What's the right number of lives to destroy over false positives from an algorithm? Is it some number other than zero? Why or why not?
Contract killers have to be up close and personal in meatspace and leave an actual homicide investigation in their wake.
Someone half way around the world could be contracted to spearfish you, take over your phone/laptop then cause it to download actual CSAM.
The first thing you'd know about it is when you were arrested, and then I don't think there's any kind of defense. If the hackers were very sloppy you might be able to hire a very talent forensics team to find traces of the attack. But beyond that you're pretty screwed and nobody will ever believe you.
It is a button that someone (without any morals and willing to take a certain degree of risk) can push to ruin your life remotely, leaving virtually no trace, and allowing virtually no defense. Nobody is ever going to believe a pedophile claiming that they got hacked.
And the intersection of this with politics and national intelligence agencies (who certainly have the skill and lack the morals) is probably a bit disconcerting in the larger stage.
There are many such buttons. So many it’s horrifying.
Fortunately, the policy choice at issue here isn’t one where there is definitive harm on the track, just risks that can be compared.
I am only half kidding :)
That doesn't make sense. The 'average' driver gets into millions of accidents. I don't want a slight above average driver driving my family around, and I don't want slightly above average drivers around me - especially those who I can't communicate with - by honking, by shouting to get their attention, and algorithms who have no fear of their own life, etc. All software has bugs and I don't want my safety contingent on developers making mistakes. The self driving car must be orders of magnitude above the BEST human driver, and there must be punitive damages in place as a deterrent, to compensate aggrieved parties, etc. At present, self-driving cars are unproven dangerous technology that is being rightfully scrutinized.
The distribution of skills amongst drivers isn't geographically even , nor is it static.
https://worldpopulationreview.com/state-rankings/fatal-car-a...
Either you're making this argument in bad faith because you just don't like self driving cars, or you don't believe in statistics as a concept.
If so, how is curbing CSAM consumption going to prevent children from being raped, exactly? And I do mean exactly. The only arguments I have heard thus far appeal to some vague link between producers and consumers, predicated on the idea that CSAM producers are doing it for the celebrity/notoriety, or for financial profit. Both of these claims are highly suspect, and seem to rest on a confusion between trading CSAM online and paying traffickers for sex with children.
You may be correct, but it's going to take more than a superficial comparison to convince anyone.
Well you just hand-waved the idea the self-driving cars can reduce harm by a "significant" margin - based on what exactly?
I don't understand, though. Are you saying that this is a reason to accept Apple's CSAM-detection?
I wanted to share a (I found) controversial thought: Self-driving cars are the US response to trains. No taxes for railways are palatable, but private vehicles on special roads and profit to be made, great for the car industry. That, IMO, is what drives (pun intended) self-driven cars.
Let’s be clear, there is ZERO empirical evidence that self-driving car technologies available to humanity today are safer than human drivers. Zero.
There are lots of theoretical advantages like “computers don’t get sleepy or drunk,” which elide the many other ways that computers still dramatically underperform people (like recognizing objects).
In contrast, CSAM scanning substitutes for... I'm not sure what. In addition to the risk of false positives, there's also the risk that the scanning technology will be used for other purposes in the future. I could easily see governments forcing Apple to scan everyone's hard drives for hate speech, 3d models of prohibited objects (such as gun parts), or communications sympathetic to certain groups. Once that door is cracked open, there is no closing it.
It is much more worse, current so called self driving cars have human drivers that intervene most of the time and saved the idiot AI from crashes but Elon will not count this near crash as an actual crash.
>but if the goal is to minimize deaths, then we should be more gung-ho about this technology than we currently are.
We should maybe try to maybe to also do the obvious quick fixes at the same time?
it would be much cheaper instead of forcing AI cars on people to force say a drunk/tired/talking on the phone detector , enforce better driving tests before giving licenses to drivers, put a tax for vehicle mass toe promote less heavy cars, enforce speed limits with tech. Do you think Bob will prefer to be forced to buy an expensive self driving car to reduce the car crashes stats or better to buy a safety device(black box) that he must install in the car.
“Destroyed lives” from false positives are at this point hypothetical. Child abuse is not. It’s fair to be concerned about false positives and ensure the system handles such failures appropriately. It’s also fair to directly intervene in the widespread circulation of CSAM.
It is better that ten guilty persons escape than that one innocent suffer. -- Blackstone's ratio
It is better that ten children get sexually abused than one innocent person comes under suspicion
/s
You'd rather 10 children be victimised than 1 person falls under suspicion? Ok...
Now you’re just arguing in bad faith.
In what way? I'm just characterising you argument from the point of view of the victims of the crime, who you appear quite happy to throw under the bus.
Well yeah, mostly because I don't understand what you mean be "revictimization", but I can easily imagine what it's like to have such suspicions leaked to the angry mob.
How about we pivot to intimate image abuse (revenge porn). The argument is the same. The victims are re-victimised every time their intimate images are shared, for the victims it's always the same stress, shame, embarrassment, etc.
However apparently the damage is already done so there's actually no issues according to the other user so what's all the fuss about...
I guess the difficult bit for me is that you talk about throwing victims under the bus, and letting them be "victimized" all over again, but I just don't understand how this solution helps victims at all. Here's the example in my head: imagine you're a victim of revenge porn. The police, using technology like the stuff Apple has developed, catch one random guy (out of hundreds or thousands more) in possession of this porn. The police probably won't tell you that they caught this guy. How does this event help you?
