Hacker News Reader

Comment by AnthonyMouse | Hacker News Reader

> It could be bypassed by modifying the file slightly, but the hashing could also get more sophisticated.

Which is the same thing as saying "have a higher false positive rate," and then we have the problem.

This is exacerbated by the same kind of system having the same application for copyright infringement, which means there will be a large community of people dedicated to defeating it. Every time they find another workaround, you have to increase the false positive rate. It only ever goes up.

This is why everyone hates Content ID. The pirates know how to defeat it, meanwhile it regularly flags public domain works and recordings of randomly generated white noise to say nothing of fair use etc.

The solution to revenge porn isn't platforms, it's why isn't the ex getting called "defendant" by a judge for making the recording without permission?

View on HN

> The solution to revenge porn isn't platforms, it's why isn't the ex getting called "defendant" by a judge for making the recording without permission?

The court system's responsiveness to sex crimes could be greatly improved if we had the collective will to do so. But the problem is both in the courts specifically (because of who the judges are, and who the legislators who write our laws are), and also the wider world. Victim blaming and dismissal/minimizing of the harm inflicted are par for the course everywhere, in the justice system, in the general population, and in many subgroups. (You'll see them here on HN all the time.)

> But the problem is both in the courts specifically (because of who the judges are, and who the legislators who write our laws are), and also the wider world.

I don't think this is it. It's a matter of resources. Police and courts are already so overburdened that it's hard to get enforcement on anything that isn't practically life or death. Have you or a friend ever had your apartment burglarized or a vehicle stolen? The police will show up, usually after several hours, take a statement, after which you'll never hear from them again. In fact I don't know a single person in real life who was a victim of a burglary or auto theft where the perpetrator was caught, save for those cases like a car jacking where significant violence was involved.

When the police and district attorney in any given city in the U.S. are looking at double-digit unsolved homicides and God knows how many unsolved rape cases, I don't think they're going to put a high priority on the fact that there's a naughty movie out there that someone willing filmed themselves in and is now regretful about. This does not mean folks don't believe these things matter, because they do. It's that relative to what the justice system is already dealing with, these things are much less important.

When the NYPD have a guess at who did around 30% of murders, compared to the London police at 90% - and despite more funding than a small army - you have to wonder if the problem is purely resources. In fact, the vast majority of police are assigned to tasks like traffic enforcement, in part because that brings in more reliable revenue.

(I'm using "guess at" to refer to clearance rate because if the cops tell a prosecutor they think you murdered someone and the prosecutor says it was obviously someone else so they're not charging you that's generally counted as cleared.)

> When the NYPD have a guess at who did around 30% of murders

This is false.

NYC homicide clearance rates for 2021 look to be over 70% for the year, with Q4 2021 at 78%. (worst quarter - 55%, best - 86%).

https://www1.nyc.gov/site/nypd/stats/reports-analysis/cleara...

> In fact, the vast majority of police are assigned to tasks like traffic enforcement

NYPD is not known exactly known for a particularly large amount of traffic enforcement, to the point that I think you'd find most residents wish they did more of it.

30% getting away with arguably our worst crime seems like it doesn't bode well for the clearance rate of lesser crimes.

I wonder if a large contributor to the higher London clearance rate is related to socioeconomic factors or surveillance laws/systems.

At a national level, the clearance rate for murders is ~61% per https://ucr.fbi.gov/crime-in-the-u.s/2019/crime-in-the-u.s.-... . Property crimes are <20%.

The FBI's UCR Data Explorer used to give you an easy to use breakdown of crimes and clearance rate at the national/state/metro/city level, but unfortunately they redesigned it and now it is both harder to use and presents less data. I guess that is one way to hide data that isn't favorable? There seems to be more emphasis on the danger presented to law enforcement officers, but the #1 cause of death for officers by a large margin, per https://www.odmp.org/search/year?year=2021, has been COVID.

