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

Suspicious Discontinuities

danluu.com

Post reply on HN
Print view View original

131–140 of 307 posts

Re: Suspicious Discontinuities

#131

Post by wahern »

tqi wrote: post #15
lolc wrote: post #9

Earlier quoted context omitted.

Please do not jump to conclusions and read the footnote which addresses that point explicitly: "drug seizure records, amounts seized do not appear to be the cause of this change."

I do not have access to the original paper, but author's assessment that the bunching is due to prosecutorial discretion is a little puzzling to me. If it is true that prosecutors are choosing to prosecute cases that are above the cutoff at a higher rate than those below, I would have expected everything above 280g to increase, rather than a solitary spike at 280g?

From Note 1:

> prosecutors can charge people for amounts that are not the same as the amount seized and then notes

If someone is caught with 500g you can simply charge 280g and make your life easier. If you charge 500g the defendant will argue it was just 450g, which even if irrelevant makes the case seem weaker to the jury--death by a thousand cuts. But if you charge 280g and present evidence of 500g, it makes the case seem more like a slam-dunk to the jury.

Re: Suspicious Discontinuities

#132

Post by Dylan16807 »

lmm wrote: post #114
SilasX wrote: post #82

Earlier quoted context omitted.

Relatedly, I learned about Universal Credit, an attempt by the UK to merge all their social benefits into one coherent system, and it seems like the cliff problem here is worse. From the BBC description: https://www.bbc.com/news/uk-41487126 >Under the old system many faced a "cliff edge", where people on a low income would lose a big chunk of their benefits in one go as soon as they started working more than 16 hours…

63% is not necessarily so bad. Income tax rates for the well off went as high as 95% at some points in UK history, and those do not seem to have been disastrous (indeed they seem to be some of the country's finest years). And beyond that you're letting perfect be the enemy of good here. Eliminating the cliffs and capping the marginal tax rate of the poor at 63% is valuable progress. If you want things to get better y…

Bringing up the marginal tax rates of the rich is a pretty weak defense of a >50% marginal rate for someone on welfare.

Yes, it's 'valuable progress'.

That doesn't mean complaints are invalid, or "letting perfect be the enemy of good". What a rubbish accusation. You're only letting perfect be the enemy of good when you use a problem as a reason to impede the weaker solution. This is a comment hoping for better. Don't shame that!

Re: Suspicious Discontinuities

#133

Post by knzhou »

There's a possibility that a discontinuity in data analysis might have an impact on fundamental physics.

Many dark matter detection experiments rely on detecting an annual modulation, i.e. an excess of detection events later in the year, when the Earth would be going into the 'dark matter wind'. This helps one distinguish the desired signal from backgrounds that don't change throughout the year, like cosmic rays or ambient radioactivity.

One particular experiment has been claiming a positive detection for two decades, leading to a lot of confusion in the field. It was recently argued [0] that this is due to a discontinuity in their data analysis. The team might be analyzing their annual modulation by subtracting out the average rate in each year, leading to a discontinuity at each year boundary. So if their detection rate is slowly going up, e.g. due to detector aging, it would look like an annual modulation, where the rate is lower in the beginning of the year and higher at the end!

0: https://arxiv.org/abs/2002.00459

Re: Suspicious Discontinuities

#134

Post by akvadrako »

crooked-v wrote: post #118
SilasX wrote: post #82

Earlier quoted context omitted.

Relatedly, I learned about Universal Credit, an attempt by the UK to merge all their social benefits into one coherent system, and it seems like the cliff problem here is worse. From the BBC description: https://www.bbc.com/news/uk-41487126 >Under the old system many faced a "cliff edge", where people on a low income would lose a big chunk of their benefits in one go as soon as they started working more than 16 hours…

With this system you're guaranteed to never end up with less total money than you started with by getting a better-paying job, which makes it definitionally not a cliff. It's just a gradual reduction in benefits as you earn more.

It’s not a cliff and sounds like a step in the right direction, but it’s still effectively a 63%+whatever marginal tax rate.

Re: Suspicious Discontinuities

#135

Post by chillacy »

tjr225 wrote: post #77
dragonwriter wrote: post #75

Earlier quoted context omitted.

> The perverse incentives around these discontinuities is one of the worst misfeatures of programs intended to help people. This is a key motivation for replacing many of those programs with UBI funded by progressive taxes.

Either that or a more progressive stance on socialist policies that provide basics to everyone. At least with socialist style policies there isn't the possibility of those spending their UBI on junk food or wine our whatever, but I really would prefer either to the current state of affairs.

Regarding the "basics to everyone" for programs, they're typically labeled "universal", and there are a number of advantages to universal programs:

1. They tend to be more resilient to budget cuts since everyone is bought in.

2. They don't have the discontinuities mentioned in the article.

Regarding UBI, studies I've seen show that people tend to spend their money on "responsible" things (as opposed to drugs and lotto tickets): https://www.businessinsider.com/stockton-basic-income-trial-...

Re: Suspicious Discontinuities

#136

Post by lmm »

Dylan16807 wrote: post #132
lmm wrote: post #114

Earlier quoted context omitted.

63% is not necessarily so bad. Income tax rates for the well off went as high as 95% at some points in UK history, and those do not seem to have been disastrous (indeed they seem to be some of the country's finest years). And beyond that you're letting perfect be the enemy of good here. Eliminating the cliffs and capping the marginal tax rate of the poor at 63% is valuable progress. If you want things to get better y…

Bringing up the marginal tax rates of the rich is a pretty weak defense of a >50% marginal rate for someone on welfare. Yes, it's 'valuable progress'. That doesn't mean complaints are invalid, or "letting perfect be the enemy of good". What a rubbish accusation. You're only letting perfect be the enemy of good when you use a problem as a reason to impede the weaker solution. This is a comment hoping for better. Don't…

> You're only letting perfect be the enemy of good when you use a problem as a reason to impede the weaker solution.

