The EU Commission accepts high error rates when checking chats • Hacker News Board

Re: The EU Commission accepts high error rates when checking chats

#371

Post by wsc981 » Tue, Aug 23, 2022, 3:45 PM UTC

ChrisKnott wrote: post #206

kurupt213 wrote: post #179

Earlier quoted context omitted.

I agree, child protection isn’t what’s driving this. It’s the deflection

What evidence do you have to support this? Everything I can see suggests that they sincerely believe this will help protect children.

The way the EU Commission handled whistleblower Roelie Post regarding child trafficking makes me think the safety of children is not the highest priority of the EU Commission. Even today she lives in fear and under a lot of pressure as can be seen on her Twitter account.

Basically the EU Commission punished her for being a whistleblower.

http://www.roeliepost.com/

https://www-vpro-nl.translate.goog/argos/lees/nieuws/2019/Ge...

Re: The EU Commission accepts high error rates when checking chats

#372

Post by coffeeling » Tue, Aug 23, 2022, 3:47 PM UTC

DangerousPie wrote: post #39

Important to note that this is not 10% of all messages being falsely flagged (= 10% false positive rate), but 10% of flagged messages being false positives (= 90% precision). As someone who works with these types of classification problems in a different context, 90% precision is actually quite good - especially assuming there is some sort of manual review process to take care of the 10%. Whether that makes this whol…

https://archive.ph/W41mf

At present the manual review process consists of "fuck you"

Re: The EU Commission accepts high error rates when checking chats

#373

Post by nullc » Tue, Aug 23, 2022, 3:53 PM UTC

thr923400230 wrote: post #168

nullc wrote: post #52

Earlier quoted context omitted.

It's easy to achieve high precision when you just define a hit very broadly. See also the political flap about teachers factually describing the existence of alternative sexuality as 'grooming'. If you want to see the lie behind any of these child protection surveillance initiatives when they talk about things like 90% precision or millions of hits per year ask them how many of those detection of vile child predators…

> It's easy to achieve high precision when you just define a hit very broadly. No, it's easy to achieve high recall when you define a hit very broadly. Precision will come down starkly.

I don't mean in the algorithm itself, I mean in your evaluation of the algorithm, where you also don't do so equivalently for recall (or don't report on recall).

Evaluate your algorithm thusly: If it made a hit, it's a grooming true positive unless its extraordinary undeniably a false positive. Absent any ground truth data you just don't evaluate recall, of if you have any test data it's only a false-negative if it's undeniably abuse. Benefit of doubt always goes to the algorithm. All hail the algorithm. All hail.

Re: The EU Commission accepts high error rates when checking chats

#374

Post by tomjen3 » Tue, Aug 23, 2022, 3:58 PM UTC

DangerousPie wrote: post #39

Important to note that this is not 10% of all messages being falsely flagged (= 10% false positive rate), but 10% of flagged messages being false positives (= 90% precision). As someone who works with these types of classification problems in a different context, 90% precision is actually quite good - especially assuming there is some sort of manual review process to take care of the 10%. Whether that makes this whol…

What that means is having some random person going over the nudes your kid sent to her boyfriend.

I wonder what kind of person would take a job like that…

Re: The EU Commission accepts high error rates when checking chats

#375

Post by monkeynotes » Tue, Aug 23, 2022, 4:01 PM UTC

paganel wrote: post #222

As an EU citizen I didn't know exactly what that Chat Control thing was, so I web searched it: > The EU wants to oblige providers to search all private chats, messages, and emails automatically for suspicious content – generally and indiscriminately. The stated aim: To prosecute child pornography. [1] Yeah, that will go down well, a central government checking our private conversations for "suspicious content". Of co…

When we put in place stuff like this to "think of the children", we tend to neglect what those children will inherit from this legislation. Every year the future for the next generation looks more and more bleak. No wonder depression and anxiety is such a huge problem.

Re: The EU Commission accepts high error rates when checking chats

#376

Post by monkeynotes » Tue, Aug 23, 2022, 4:06 PM UTC

pharmakom wrote: post #308

Dudeman112 wrote: post #283

Earlier quoted context omitted.

