News for hackers. Stuff that matters.
Re:The optimal amount of fraud is non-zero (2022) by GuB-42 on Sunday January 07, 2024 @07:03PM (#38906563)
Just like everything. The optimal amount of crime, unemployment, accidents, deaths, etc... is not zero.

For example, when building a road, there is a certain chance that an accident will happen, that some people will die, and that could have been prevented. For example, by installing a guard rail, enforcing speed limits, or by taking a different path.

But installing that guard rail for a one in a million chance that something bad happens is money better spent elsewhere, like improving safety where it matters more, and people tend to dislike the resulting taxes. Enforcing speed limits have a cost too, that can be recovered from the fines that result from it. But the goal is not to bankrupt your citizens with fines, and constant surveillance is not very popular. And the different path you are planning may go though people homes, relocating people is also expensive, and usually not very popular for the people in question.

So, we tolerate a few accidents and deaths over a dystopian society.

Re:The optimal amount of fraud is non-zero (2022) by mlyle on Sunday January 07, 2024 @07:50PM (#38906907)
> But installing that guard rail for a one in a million chance that something bad happens is money better spent elsewhere

Just for clarity -- guard rails, where we choose to place them, probably cost $500k to $2M per life saved over their lifetime, while the value of a statistical life in the US is >$10M. This comparison ignores costs of non-fatal outcomes (injuries, disability).

It seems like we should be deploying more guardrails, even though their marginal return would be less than our current average.

Re:The optimal amount of fraud is non-zero (2022) by kevindamm on Sunday January 07, 2024 @08:12PM (#38907056)
The problem the article poses is actually not one of diverting costs, despite how GP's metaphor would indicate... It's that with fraud, only the most draconian systems can be effective at stopping all fraud.

In the above metaphor, it's more like guard rails are already in place, the speed limit is 1/10th a reasonably safe velocity, and the only way you can stop the remaining death-every-decade is to make everyone walk the road instead of driving it, and guarded so that you don't encounter any strangers along the way either. Very safe. But now nobody even wants to take this road. But it's so safe! Zero deaths, no injuries, ever.

Sure, the expenses/value argument can be made, too, but that's not ultimately what makes it nonzero. Even if you had limitless resources to apply to the anti-fraud, the only way you're getting nonzero fraud is if a large amount of legitimate customers are inconvenienced or outright denied as well. This is due to how easily fraudsters can still find processes and marks to make it worth their time, and regulations + policies are ever evolving to keep up, but a lot of it comes down to a cost/benefit analysis by the business. The internet just scales this up by several orders of magnitude too.

Re:The optimal amount of fraud is non-zero (2022) by mlyle on Sunday January 07, 2024 @08:15PM (#38907074)
Sure, I was commenting just on guard rails. If you want to know what I think about the topic of the article, my earlier comment is more informative.

https://news.ycombinator.com/item?id=38906842

Re:The optimal amount of fraud is non-zero (2022) by yardstick on Monday January 08, 2024 @02:37AM (#38909157)
> while the value of a statistical life in the US is >$10M

Take the wealthiest top 10,000 Americans out of the equation, and what’s the value of a statistical life in the US?

$10M seems to be a weird number in my mind.

Edit: Well that was a fascinating rabbit hole: https://en.m.wikipedia.org/wiki/Value_of_life#:~:text=In%20W....

Re:The optimal amount of fraud is non-zero (2022) by robertlagrant on Monday January 08, 2024 @05:53AM (#38910386)
> Take the wealthiest top 10,000 Americans out of the equation

They are much less likely to be the recipient of all the money spent on entitlements, no? A giant bucket of money is spent on regular Americans.

Re:The optimal amount of fraud is non-zero (2022) by bigger_inside on Monday January 08, 2024 @06:16AM (#38910516)
what? no. Government money is a direct pipe to transnational corporations and the billionaires that run it. Welfare payments are peanuts compared to it.
Re:The optimal amount of fraud is non-zero (2022) by robertlagrant on Monday January 08, 2024 @06:20AM (#38910545)
You're confusing two things. If the government pays for healthcare, that is for people, but goes to organisations to implement what's being paid for.
Re:The optimal amount of fraud is non-zero (2022) by bigger_inside on Monday January 08, 2024 @06:25AM (#38910576)
like trillion dollar bank bailouts, subsidies awarded through ruinous competition among jurisdictions, trillions and trillions of dollars of defense spending for which the government will invent endless wars.. like that?
Re:The optimal amount of fraud is non-zero (2022) by robertlagrant on Monday January 08, 2024 @07:01AM (#38910789)
If you like, you can look at the data[0] I'm looking at. 15% on defence; 67% on health and related; education; social security; etc.

[0] https://fiscaldata.treasury.gov/americas-finance-guide/feder...

Re:The optimal amount of fraud is non-zero (2022) by jokethrowaway on Monday January 08, 2024 @08:22AM (#38911541)
Sure, let's assume 100B

What's the value provided to citizens by spending 67B on health in the USA in terms of value provided? 10B in a country where prices are not the result of insurance companies colluding with the government?

Re:The optimal amount of fraud is non-zero (2022) by robertlagrant on Thursday January 11, 2024 @05:00AM (#38950098)
I'm replying to this:

> Government money is a direct pipe to transnational corporations and the billionaires that run it. Welfare payments are peanuts compared to it.

I don't see the point in hypotheticals. You're restating the politics you've absorbed over years of consuming a certain slant of media. There's no point in that, and no point in me doing the same.