Yeah, I just gave the 30% unsolved in response to the comment about NYC having 70% clearance rates on homicides.

How much of that clearance rate is railroading innocent people through the justice system and an overworked underfunded public defenders office?

You're right, thank you. I mis-remembered, in NY it's major crimes that's around 30%.

Isn't London well known for having way more surveillance cameras than most cities?

Maybe in the early 2000's, when stories about it kept making evening television, but nowadays, the rest of the developed world is roughly on par with it.

I agree that lack of resources is a contributing factor.

> I don't think they're going to put a high priority on the fact that there's a naughty movie out there that someone willing filmed themselves in and is now regretful about.

This characterization of revenge porn is exactly the victim blaming and minimization of harm inflicted that I was talking about. Consenting to film a sex act is not consenting to have it published to the wider world.

> This characterization of revenge porn is exactly the victim blaming and minimization of harm inflicted that I was talking about.

While it certainly isn't minimal in terms of the harm inflicted, it very much is minimal compared to the kinds of things that police departments and courts are already dealing with.

There's also a weird and very unproductive ethos that's evolved around the concept of victim blaming. Something my not be your fault, but that doesn't mean we're all free to be careless with zero chance of consequences. If I'm walking through a neighborhood at night that I know to be dangerous, it would be very foolish of me to be carrying several thousand dollars of cash on my person. That does not mean it's my fault if I get mugged and lose my money - it's purely the mugger's fault. However, my actions were stupid and I should have known better. Both those things can be true.

Many judges, police, and prosecutors agree with you that the victims "should have known better" and that their cases are not important in the grand scheme of things, which is why it is more difficult at every step of the way to get any sex crime through the court system.

Many in the media also agree with you. So do many in the wider society. (Including many in the ridiculously male-dominated HN.)

All of these factors contribute to why if a victim even goes to court, not only is the likelihood of actual consequences for the perpetrator absurdly low, but the experience of going down that path will be horrible. Instead of sympathy, they get an avalanche of contempt and shame dumped on them.

This is the thin end of the wedge fallacy. It's disproportionate to load of the possible extreme outcomes of a - true - idea on that idea. The idea is fine. Other things you list that are bad and not that idea are not fine.

Your reasoning is too convoluted and unclear to argue against.

Can you be specific? What do you need clarifying?

Low convictions isn't due to victim shaming, it's due to lack of evidence. It's generally a he said she said situation with no witnesses or evidence.

We're talking about revenge porn here. Who uploaded a file is not he-said/she-said, there are going to be logs. And with regards to the question of whether there was consent, "I don't have it in writing but she said it was OK to upload our sex videos to pornhub" should be treated with extreme skepticism.

Piggybacking on this point, in a civil lawsuit you only need prove that a "preponderance of the evidence" (51% or more) supports your position. (The burden is greater for criminal cases, of course).

“greater” is an understatement. It’s “beyond all reasonable doubt”.

Meanwhile, if the video is of consensual sex, the main problem is really that society looks down on people having sex. If it had all the reputational impact of a cooking video, she’d be less likely to care (though she certainly can care about anything she wants).

>And with regards to the question of whether there was consent, "I don't have it in writing but she said it was OK to upload our sex videos to pornhub" should be treated with extreme skepticism

I don't think you're gonna find many cases for non-professionals where they had consent and it was given in writing.

I don't actually know, it's just a guess given observed human behaviour.

If that's the scenario, why doesn't the burden of proof fall on the person doing the upload of mortifying materials? Shouldn't the uploader protect themselves by getting permission in writing, similar to how victims are constantly badgered to take every action to protect themselves?

>why doesn't the burden of proof fall on the person doing the upload of mortifying materials?

Because the people of your country didn't make your politicians codify that into your law.

>Shouldn't the uploader protect themselves by getting permission in writing, similar to how victims are constantly badgered to take every action to protect themselves?