> there's actually no issues
It's not that there are no issues: the distribution of CSAM (and of revenge porn) are terrible crimes, and iirc they're increasing in prevalence. I just think that Apple's solution doesn't help victims, and seriously discomforts millions of people.
As you've seen, many people do see this as privacy invasion. The feeling that folks somewhere far away might be able to see your photos (even low resolution thumbnails) without your knowledge isn't a good feeling to inflict on the world. And the feeling that your phone could report you to the police at any time isn't a good feeling either, even if you know you've done nothing wrong.
I value privacy above most things, for example you'll not find me on any social media platforms and the ones you do don't link together. If these people valued their privacy that much they'd not be using Apple devices and services in the first place is my thinking. I'm pretty sure Apple have full access to anything in iCloud which for most users is where all their stuff sits, where's the expectation of privacy?
Whether a rational person would accept chance X of wrongfully being convicted of a crime to decrease the chance of being a victim of crime by Y obviously depends on the values of X and Y.
Is there any evidence such things actually reduce the production of CSAM? Or is it like the war on drugs where drug production is as high as it's ever been.
Therefore, the argument is one of degree. I agree with you that Apple's CSAM-detection is going too far, and this is what we should be articulating. Chanting "not even one" is not particularly convincing, nor sensible.
Also… The goal of law enforcement is not to vengefully destroy the lives of people who commit crime, but that’s a whole different can of tuna. Still worth noting, because it hints at a larger problem about how we approach heinous crimes.
Now you're probably going to then cry about incentivising or normalising CSAM - but that's a different argument. And if you then try to argue that the normalisation of CSAM would somehow encourage people to abuse children, well then you're really off into the zero-evidence weeds. Go look at porn research (the actual Google Scholar/JSTOR/Elsevier kind), and you'll see that almost everybody who looks at porn neither wants to nor would actually do what they see in porn, if they were given the opportunity. Surprise, surprise, most people wouldn't actually get gangbanged/bukkake'd/fucked by their sibling, mom, dad, grandpa/pooped on by their next door bespectacled red-headed neighbour, etc.
Nor is there any evidence that inadvertently coming across CSAM turns people into pedophiles (news flash: pedos were turned on by kids long before they were ever exposed to CSAM on the internet), and porn itself is almost invariably used for fantasy or as something wholly unrealistic that people get off to precisely because it's unrealistic. Even though it might be unsavoury to do so, we could follow this reasoning to its extreme but undoubtedly true conclusion and state that there are individuals who get off to CSAM notwithstanding that they would never themselves abuse children.
So to recap, 0 children would be saved from harm because the CSAM itself is ex post facto; it wouldn't de-incentivise CSAM because the demand and markets for CSAM and pedophilia existed long before the internet was a thing, and pedophiles will find avenues around dumbass implementations like Apple's scanning (TOR, anyone? not using an iPhone/Mac?); and, finally, just because somebody looks at CSAM doesn't mean they're an actual pedo or would ever harm children themselves. The fact that possession of such material is illegal is not to the point - Apple is not the police, and the police and other executive agencies need warrants for this kind of thing (in common law countries this notion is more than 400 years old).
Meanwhile, we know the false positive risks are not insignificant - look at the white papers yourself, or just look at the numbers that smart people have crunched. The best part is that even though Apple says the false positive rate is 1 in a trillion accounts, people's photo libraries are exactly the sort of thing you can't extrapolate statistically. Maybe your Aunt June really likes photos of her nephews in the pool, and she's got a library with 300 photos of her nephews half naked and swimming when they visited her last year. Apple has no fucking clue whether they would or would not trigger its scanner, because it currently does not have access to Aunt June's unique photos to test vs its database. Apple quite literally doesn't know what the fuck will happen. I and many others find that abhorrent when you consider the effect that even the mere accusation of pedophilia has on a person's life. And that isn't even to start the discussion of what kind of precedent on-device scanning would set for other subjects (political and religious dissent, for example - if not in America then in places like China).
Apple and everybody else can fuck right off with this Orwellian shit.
I do hope you get whatever support you need for something that's apparently affected you. Take care.
Circulating images of a minor child engaged in sexual abuse do not constitute an ongoing harm to that child? That's a fascinating viewpoint.
> Apple and everybody else can fuck right off with this Orwellian shit.
Right along with people who think child abuse images should be okay to keep as long as you aren't the one who made them.
But let's assume you're right.
You think that pedophiles won't find ways to share content even if every tech company in the world implemented this? You think pedophiles don't and won't have terabytes of child porn backed up on hard disks around the world that will be distributed and circulate for the next millennium and beyond, even if it has to be carried around on USBs or burned to CDs (which aren't exactly amenable to CSAM scanning), and then saved to offline computers? Put another way, even if the internet shut down tomorrow, plenty of pedophiles around the world would continue jacking off to those images and sharing them with their buddies - you don't need the internet for that.
Further, even if you could convince me that it's harmful in the sense that I understand the word, I'm not sure I'd ever be persuaded that the amount of harm could be sufficient to outweigh the harms that would result from the scanning itself.
Happy to listen, though.
When you do it, it does. When law enforcement does it, apparently not...
By this "moral" reasoning you're also fully supportive of revenge porn.
But prosecuting third parties for having it is asinine, IMO.
The idea that all of this effort (and all of the direct discomfort inflicted on Apple users) will do anything to stop child abuse is just as hypothetical.