Am currently living in England, can confirm things they don't like didn't disappear and things have, in fact, been getting worse But hey, happy fishes and being able to flaunt that stylish passport on the massive customs queue Maybe I should get myself a job in the continent and move there...

UK government is working hard to replace all the worst aspects of the EU with homegrown alternatives.

Examples? And what about replacing all the good things the EU provided with homegrown initiatives?

Re: The EU Commission accepts high error rates when checking chats

#377

Post by rob_c » Tue, Aug 23, 2022, 4:45 PM UTC

n4r9 wrote: post #56

isaacfrond wrote: post #8

With 90% accuracy, and assuming the incidence of true grooming in random conversation is way smaller, they are setting themselves up for the base rate fallacy. https://en.wikipedia.org/wiki/Base_rate_fallacy

Good to know there's a name for this. It's pernicious. I came across it recently when someone said "if vaccines are effective then how come 60% of COVID patients in hospital have had one?".

No in this case there is an abuse of language which has led to the poisoning of the conversation somewhat.

There, the reason being the broad change in the correct definition of the word "vaccine". In former years this would be a preventative treatment that was probably 90%+ effective for multiple years for most people for almost all symptoms and almost all ability for infection to spread. The key being stopping the spread as this is why vaccines are adopted in the first place historically.

For some reason this particular treatment is given this grandiose title despite unfortunately not really working at reducing spread, and only demonstrably reduces severe symptoms for those who are clinically at risk. It's a preventative treatment which works very well and should be applauded but the problem is people are now calling it what it's not which causes problems.

Re: The EU Commission accepts high error rates when checking chats

#378

Post by coffeeling » Tue, Aug 23, 2022, 4:56 PM UTC

neurostimulant wrote: post #126

sofixa wrote: post #43

Earlier quoted context omitted.

But each improvement in communications tech (landlines, mobile phones, internet) have enabled new, faster, easier, better communication methods. Medieval criminals would have to physically meet to coordinate a coach robbery; nowadays a terrorist attack can be coordinated fully remotely and anonymously. There is no point in denying the facts - that encryption and the ease of use of the Internet enable criminals to coo…

Without encryption, criminals will have even easier time stealing personal information and breaking into accounts. Limiting encryption use may hurt certain types of criminals, but also helps another types of criminals, mainly cyber criminals.

A huge chunk of things are already encrypted in that sense - the data for all big tech companies for example is encrypted in transit, decrypted, analyzed etc. on their servers, and then encrypted again when saved to disk and not immediately used for anything. You'd have to compromise both their at rest data storage and the storage for their local decryption keys to steal useful data. That does mean that data on a provider that is thoroughly compromised is vulnerable, and that you obviously have to trust the provider itself, and judging from the recent NYT article (and Google's reputation wrt closed accounts in general) at least Google is not at all worth that trust.

End to end encryption / Trust No One encryption setups are much simpler in that regard - only you have the keys, so you don't have to worry nearly so much about the provider being compromised or just not giving a fuck about paying customers. Plus you just don't have to worry about monitor people creeping on your photos and such.

Re: The EU Commission accepts high error rates when checking chats

#379

Post by _-david-_ » Tue, Aug 23, 2022, 5:06 PM UTC

erdos4d wrote: post #255

This will never fly in Germany. Those people still have internet cafes without surveillance cameras so they can do their computing anonymously. Paranoia is a way of life there, for good reason.

How many people actually use them? If 1% of people use them that probably won't be enough to stop anything.

Re: The EU Commission accepts high error rates when checking chats

#380

Post by erdos4d » Tue, Aug 23, 2022, 5:08 PM UTC

_-david-_ wrote: post #379

erdos4d wrote: post #255

This will never fly in Germany. Those people still have internet cafes without surveillance cameras so they can do their computing anonymously. Paranoia is a way of life there, for good reason.

How many people actually use them? If 1% of people use them that probably won't be enough to stop anything.

Enough to keep one on every block open, in my experience. They really are everywhere.