I dunno, do you get written permission before having sex so you'll be sure you won't get falsely accused of rape afterwards?

People will only follow what is culturally ingrained or enforced on them, and getting permission in writing isn't a thing right now.

As someone with a friend who was on a jury that was overseeing a sexual assault casae, this is categorically false. You can do this by the numbers too, which show that it is false.

There’s also the small issue of the First Amendment. Anyone that thinks these revenge porn laws are going to get more than 2 votes with this Supreme Court - that’s pretty 1A absolutist - isn’t thinking clearly.

The first amendment wouldn't come into play here. Copyright and consent laws would.

IANAL. Copyright and all associated rights of a photo/video by default belong to the photographer/videographer. (https://photofocus.com/business/photographer-or-model-who-ow...)

There are multiple factors that are involved, but generally speaking, consent laws end at the production of the photo/video. Distribution falls under copyright. Unless the subject was surreptitiously recorded with the intent of distribution or blackmail or was trespassed upon for the recording to be made, there is no copyright or consent violation.

And yes, the First Amendment does apply. In fact, there have been cases where such laws have been struck down on the basis of the First Amendment (https://thehill.com/changing-america/respect/equality/475968...).

There has long been a debate in the judiciary about the extent to which there exists a privacy carveout in First Amendment rights against disclosure of private information. What this means is that depending on judicial philosophy, either any such law must be carefully tailored (which seems reasonable and desirable), or any such law is inherently unconstitutional (which seems to me to offend common sense):

https://mtsu.edu/first-amendment/article/1532/revenge-pornog...

> Under current First Amendment guidelines, any laws regulating revenge pornography would have to be narrowly drawn and made in pursuit of a “compelling state interest.”

Unfortunately, I agree with tiahura that there's little question how the current SCOTUS majority would rule — it's unlikely that they will find a "compelling state interest" in protecting the victims of revenge porn. The result will be that any victim without the resources and circumstances to bring and win a civil suit will be SOL.

Whether there is a debate or not is not what I was addressing. I'm simply pointing out that, under current Federal law, that's not the case.

But If you want my opinion as an individual, I wouldn't agree with a privacy carveout. Firstly, because "revenge porn" doesn't qualify under currently prescribed limitations (defamation, incitement to imminent lawless action, obscenity, "fighting words", etc.) and I'm skeptical of adding any more. Secondly, most "revenge porn" legislation is overbroad and does not require a mens rea to prosecute. They are strict liability laws in all but name. They amount to prior restraint and violate the requirements of content neutrality and strict scrutiny towards limiting speech. Thirdly, these laws can be used to silence journalists reporting on relevant events (such as the Katie Hill debacle).

"Revenge porn" laws do nothing but coerce individuals into becoming bearers of secrets. Even if they had no interest in doing so to begin with.

I would also like to add that not every person charged under "revenge porn" laws is the stereotypical ex-boyfriend uploading sex tapes to Pornhub. It's in situation like Bethany Austin's where rights are needed most. https://www.cato.org/blog/picture-worth-thousand-words-priso...

Do you support any consequences for revenge porn?

Before one can answer that, one would have to know what a proper legal definition of "revenge porn" is.

If you're referring to today's overbroad interpretation that has swept legislatures across the country, then I would say there should be no legal consequences at all. I'm firmly convinced those laws should be repealed and the cases involved reheard.

However, if you're referring to something much narrower and applicable with in the framework of the First Amendment as it is currently, then I could consider a limited and proportionate civil fine appropriate.

For example: a man records a video of his girlfriend performing a solo sex act. Before the recording, they discussed the matter via text and it was clear from the written record that the video was only for their own use and would never be shared with anyone else. Later, he uploads the video to PornHub, in direct contravention of her wishes and their previous understanding.

Should there be consequences?

Sounds like a contract violation to me under current law.

