going to be precise, the word "data" is plural and its singular is "datum". But treating it as a substance noun does not change the argument I made in any way - "I gave away the water" does not mean "I gave away some of the water (and kept some for myself)".

They're being very specific about images and biometric data, and they're weaseling those words for a reason. If they deleted all data about such scans, they'd say that.

To do the lookup they're still hitting a database somewhere to know you aren't a concern, which is prime time to log that you were seen, when, and where. Then your "images and biometric data" can be "immediately deleted" since they've gotten anything else they want now.

This is just a theory, but I bet I'm not far off.

Digging in it cameras were deployed 21 times for about 4 hours each time ... matching against a watchlist of around 400 faces.

Sounds like it was intended to achieve the outcome it achieved - a near zero false positive rate. If I were a betting person, I would wager the threshold for a match was set comically high.

https://www.btp.police.uk/police-forces/british-transport-po...

OK, so turn them on 100%, scale (your numbers here) the results out to 50 million faces. Now you have still caught 0 criminals and you have 500 false positives.

And they allow people to walk around them with a labelled route, so you see all the criminals just walk around it.

Perhaps not many criminals use train stations during the day, for example, people probably deliver drugs by bike or car so they don't walk around a busy place stinking of weed.

How many known criminals have biometrics of them in the first place?

Of the criminals that do have identifiable biometrics, do they operate in that area?

We also don't know the false negative rate. Perhaps it's very high in order to keep the false positive rate as low as possible.

These are things pulled out of thin air that could easily explain why 500,000 face scans (not people) provided no arrests/identifications.

The biggest reason probably being as you said... it was clearly marked. Honestly I would be shocked if they found a criminal dumb enough to walk past it, but not dumb enough to have yet to be caught.

> people probably deliver drugs by bike or car so they don't walk around a busy place stinking of weed.

A year or two ago there was a "shock horror" piece in one of the far-right extremist tabloids, possibly the Daily Mail, claiming that Glasgow was so lawless "you could get a gram of coke delivered quicker than you could get a pizza delivered".

Shut down by one person's tweet - "yes, because one of them arrives in a Vauxhall Corsa and one arrives in an Audi RS3"

Simple solution: order your pizza from your coke dealer.

100 quid pizza that comes with a "free gift".

I also kind of wouldn't expect that my coke is made to order and would hope that my pizza is.

> I also kind of wouldn't expect that my coke is made to order

I mean I expect you could get that, for the right sort of money.

Maybe not in Brigton, but certainly in the wealthier parts of the South Side.

How many unique faces though

It's called sampling. It works. :)

Reply on news.ycombinator.com