Hacker News Reader

Comment by lewdev | Hacker News Reader

Why ban it? It's not the facial recognition that's the problem; the problem is that the justice system believes that it's enough to convict someone of a crime. It's hard to believe that prosecutors can be like, "The computer says it's you, so you're guilty." Especially when there's probably a known margin of error.

View on HN

because that's the only option. There is no scenario in which it's allowed in a narrow set of circumstances that isn't going to get abused by law enforcement. Either you ban it, or law enforcement is going to find ways to abuse the right to use it, including appointing people whose job it is to say "this was fine".

If this was a reasonable society, with a reasonable police force: absolutely, narrow definitions of permitted use should be reasonably fine. But the US isn't, and the US doesn't, so in the US, it can't be.

(but then you look at countries where it might be fine, and it turns out that they don't need facial recognition to have a decent enough catch rate, kind of making the whole thing moot. No police force should use it, for wildly different reasons)

Except this is allowed in a narrow set of circumstances - to look for missing children. So they can probably dragnet everyone they see to find the missing children or abductors. Now the residents need a way to prevent the police misusing it by sneakily letting it find rapists and murderers.

Welcome to the security vs. privacy debate. Glad to have you here, please read the pamphlet and please don't resort to a "won't anyone think of the children" argument. Because the statistics around facial recognition just don't work for that argument. However many children you think are saved with this: it's nowhere near that figure, if not flat out zero.

There's far more than the racial component. Do you want the government tracking you? It helps authoritarian rule (who always victimize minorities first and more harshly). I think a big problem with this topic is that we're only focusing on the "it doesn't work" aspect and not even considering the "but what if it perfectly works, do we still want it?" question. Humans haven't historically been great at dealing with questions that rely on foresight, but that doesn't mean we should ignore those questions.

Exactly. No one wants to live in a word where 100% of crimes are prosecuted 100% of the time. Not really.

Do you want to receive a citation every time you exceed the speed limit by 1 mph? Ever pull off into a parking lot to check your phone or look for something in your car? Trespassing. Ever lose your temper and yell at someone, even briefly? That's assault in most jurisdictions.

Unfortunately, the absolutist nature of the criminal justice system means it does not do a good job of integrating nuance.

Take for examples:

How unimpeachable police testimony is.

How hard it is to overturn a conviction.

How hard it is to dislodge forensic techniques that are known to be bunk.

I could go on. Probably the best correlate we have to this is the rejection of the polygraph and boy in retrospect are we just lucky about that. If King County manages to be the first in such an action against facial recognition they may be worthy of a great deal of praise.

> It's hard to believe that prosecutors can be like, "The computer says it's you, so you're guilty."

Why is that hard to believe? The article very explicitly and clearly quotes a lawyer who says that is pretty much exactly what is happening.

And regardless, even if there weren't a direct quote from a lawyer on this topic in the article, it still shouldn't be that hard to believe that a computer's output is going to influence the court cases at all. That's not how humans work, for one - people will include everything they 'know' about a situation when making a decision. You can't have a computer producing output and then people ignoring the output and then calling the computer system useful. If people are ignoring the output, then why have the system? And if people aren't ignoring the output, then, the (FALSE) output is being considered in decision-making of which innocent minorities are to be jailed.

Because even the hassle of being hauled into jail for someone to realize the software got it wrong can cause someone to lose their job, and since the software remains flawed it’ll keep picking them!

Because people don't enjoy the government knowing what they're doing all the time? It's a bit creepy that in places like London the government knows where everyone is 100% of the time.

exactly. There must be a person in the loop. Facial recognition is just a tool and a very useful one. Banning it is silly virtue signaling. A proper way to address all sorts of possible problems with it is by requiring qualified people in critical decision points within the overall system.

That's right. But the justice system can't be fixed. So we prevent it from using these tools.

It's to prevent some halfwit police officer talking to a halfwit judge and prosecuting some innocent fellow. #donttreadonme

Reply on news.ycombinator.com