Live data from Hacker News

Google, Meta, others will have to explain algorithms under new EU legislation

theverge.com

211–220 of 222 posts

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#211

Earlier quoted context omitted.

We're doing credit scoring at the Danish bank i work at. One of our requirements is that the model and architecture has to be able to provide explanations for why yohr rating is whatever it is. Both to regulators, internal auditors, and customers. Personally, i thimk denying people a loan is a pretty impactful decision on peoples life. They deserve a reason.

Fellow dane here. I guess i shouldn’t be surprised, it’s just not something I’ve encountered personally yet (that I know of). Do you know if a bank is obliged to give you your credit score if you ask for it?

I actually don't know if we have to give you the final "credit score". I'm more on the technical infrastructure side. For our project though, it's my impression that we aren't aiming to provide the number to the customer (It's a ranking from 1-10, binned arbitrarily based on internal rules, so practically meaningless to any individual). The credit score can be manually overwritten by individual advisors, and may be adjusted to fit sector wide government mandated risk levels, so its use beyond internal calculations is limited.

We will rather try to provide explanations for specific decisions, so if we deny you a loan we want to be able to tell you what pieces of hard data we are basing that decision on. Try to make it actionable, instead of the opaque numbers that lead to a feeling of "the computer says no." It's still an open question of if we will be able to provide that insight.

Legally I don't know if the credit score, or any of the components (LGD and PD) count as "personal data". If it does, we are obliged to provide it. Basically no customers make those kinds of requests so there's not really any internal discussions around those questions. As far as my bank is concerned, I'm sure we'd just tell you if you asked. We're bad at secrets.

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#212

Earlier quoted context omitted.

How does a ban on personalized algorithmic feeds work if each user is subscribed to a different set of others?

"news from friends" might be ok if it is presented without algorithmic curation - i.e. strictly on time order or similar.

Strict chronological runs into the issue that some people post far more frequently than others and can flood the feed. Perhaps some minimal algorithm to up-rank infrequent posters would be helpful without being engagement-driven.

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#213
post #151
post #84

Earlier quoted context omitted.

After the last iOS update, Apple nagged the shit out of me to setup Apple Pay, for two days. No way to say ‘fuck off’ - only ‘remind me’. No obvious way to stop the nagging. Finally I gave them just the tip, and then pulled out before the money shot, and that seems to have shut them up for now.

Apple Pay is legitimately useful though. You can use it to pay at physical businesses if you forget your wallet (double click and face ID to turn your phone into a "tap to pay" card basically). There are also lots of apps/sites that support it so you don't have to type in your card number or even your shipping info sometimes.

Amazon prime is also very useful. I would say it's more useful for more people than Apple pay. But that doesn't excuse the bad behaviour.

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#214

Earlier quoted context omitted.

There are three basic reasons people don't get a loan: - They already have too much debt - They don't earn enough to pay it off - They have a history of not paying off their debts. This should not be hard to explain to most people.

> - They have a history of not paying off their debts. If only. In practice it's more like "They don't have a history of paying off their debts", which crucially means if you don't have a history of being in debt then you have no history of paying it off, and therefore you're considered high risk. Thus you get otherwise-nonsencial behaviours like taking out a loan only for the reason of paying it off.

Having to have had debt to prove you can pay it off is a weird thing that's going on in America. Here in Europe we don't really do that. If you have avoided loans most of your life that's a good thing.

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#215
post #207

Earlier quoted context omitted.

Cookie consent forms were a response to the "cookie law" passed in 2009[1]. If they're now illegal, that's on the EU for making them vague or not strict enough. But my point is that fighting this with laws is: - too slow, since by the time governments catch up that something should be done, a lot of harm has already been inflicted upon users. And by the time laws do come to pass, tech companies have grown in power an…

> Cookie consent forms were a response to the "cookie law" passed in 2009 > were a response to the "cookie law" passed in 2009[1] Your link clearly states: "Receive users’ consent before you use any cookies except strictly necessary cookies". For everything else you need to ask for consent with "No"/"Reject" being clearly labeled and being the default option. Yes, it's that easy. > too slow, since by the time governm…

I'm not saying "law is bad". I'm saying that the process of prohibiting a behavior after it happens is too slow and ineffective for Big Tech. As it was for many other industries before, but particularly now for Big Tech.

What needs to happen is for privacy-minded tech people to propose and lobby solutions to governments that make it impossible for companies to violate these rights in the first place, and then governments making it a law for this technology to be used by all companies. E.g. the DNT header could've been one such solution, but the fact it was never made part of a law is what led to it being abused for ironically tracking itself, and now abandoned altogether.