If your response to the weak solution is to talk about how bad it is, you're setting the wrong incentives.

Re: Suspicious Discontinuities

#137

Post by chillacy »

dragonwriter wrote: post #75
btilly wrote: post #74

The perverse incentives around these discontinuities is one of the worst misfeatures of programs intended to help people. Here is a real example. My niece wound up with 3 kids, divorced, with inconsistent child support payments. So she went on food stamps. She was looking for a better job. She found one, interviewed, and they wanted to hire her. But the salary that they offered was $0.50/hour too much for her to stay…

> The perverse incentives around these discontinuities is one of the worst misfeatures of programs intended to help people. This is a key motivation for replacing many of those programs with UBI funded by progressive taxes.

A Negative Income Tax which phases out would work too (though I think a UBI would be more popular and thus resilient to budget cuts, as Alaska has shown).

The wonderful thing about cash (UBI, NIT, EITC) is that it's easy to divide up and low-cost to distribute.

Re: Suspicious Discontinuities

#138

Post by peteretep »

Dylan16807 wrote: post #129
peteretep wrote: post #122

Earlier quoted context omitted.

How else would you see this working, exactly? You've got to make sure people who need the benefits and can't work get supported, you need to not discourage people from working (and most people _do_ want to work), and you also can't just (afford to) unilaterally extend benefits to a whole bunch of people who don't currently get them. So, concrete proposal on how you'd structure this?

The exact same way as the new system, except instead of 63% it's a much smaller number. Edit: Why are people downvoting this? I don't understand what's so confusing about SilasX's argument. Removing cliffs and changing it to a smooth dropoff is good. But that dropoff doesn't have to be £.63 in benefits per £1 earned. That's a pretty big effective marginal tax rate! Why not £.50 or £.30? It should be extremely easy to…

> The exact same way as the new system, except instead of 63% it's a much smaller number.

If it's less than 63%, you "unilaterally extend benefits to a whole bunch of people who don't currently get them".

Let's say benefits for the unemployed are £1,000 a month (chosen to be an easy number to work with). You lose 63p of benefits per extra pound you earn. The government gives you (some) benefits up to the first £1,587 you earn. If you go to 50%, the government gives people earning up to £2,000 a month benefits, and so on, which probably includes a bunch of people who don't currently get any.

You can't change 63% without either paying much more in the way of benefits overall, or without reducing the benefits of those who can't do any work.

> It would technically bring more people into the system, but they'd be at the top end, and therefore only be receiving a small fraction of the current levels of benefits

That's not how the maths works. Assuming an even distribution of people at each wage level, changing this to 30% effectively doubles overall spend.

Re: Suspicious Discontinuities

#139

Post by PhantomGremlin »

iron0013 wrote: post #84
thrower123 wrote: post #78

Earlier quoted context omitted.

Government paternalism is terrifying. People are going to screw up, and make bad choices, but having choices is better than being treated like a child your entire life.

Speaking of children, that’s who’s gonna starve if you decide to let them suffer the consequences of their parents’ poor choices. I would have personally starved to death dozens of times over if not for food stamps.

I would have personally starved to death dozens of times over if not for food stamps.

I understand the point you're making, but you're more than hyperbolic about it. You can personally starve to death only once, and that single time would be pretty tragic by itself.

Re: Suspicious Discontinuities

#140

Post by chillacy »

taneq wrote: post #127
toomuchtodo wrote: post #52

Earlier quoted context omitted.

This is depressing to people who expect the dating and sexual marketplace to be some sort of meritocracy, or equality to be exhibited. It's more of a free market than some care to admit. The gritty truth is: people are depreciating assets, and some lose their value (perceived or actual) faster than others. One of the founders of Okcupid (a data scientist) wrote a book, Dataclysm [1], using data from Okcupid's platfor…

> some sort of meritocracy What definition of merit are you using that it's not? > most people don't describe themselves as racist, but a surprisingly high number of people are Calling people racist for preferring to date a particular race is just silly, you might as well call people sexist for preferring to date a particular gender.

It really depends on the reason to prefer to date a particular race.

A: If it's aesthetic, sure (though one might ponder to what extent euro-centric beauty standards are indicative of inherent attractiveness vs a desire to fit into the current dominant culture: fair skin is now very popular world-wide, skin lightening is a big thing in Asia, the most popular plastic surgery tends to try to replicate european features, etc)

B: if it's because you think someone else's genes are inferior to yours... that might be the actual definition of racism. In his final book Where Do We Go From Here, MLK calls this out:

> Often white liberals are unaware of their latent prejudices. A while ago I ran into a white woman who was anxious to discuss the race problem with me. She said: “I am very liberal. I have no prejudices toward Negroes. I believe Negroes should have the right to vote, the right to a good job, the right to a decent home and the right to have access to public accommodations. Of course, I must confess that I would not want my daughter to marry a Negro.” This lady could not see that her failure to accept intermarriage negated her claim to genuine liberalism. She failed to see that implicit in her rejection was the feeling that her daughter had some pure, superior nature that should not be contaminated by the impure, inferior nature of the Negro. It is the Teutonic Origins theory warmed over. The question of intermarriage is never raised in a society cured of the disease of racism.

Post reply on HN