Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Dutch scandal serves as a warning for Europe over risks of using algorithms

politico.eu

Post reply on HN
Print view View original

161–170 of 197 posts

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#161

Post by justin_oaks »

sfifs wrote: post #133
Kaotique wrote: post #19

Earlier quoted context omitted.

On the other hand one of the main triggers for this whole operation was the mass fraud by people from Bulgaria who netted a lot of benefits by gaming the system. They traveled to the Netherlands, applied for benefits and then went back to Bulgaria and used Dutch bank accounts and debit cards to get the cash. The tax office never wanted that to happen again and started this program to find fraudsters but ended up clas…

Perhaps the problem is setting the wrong incentives. The way to deter fraud is not necessarily to make the benefits administration overly strict but to directly dis-incetivize fraud through stiff penalties. For instace if benefits fraud had a penalty of say 10 to 20 years imprisonment if convicted, the number of people trying to commit fraud would very likely go dramatically down.

I would think "higher chance of getting caught" would be better in deterring any particular crime than "stiff penalties".

My first google search "harsh penalty deter crime" had this as the top:

>Increasing the severity of punishment does little to deter crime. Laws and policies designed to deter crime by focusing mainly on increasing the severity of punishment are ineffective partly because criminals know little about the sanctions for specific crimes.

https://www.ojp.gov/pdffiles1/nij/247350.pdf

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#162

Post by justin_oaks »

jacquesm wrote: post #26
eesmith wrote: post #17

Earlier quoted context omitted.

It's shocking to read, and reminds me of the British Post Office scandal. A €3.7 million + €2.75 million fines are a slap-on-the wrist for what happened. Even €30,000/case seems a pittance compared to its effects on people. It mentions there were "secret blacklists of people for two decades" and that "authorities [were] hiding information or misleading the parliament about the facts". I don't see how regulating algor…

Yes, and besides it is the one government institution fining another, so there is really no consequence for any of this.

It kills me when government institutions get fined as a punishment. It's just the tax payers that are being punished.

We as a society need to do better to incentivize good behavior for those in trusted government positions, and hold the individuals responsible when they betray that trust.

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#163

Post by pessimizer »

legulere wrote: post #116
wallaBBB wrote: post #4

By reading, seems that the algorithm is just a scapegoat used in the title. This is racial discriminations dating way back, purposely introduced in the algorithm, and supported for 6 years. > Having dual nationality was marked as a big risk indicator, as was a low income. >In 2020, Trouw and another Dutch news outlet, RTL Nieuws revealed that the tax authorities also kept secret blacklists of people for two decades,…

Algorithms calculating the probability of some event of a person will always be discriminatory, because the probability depends on which groups you belong to. The problem is that algorithms are an easy way to hide behind. Think of an algorithm to check wether to do a police check for a person. Statistically black people are more likely to commit crimes. The police could easily hide behind an algorithm and say that th…

> Statistically black people are more likely to commit crimes

Statistically black people are more likely to be arrested and convicted for crimes. So if you use an algorithm to determine who should be arrested and convicted, how long they should be sentenced, and likelihood of recidivism, and seed that algorithm with historical information about arrests and convictions tagged by race (even inadvertently through names or addresses), you end up permanently encoding untrue information like "black people are more likely to commit crimes."

an example from a particular class of crimes:

Study: Whites More Likely to Abuse Drugs Than Blacks

https://healthland.time.com/2011/11/07/study-whites-more-lik...

Black and white Americans use drugs at similar rates. One group is punished more for it.

https://www.vox.com/2015/3/17/8227569/war-on-drugs-racism

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#164

Post by pessimizer »

DavideNL wrote: post #128
logifail wrote: post #122

> Having dual nationality was marked as a big risk indicator Dual nationality indeed has its ups and downs. FWIW, our children have dual nationality, and on top of that, are all bilingual. On enrolling them in their first school the school insisted on tweaking the enrollment data we'd submitted to list the locally-spoken language second, and so put the non-local language first, as "mother tongue" (and yes, that's exa…

> but the school wasn't having it That's rude. I would probably consider getting the school to communicate this in writing, and then report it for fraud at some point; when the kids leave the school, or when I'd be sure it couldn't cause any issues for the kids...

> I would probably consider getting the school to communicate this in writing

You would only catch the stupidest, most pitiable people with this trap.

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#165

Post by Arrath »

Jensson wrote: post #113
cjrp wrote: post #102

Earlier quoted context omitted.

Why would the required budget vanish? People are greedy. Why claim UBI for just you when you could claim for your fictitious family of 6?