Re:The optimal amount of fraud is non-zero (2022) by concordDance on Monday January 08, 2024 @03:24AM (#38909433)
I would like to emphasise that the parent has said "where we choose to place them", so they're not contradicting the grandparents metaphor, just giving an interesting FYI.
Re:The optimal amount of fraud is non-zero (2022) by gwern on Sunday January 07, 2024 @07:57PM (#38906945)
End-to-end principle in action. You could try to make every stage super-reliable and fraudproof... but that yields a terrible overall system compared to a much lighter touch and going after complete transactions which turn out bad.
Re:The optimal amount of fraud is non-zero (2022) by tgv on Monday January 08, 2024 @03:05AM (#38909326)
> The optimal amount of [...] deaths [...] is not zero.

That's not true. It assumes that there is an objective, known price for a life. Since you can't have e.g. 0.1 deaths, there's no guarantee that the optimal amount is at least 1. It's more that it's the outcome of haggling, politics, and the willingness to turn a blind eye to the consequences of stupidity.

Re:The optimal amount of fraud is non-zero (2022) by concordDance on Monday January 08, 2024 @03:22AM (#38909418)
There's a price for life in a given society though. Due to resources being finite and wellbeing mattering. In the UK I believe NICE uses £30k/QALY as the price.
Re:The optimal amount of fraud is non-zero (2022) by robertlagrant on Monday January 08, 2024 @06:00AM (#38910414)
Yep, although short term end of life years are deemed more valuable (e.g. if you have 3 months to live and we can get you another 3 months, we'll pay more for those 3 months)[0]. And the number is a bit flexible based on other factors, I think.

[0] https://www.tandfonline.com/doi/full/10.1080/14737167.2017.1...

Re:The optimal amount of fraud is non-zero (2022) by tremon on Monday January 08, 2024 @08:51AM (#38911871)
£30k/QALY

nit: pretty sure that should be £30k * QALY.

Re:The optimal amount of fraud is non-zero (2022) by concordDance on Monday January 08, 2024 @10:14AM (#38913079)
Imagine 10 km/h, the more hours you have drive the more kilometers you've driven. Similarly, with £/QALY the more QALYs you've paid for the more £ it will have cost.
Re:The optimal amount of fraud is non-zero (2022) by nox101 on Monday January 08, 2024 @04:45AM (#38909918)
Maybe that's a poor way of stating the problem. In a zero tradeoff world, 0 deaths would be optimal. We don't live in that world though. For example, flying is supposedly safer than driving. Airplane crashes happen. We could demand zero deaths from the planes which would drive the costs up, causing more people to choose to drive, causing more deaths.

So, in relation to airplanes, we can't choose zero deaths in the real world. We have to choose some imperfect trade off, otherwise we'll cause more deaths.

Re:The optimal amount of fraud is non-zero (2022) by michaelt on Monday January 08, 2024 @05:53AM (#38910383)
> Since you can't have e.g. 0.1 deaths

True, but you can have 0.1 deaths per year, by having 1 death per 10 years.

Of course, I agree there are some things that don't work well for that - nobody's going to be happy accepting "1 nuclear meltdown per 30 years" even if statistically it'd mean fewer deaths than coal.

Re:The optimal amount of fraud is non-zero (2022) by ncruces on Monday January 08, 2024 @04:40AM (#38909874)
You could totally avoid that accident by not building the road in the first place, which typically is the first level of paralysis we need to overcome.
Re:The optimal amount of fraud is non-zero (2022) by konschubert on Monday January 08, 2024 @07:50AM (#38911225)
That being said: GPS based speed limiters are cheap, effective and not dystopian as long as they don't send telemetry back home.

> But the goal is not to bankrupt your citizens with fines,

There is this one crazy trick to avoid speeding tickets. Cops hate it.

Re:The optimal amount of fraud is non-zero (2022) by thworp on Monday January 08, 2024 @08:22AM (#38911538)
> GPS based speed limiters are cheap, effective [...]

While this is technically true (the devices themselves are cheap and effective), the data required for them to work well doesn't exist. If the data exists at all, it is usually horribly out of date. So in practice there will be many cases where these devices limit you to the 30mph from last year's big construction project. Then you get rear-ended because nobody expects you to slow down 50mph for no reason at all.

Re:The optimal amount of fraud is non-zero (2022) by JeremyNT on Monday January 08, 2024 @10:58AM (#38913835)
I feel like this "bad data" problem would disappear immediately if there were real world consequences for it being incorrect.

Which is to say, if you were to build out a system that limits speed based on some authoritative database of speed limits, then suddenly there's an incentive to make sure that database is actually correct (where there was no such incentive before).

Re:The optimal amount of fraud is non-zero (2022) by konschubert on Monday January 08, 2024 @08:24AM (#38911568)
Okay, then let's just ignore construction projects and other temporary speed reductions.

Maybe let's start with residential streets only, using the residential speed limit. And let's say only those residential streets that are at least 100m away from a faster street.

I'm sure it will be possible to find a solution that eliminates false positives.

And in the worst case - it's just about acceleration. It's not like the car would abruptly break.

Re:The optimal amount of fraud is non-zero (2022) by konschubert on Monday January 08, 2024 @08:27AM (#38911597)
> Then you get rear-ended because nobody expects you to slow down 50mph for no reason at all.

Even in this rare fictional scenario, I don't agree that the costs outweigh the benefits. There are many vehicles that cannot go above 30 kmh and somehow they don't get rear-ended either. Plus, mid-term, the other cars will also have the speed limiter installed.