As another commenter examined, that sounds like a contract violation and would already incur a civil fine. In such a case one would have to prove at least three elements:

* that there was an oral or written contract

* that said contract was made by two informed and sober parties

* that one party uploaded the video to Pornhub against the wishes of the other and not anyone else.

Well, although civil suits are an utterly inadequate stratagem given the scope of the problem (which includes millions of victims who are not practially in a position to pursue and win a suit), at least it's something.

Criminalizing speech would not solve the issue. The example you cite involves two private parties engaged in a contractual dispute. There are millions of those every week. Barring the involvement of criminal behavior such as theft or extortion, a civil court is the only forum to adjudicate these matters.

The point of a court, criminal or civil, is not to invent legislation favorable to the most sympathetic party. The point is to interpret the law within the constraints it was written under.

As a practical matter, throwing the matter to civil court ensures a vast epidemic of revenge porn. There are a lot of people who don't think that's a big deal. But there are also a lot of people who do: 46 US states have have passed laws criminalizing revenge porn. There are lots of problems which are theoretically resolvable by civil courts but where the results would be too inefficient or incomplete and so criminal statutes have been passed. Criminal legislation to stem revenge porn is happening, even if you would rather it not.

"Revenge porn" laws have been on the books for about a decade or so and haven't slowed it down at all. In fact, all they've done is push it further underground. The only practical effect has been on the media of distribution, not if or to what degree "revenge porn" is distributed. That most legislatures across the country favor over-broad criminalization does not make your claim meaningfully representative of the people who are subject to these laws or of the courts who will ultimately adjudicate the issue. Every case is tried on its own merits, not ones prescribed by popularity or mere practicality.

> In fact, all they've done is push it further underground.

Deplatforming revenge porn is a win. Having less revenge porn on PornHub is a win.

You're welcome to disagree, just as advocates for Voat, Parler, etc. would — and as companies such as Manwin who would prefer to exploit the victims of revenge porn for profit would.

> Every case is tried on its own merits

The vast majority of civil cases don't even become cases let alone get tried on their merits. For a huge swath of the population who do not have the expertise, the resources, the time, or the life circumstances to obtain a lawyer and pursue a case, civil courts are not practically accessible. In contrast, while reporting to criminal authorities has all kinds of issues including that those authorities may be unsympathetic, the criminal system at least does not place such outrageous demands on those who report a crime.

Holding civil court aloft as a sufficient remedy for revenge porn may not be disingenuous if you actually think that a typical victim could sue, but it's wildly unreasonable.

> Deplatforming revenge porn is a win. Having less revenge porn on PornHub is win.

PornHub didn't deplatform "revenge porn". It nuked more than half of its vast library of legal content under outside pressure and questionable allegations of hosting what may or may not have been "revenge porn". Even when there has never been a law in Canada that has defined in any way, shape, or form what "revenge porn" is. Millions of videos became martyrs for a cause of dubious effectiveness and even more dubious reasoning. On a practical level, the indiscriminate purge of content did nothing but punish innocent users. If you think that substituting due process with such massive collateral damage is a "win", then one cannot conclude your opinion to be reasonable at all.

> The vast majority of civil cases don't even become cases let alone get tried on their merits

For vast majority of cases, civil and criminal, filings fail to pass muster for multiple reasons. Common ones are Failure to Make a Claim or Lack of Subject Matter Jurisdiction among other legal errors. Many claimants have a chance of refiling, however legal standards shouldn't be lowered on grounds of sympathy based upon hitherto unsubstantiated claims. That only encourages witch hunts and fallacious charges.

I'm not sure how you reached the conclusion the first amendment applied. The MA SC ruled, not that Casillas had a first amendment right to post "revenge porn", but that the law, as drafted, was too broad.

We have insanely well established laws and precedent regarding expectations of privacy, and laws regarding two party consent. Other states have successfully narrow laws (see CA).

The First Amendment thing is red herring nonsense, sadly boosted by one zealous and incompetent legislature.