How would you manage to register 5 fake family members with the government? Governments already track its citizens in basically every country, it is really hard to trick them into thinking more people than that exists.

Its more like not registering the departures of family members from this mortal coil, and is a persistent (if relatively small scale) problem with Social Security payments in the States.

Which, one again, circles to the point that the money, time, and effort spent on fighting such low level individual fraud typically vastly out-spends the actual fraud they're trying to combat.

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#166

Post by kkfx »

That's not really an algorithm usage issue but an administrative ones. Really. The issue does not arise from an ineffective/wrong system but for the absence of quick, powerful and effective human backup.

A small example: automated procedures, if well designed tend to be really effective *in means* witch means they are good. BUT they tend to have hard to spot (or even evident, but no one care up front) loopholes and corner cases that makes not them but their use a nightmare. In those case the solution is simple: a human that pick all issues not automatically treated.

The *administrative* issue is this: those who choose automation must know it can fail and so must be prepared for a normal and regular and effective backup.

Personally if a day my country tax agency ask me 100k euros for a clear error I feel no specific scandal, anyone can makes error, but I expect to have a phone number to call where someone pick the call quickly and in little time we can sort out the issue. It's not a life-or-death thing, and it's complex, issues normally happen and that's is. If I can't call because no one answer or they answer but they are powerless or there is no quick solution than it's not the erroneous claim the scandal, but the absence of means to sort it out rapidly.

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#167

Post by jokoon »

I'm on welfare in france, and sometimes I wonder if the government is going to find some excuse to stop giving me money.

Could be "not applying to enough job offers", while most employers are always complaining about candidates not being skilled enough, which is why unemployment is so high.

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#168

Post by jacquesm »

Kaotique wrote: post #79
styluss wrote: post #63

Earlier quoted context omitted.

> mass fraud by people from Bulgaria turned out that in the period 2007-2013, 805 Bulgarians From https://nl.wikipedia.org/wiki/Bulgarenfraude

Thank you for pointing that out. I guess I should not have used the mass adjective.

The Bulgarians were very visible, but not the main reason:

https://www.vpro.nl/argos/lees/onderwerpen/artikelen/2021/bu...

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#169

Post by jacquesm »

rocqua wrote: post #50
yunohn wrote: post #49

Earlier quoted context omitted.

> The 'bulgaren-fraude' was the cause for the government to change policy (and I believe law) to be much stricter on fraud and be much more vigilant when trying to detect fraud. Whatever the stated reason was, they failed horribly at execution - by unfairly stereotyping and discriminating against innocent people. There lots of other ways to combat tax fraud.

Fully agreed. But the stated reason is still relevant, if only to check what the history and reasoning of the original plan was. Here it informs us how the tax office came to see citizens as adversaries, when that was not (or less so) their stance before 2016.

https://www.vpro.nl/argos/lees/onderwerpen/artikelen/2021/bu...

Re: Dutch scandal serves as a warning for Europe over risks of using algorithms

#170

Post by skrbjc »

pessimizer wrote: post #163
legulere wrote: post #116

Earlier quoted context omitted.

Algorithms calculating the probability of some event of a person will always be discriminatory, because the probability depends on which groups you belong to. The problem is that algorithms are an easy way to hide behind. Think of an algorithm to check wether to do a police check for a person. Statistically black people are more likely to commit crimes. The police could easily hide behind an algorithm and say that th…

> Statistically black people are more likely to commit crimes Statistically black people are more likely to be arrested and convicted for crimes. So if you use an algorithm to determine who should be arrested and convicted, how long they should be sentenced, and likelihood of recidivism, and seed that algorithm with historical information about arrests and convictions tagged by race (even inadvertently through names…

For the first article: "The study, which was published Monday in the Archives of General Psychiatry, controlled for variables like socioeconomic status because rates of severe drug problems tend to be greater amongst the poor. Despite this, Native American youth fared worst, with 15% having a substance use disorder, compared to 9.2% for people of mixed racial heritage, 9.0% for whites, 7.7% for Hispanics, 5% for African Americans and 3.5% for Asians and Pacific Islanders."

You can get into a debate about the merits of controlling for factors, but the title is somewhat misleading. They are suggesting that whites have higher rates of drug use when you control for factors like socioeconomic status, not that the hard numbers show that whites use more drugs than blacks.

The second article is literally two paragraphs with no actual data other than an infographic. I think you need some better sources if you want to make the claims you are trying to make.

Post reply on HN