We're in this mess because governments fundamentally don't understand technology and how to police it. Either that, or they're willfully complacent with the status quo because it benefits them as much as the corporations.

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#216
post #215

Earlier quoted context omitted.

> Cookie consent forms were a response to the "cookie law" passed in 2009 > were a response to the "cookie law" passed in 2009[1] Your link clearly states: "Receive users’ consent before you use any cookies except strictly necessary cookies". For everything else you need to ask for consent with "No"/"Reject" being clearly labeled and being the default option. Yes, it's that easy. > too slow, since by the time governm…

I'm not saying "law is bad". I'm saying that the process of prohibiting a behavior after it happens is too slow and ineffective for Big Tech. As it was for many other industries before, but particularly now for Big Tech. What needs to happen is for privacy-minded tech people to propose and lobby solutions to governments that make it impossible for companies to violate these rights in the first place, and then governm…

> I'm saying that the process of prohibiting a behavior after it happens is too slow and ineffective for Big Tech

Do you realize that all laws happen after something happens? Even your proposed solution of tech people coming up with something would also happen after the fact?

> What needs to happen is for privacy-minded tech people to propose and lobby solutions to governments that make it impossible for companies to violate these rights in the first place

Ah yes, the magical technical solution that is impossible to violate.

Good thing that you mentioned DNT. Do you know that DNT ended up being used for browser fingerprinting and hence tracking?

Had DNT been codified into law, you'd be complaining on HN that the law is bad and governments don't understand technology.

> Either that, or they're willfully complacent with the status quo because it benefits them as much as the corporations.

wat. GDPR is literally aimed against the status quo. I wish it was more rigorously enforced, of course.

Also, it doesn't apply just to the web. It asserts right to privacy as a fundamental right.

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#217

Six percent annual turnover for non-compliance seems too low. Should be six percent for first offense, 12% for second, 25% for third, etc. Until the company fixes it's compliance or becomes insolvent.

This opinion is terrifying to me. Why has our culture become so authoritarian?

If an individual continues to defy the law they get sent to jail. Why should corporations get away without similar escalating consequences?

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#218
post #169

Six percent annual turnover for non-compliance seems too low. Should be six percent for first offense, 12% for second, 25% for third, etc. Until the company fixes it's compliance or becomes insolvent.

Fines are fine balance between motivational and liquidating. I think 6% is quite a lot, even if one has 40% margin. Investors will be highly distraught and seek remedies from the current management. But for instance at 20% they will blame the regulators and push the company to fight in courts. In any way, government wants to motivate change in behavior not taking companies out of business.

> In any way, government wants to motivate change in behavior not taking companies out of business.

Taking maliciously noncompliant comanies out of business can be a way to motivate others to not try to skirt the law.

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#219
Before people get overly excited about this - it will be very important to see how exactly it's worded in the legislation itself.

Anti-discrimination legislation has already made black-box algorithms illegal if they are deciding on anything that a user might take objection to - so for most use cases this is not a big change.

As for - the recommender systems will have to not be based on profiling - unless we're talking about removing recommender systems based on data altogether - it will be interesting to see what the legislation considers profiling. If I tie your recommendations to the last viewed piece of content (content contextual recommendation), is that profiling? It's arguably worse for the user and for society more than profiling recommendation. If the recommendations are based on your explicit categories is that not profiling? Yet it's the principle used in news aggregators for the last 30 years.

The wording is going to be important here.

Re: Google, Meta, others will have to explain algorithms under new EU legislation

#220
post #127

Earlier quoted context omitted.

Don't let the prose fool you. They're doing this because what they did before was in violation and the walls were closing in. This reminds me of supermarkets in Germany loudly announcing that they would abandon plastic bags to save the environment ... a few weeks before legislation came into effect banning them from selling plastic bags. Why wait until you're potentially facing fines if you can move slightly ahead an…

That's basically WAI though. Plastic bags were banned because of the environmental impact. For some particularly naive customers they might think it's a feel-good initiative from a brand they now love a little more, but at the policy level it's really not important how various implementers want to spin it. What's important is that the outcome is a reduction in plastic bag use.

Oh, I'm not saying any of these are bad. I'm just saying you shouldn't thank Google (or the supermarkets) when they try to spin this as charity. Some people might not want to hear this, but the thing doing this is not corporate volition but regulations.
Post reply on HN