Let's not claim that police are ONLY investigating rapes and murders all day. There's a lot of other time spent on stuff like pulling people over for expired tags or other more-minor-than-this shit.

> expired tags

There are really only two things the government actually cares about:

1. staying in power

2. collecting taxes

So yeah, they're not going to overlook expired tags.

Traffic stops that let them seize cash will definitely be a priority. So even your broken taillight means revenue.

This is very true. I lived in a fairly high-crime city for a while. The police never really investigated burglaries or car thefts because of a lack of resources. But man, leave your car parked on the street for a couple of minutes after the meter expired, and BAM! a cop would appear out of nowhere to write you a ticket.

Police have all the time in the world for drug crimes and pulling over people because “they look suspicious”.

> In fact I don't know a single person in real life who was a victim of a burglary or auto theft where the perpetrator was caught, save for those cases like a car jacking where significant violence was involved.

Me! My car was broken into and my audio deck was stolen. I had to report the crime to the police for insurance purposes. I didn't even give a proper description of my deck but the police called me back a few weeks later saying they found it inside a stolen car and mailed it back to me.

But maybe violence had occurred, .. maybe some shaddy shit went on with that stolen car that led them to investigate it.

That's fantastic that you got your stuff back. I'm very happy that my absolutist statement was proven wrong in this case.

"The court system's responsiveness to sex crimes could be greatly improved if we had the collective will to do so. But the problem is both in the courts specifically (because of who the judges are, and who the legislators who write our laws are), and also the wider world."

Responsiveness could be improved on just about any crimes if we had the will to do so. Just look at how many murders go unsolved (about 38% unsolved). Then look at the difference in response the system makes when it's some nobody whose murder didn't even make the news vs a cop/judge/[system participant] who was murdered.

Frankly, who the people are that comprise the system are doesn't matter. The system is not appropriately sized to handle current criminal and civil case volumes. In many cases you wait years for trials. A majority of people accept plea deals, including the innocent. There's very little justice in the system. Between 2 and 10 percent if the incarcerated are wrongly convicted. I assume this is even higher for plea deals that avoid jail time.

And on victim blaming, both "sides" do it. It just depends on which issue it is.

> Frankly, who the people are that comprise the system are doesn't matter.

The fact that the system cannot cope with the volume of cases means that the people who are in a position to prioritize which cases move forward is critical. That's the very nature of selective enforcement.

And that's largely irrelevant to the topic at hand (revenge porn) as emotional crimes will always be at a lower priority than violent and physical crimes.

Selective enforcement (law enforcement or prosecutorial discretion) is a larger issue that should be addressed. By unevenly enforcing the law, they are likely exhibiting biases, they will end up eroding what trust there is in the system, and it will allow bad laws to continue to stand (if everyone was being screwed over then they would demand the law change).

Basically, we should be fixing the system, not just shifting priorities based on some individual's preference.

Revenge porn often has devastating impact on the emotional health of its victims[1], with half reportedly considering suicide. But it is not strictly an "emotional crime", since publication loops in many people beyond the victim and the perpetrator — there are frequently reputational, professional, and economic consequences for the victim as others become aware of the published materials.

There is a huge variety of potential problems which require the attention of legislators, law enforcement, and the courts, and priorities change all the time. It's not as though the only recourse we have is to increase the system capacity.

[1] http://jaapl.org/content/44/3/359

"there are frequently reputational, professional, and economic consequences"

Which can and should be handled under civil cases.

You can extend many these problems, health effects, and the thoughts of suicide to many other issues too - victims of rape, assault, harassment, burglary, etc.

So we end up with violent and property crimes which also contain these emotional damages, vs just an emotional crime. You would need to prove that this emotional-only crime is somehow more important and these other crimes that contain those damages plus violent or property damages. It's not that there isn't harm done by revenge porn, but that the damage done by others is objectively worse because it contains additional elements.

"It's not as though the only recourse we have is to increase the system capacity."

Then please, do explain. Because the only logic I see is that you ignore other crimes to pursue these, or you increase capacity. And frankly, ignoring any crimes undermines the entire system due to unequal enforcement and ostricizing some victims but not others by not giving them the recourse due under law.

> Then please, do explain. Because the only logic I see is that you ignore other crimes to pursue these, or you increase capacity.

For example: There's a difference whether the Attorney General is Bill Barr or Merrick Garland — and it does indeed involve choice as to which cases to pursue.

The same holds true for many other offices in law enforcement and the courts. Who the district attorneys are, who the judges are, who sets budgets for police and who decides how those budgets are allocated and law enforcement resources are deployed... all of that matters.

Which isn't contrary to what you quoted - they are ignoring some laws in order to pursue others. Unequal enforcement is wrong and should be ended. This should be fixed through increasing capacity.

Given that property rights usually end up trumping human rights in US law, DMCA'ing revenge porn might work better.

I wish I were joking.

Property rights are human rights.

Suppose I dropped by, tossed you out of your house into the street, and took possession of everything you had.

You'd certainly feel your human rights were violated.

> Revenge porn often has devastating impact on the emotional health of its victims[1], with half reportedly considering suicide.

This obviously is a very important issue to you. While massive enforcement that punishes offenders and gives justice to victims would be the ideal, from a technical and legal perspective, that's very hard to do. Why not take the less morally righteous, but infinitely more practical, path of convincing folks to stop making porn movies of themselves? If our main goal here is to alleviate suffering, the discussion should center on how best to prevent this from happening, regardless of what the morally correct solution would be in an ideal world. While there is nothing morally wrong about making a sex video, it's a classic high-risk, low-reward activity. If everything goes perfectly, you have a video that you will likely never watch again. On the other hand, that video may find it's way online and ruin your life.

> Why not take the less morally righteous, but infinitely more practical, path of convincing folks to stop making porn movies of themselves?

A ton of reasons[1], including that such preaching is ineffective and so the only impact of doing it is to justify excoriating the victim later and excusing failing institutions.

[1] https://en.wikipedia.org/wiki/Victim_blaming

Who said the video was made without permission? That is a rare case. The far far more common situation is where both parties consented to the video at the time, but later one side uploaded/shared it without permission.

Just look at all the pictures/videos from that iphone celeb hack a few years back. They all consented at the time. Many even took the pictures themselves (selfies, very common with teenagers too). The harm came far later when the material became public.

I don't know where you got "made without permission" from in the post you're responding to. This discussion appears to be about "uploaded without permission" in terms of the OP and the one you directly replied to.

And then even in that even-more-obviously-bad made-without-permission case, the argument here seems to be "we can't make tools to handle these [assumed rare] cases because false positives"?

Is the harm of an uploader having to follow up with a human about a false positive seen as WORSE than the harm of having these videos uploaded over and over and over...?

> Who said the video was made without permission?

Not me — that was a quote from the comment I was responding to. :) I agree with the implication that the question should have been about publishing rather than creating.

> Who said the video was made without permission?

So something the legal system doesnt take into account is people changing their minds.

Why do people changed their minds, because their chemistry changes, this is seen as aging or might even be a short term change like getting drunk, going through stress, even having people you dont want to know, knowing about it. So it shows even the law isnt perfect.

I would say that the legal system is open to revokable consent, it just does not currently have the capability of having that record

Similar to how the legal system recognizes that verbal business contracts are valid (a form of consent), it just does not have a way of proving them essentially rendering them invalid.

So it boils down to exploiting peoples knowledge and when you look at the dire educational system in many parts of the world, all I can conclude from it, is its just good enough to make you a slave.

I think law and sofware are very similar in this. Basically read the parent post, except replace words of tech with law and court process ones.

As you increase the number and the precision of laws, you inevitably increase false positives.

First you will hit the resource issue. Secondly, you will see laws being weaponized.

If you can pay a lawyer 1k/hour, you can get anyone arrested. They will find a way. More laws you have, more ways they can use. See the case of Steven Donziger: https://twitter.com/SDonziger

I don't follow the argument that the false positive rate necessarily goes up if you notice some successful attempts to defeat your system and update your system to detect some attempts. Is this some statistical law you're referencing, or do you just mean that people tend to do naive updates to these detection systems to make them more sensitive (not necessarily "more sophisticated")?

It’s just a convenient and lame excuse. Slippery slope stuff is the worst/laziest argument.

YouTube has definitely been flagging more false positives over the years, as pirates have gotten smarter. When you broaden the algorithm to flag more bad content, you also end up flagging more good content unless you narrow it in some other way.

But the issue with YouTube etc. isn't flagging innocent people, it's flagging innocent people and then rejecting their appeals. It should be pretty easy for PornHub etc. to get videos "verified" by getting explicit consent from all parties filmed, so that they cannot be flagged and removed from the system.

In fact, I really don't see any reason against a law requiring all porn sites to require verification for all videos (w/ retroactive revoke on the hosts). Nobody's being censored, and if you agree to be in a porn video you kind of waive your right to anonymity.

> Nobody's being censored, and if you agree to be in a porn video you kind of waive your right to anonymity.

Then the host gets hacked and your government ID gets published as a creator of amateur pornography. The risk of that is de facto censorship by chilling effect.

The thing is, this was technically required in the US under USC 2257 for the past two decades: https://www.law.cornell.edu/uscode/text/18/2257

Although I think that says "produces" rather than "publishes" or "distributes", so pornhub themselves aren't liable but the boyfriend "producing" the film has committed an offence. Not that there's any will to enforce this.

Closer to home, Scotland explicitly banned revenge porn, and has actually achieved some prosecutions.

> Is this some statistical law you're referencing

There is a natural law in pattern recognition governing the split between sensitivity and specificity in scenarios with uncertainty, though I don't know if it has a special name. Catching more edge cases in a poorly-defined group requires allowing the inclusion of more things that look a lot like edge cases, but cases on the edge of a group are by definition similar to things outside of the group. The more you want to eliminate false negatives the more likely you are to produce false positives.

In binary classification, this is basically a precision/recall trade-off. More precision (ie, fewer false positives) necessarily means less recall (ie, more false negatives). Classifier statistics like AUC basically tell you how good the classifier is in terms of how bad the trade-off is.

Adversarial situations - any time humans are trying to trick your classifier, in other words - will always make the trade-off worse.

More to the point, I don't buy that a higher false positive rate isn't worth it in this application. From a social perspective, the inconvenience of the false positive rate seems so small as to be impossible to measure on a scale weighed against the harms of uploading revenge porn.

I know the companies are optimizing for other metrics where the inconvenience might be measurable but I think that's a failure of regulation then. The platform could be made liable like we do with copyright material, for example, and then like YouTube they would be incentived to accept a higher false positive rate and automatically filter repeat offenders out.

Isn't a high false positivity rate favorable to revenge porn being uploaded? Shouldn't it be on the uploader to prove that a scene depicting sex is legally filmed rather than assuming that it is?

> Shouldn't it...

Yes. Will it? No.

What about this: Take their IP. Make sure that, to them, it displays as uploaded. They can watch it. It registers natural-looking traffic. It does whatever else is necessary to satisfy them that it's on the internet (besides Google SEO; I'm not sure if that's expected). Otherwise, it doesn't exist and doesn't show for anyone else. In other words: blackhole it, shadow-block it. I feel like that has a chance of working.

IP addresses can be obscured.

This inhibits journalists or activists that may want to obscure their identity.

In the context of porn, perhaps the person uploading the video is the performer and does not want to attached their real ID to the performance.

We had a problem with anonymous trolls on a message board I managed decades ago. We created a simple profiling system and combined it with shadow-banning. It was highly effective. Surely in the decades since and with the profitability of profiling, this could be done here too.

In the case of trying to block a specific video, it could be even more effective, because as soon as that video gets uploaded, you can assume the 'new' person is the same person as the existing profile—it doesn't really matter if you're wrong because even if it's two people, you're trying to stop them both.

I'm not sure how this would in any way inhibit journalists or activists except that, they too, would not be able to upload the disallowed video to your site.

I don't think there are perfect answers here. Someone might use the same IP address of a bad actor (e.g. they share a VPN provider). One person may use the provider to watch Netflix, another person may use it to hide their identity to do bad things.

From my limited experience in this space (I moderate a tiny social app that has problems with bad actors), IP address blocking had high false positives and low false negatives.

Even back then using IP address only wasn’t how we profiled. If the poster meant literally basing this on IP only, then yeah, I sort of agree.

But a false positive here just means someone can access the video who isn’t supposed to access it. Without the system in place anyone can access it.

I would do something like this, should work for any type of image.

Original -> Upscale and Downscale.

Alter the frame sizes ie 4:3, 1080p, 4K

Encode all video formats in all video codex's.

Make stills for timelines which are likely to be interesting.

Use a variety of background filters like those seen in Zoom calls to obscure backgrounds and then use all the stills with and without backgrounds, and also alter the colouring/whitebalance.

Hash Stills and you will have loads of stills!

Train an AI to look for the stills allowing for colour changes like a slight filter to alter the colouring, or maybe do it in black and white.

When still is suspected, hash the AI found stills against known stills

Obviously different websites allow uploading of different formats/codexes which is a filter in itself and frame sizes

Because of editing, an AI needs to scan the film, when a still is found, make a copy and hash it to see if there is a match.

At least thats how I would do it.

I think you can spend too much time looking at other stuff AI & training data and sometimes its the background objects which are more important, like clothing, jewellery & furniture, but these new background filters seen on Zoom calls also makes it harder for the AI stills recognition.

Expensive in terms of cpu processing, but probably a better way to spend money than on human wages going through stuff/moderating.

You should look into Scale-invariant feature transform.

Its beyond me now, I'm on a different more explosive short lived path, but the idea is out there for someone else to have a stab at.

How many GPUs do you have for all of this?

> At least thats how I would do it.

What's the problem exactly with a higher false positive rate there?

We're not talking about a crime enforcement unit here. We're talking about access to a platform, which doesn't owe it to anyone to provide it in the first place.

You can take down 90% of content on Pornhub without any loss to anyone; adult models would gladly take a 1% false positive rate on new uploads vs. content getting rampantly stolen and shared without consent.

The solution to revenge porn might not be platforms, but platforms can easily make this problem an order of magnitude smaller.

That's to say, platforms can be the most significant part of the solution.

> The solution to revenge porn isn't platforms, it's why isn't the ex getting called "defendant" by a judge for making the recording without permission?

I imagine the repeated uploads aren't by the ex, after the first one

> The pirates know how to defeat it,

Is this true? I never see feature-length movies on YouTube for example.

They're out there for sure. The problem is, entities that own feature length films usually have more pull (money and lawyers) to get it removed. Furthermore, other easier avenues exist for sharing movies.

https://www.reddit.com/r/fullmoviesonyoutube/

Yeah, I pity the poor soul who has to ID content for takedown and goes into it thinking "This seems to be such a simple problem to solve: record a hash of the video file".

> The solution to revenge porn isn't platforms, it's why isn't the ex getting called "defendant" by a judge for making the recording without permission?

I would be happy to see more of that (expanding 'recording' with 'posting'), but the law itself has certain protections against false positives, that we need to be careful about.

There are regularly stories in the news about judges giving slaps on the wrist for full on sexual assault. In comparison to that, what would a punishment for this kind of thing be? Absolutely nothing.

Reply on news.ycombinator.com