Hacker News new | past | comments | ask | show | jobs | submit login
Fluctuations in silver prices can predict the probability of terrorist attacks? (twitter.com/benmarrow)
122 points by ninetax on Oct 8, 2019 | hide | past | favorite | 56 comments



I'm going to go out on a limb here and say that they talked to absolutely no one local (Pakistani or Muslim) before they published this. The threshold before people pay zakat is like $300 so fluctuations in silver price has very low effect. Do you think someone with $300 in savings is giving donations?

Government only deducts the zakat from the checking account. I am from Pakistan and very few people(basically salaried employees for corps/govt which is a very small portion of population) have bank accounts and the main use is to just withdraw the salary immediately every month. People who are on the threshold of $300 lifetime savings have almost negligible chance of owning a bank account.So I am going to call BS on this paper. There are so many problems with the data in this paper that I can't even. Also quite racist to say factually that a lot of charity money goes to fund terrorist attacks and the proof for this is nada. They are using metrics for terrorist attacks in Pakistan (most of which used to happen indiscriminately in public places in metros where most people with bank accounts live). Do you really think that people living in metros are bankrolling indiscriminate terrorist attacks against themselves?


There is a cottage industry within academia for this kind of sophistry around terrorism.

I studied IR at a uni with a fairly good research reputation, one of the hot projects (funded to the tune of 7 figures) at the time (some time ago now) was the physical identification of terrorists i.e. by various body measurements...I am 100% serious. And this guy was not an idiot, in the conventional sense at least, PHd in Biology from Oxbridge, "genius", who (at the time) was talked about in breathless tones of wonder by the science departments (and ridiculed everywhere else).

Every few years a scientist will say: "Terrorism isn't solved...but I have a formula"...stop, we know why terrorism occurs but it isn't as simple as your calculus homework.

And I agree, it is totally unfounded to suggest that charities are funding terrorism (or that people who would give to charities and somehow be unaware of that fact, if it was the case).


I think it may be possible to get some signal from body measurements. Skin color, sex organs, teeth health (wealth proxy), scars/injuries (been in fights before?), maybe you could find some marker of chronic stress.

But yeah it's going to be pretty poor and not really actionable.


The "primary" indicator was length of ring finger relative to index finger, as in - https://www.sciencemag.org/news/2019/06/talk-hand-scientists... - at the time, this was considered cutting edge...I understand it has been mostly discredited now.

You have also misunderstood slightly. It wasn't stuff like scars, or things correlated to wealth (or race...much good that would do in Northern Ireland or the Basque...I take it you don't know much about terrorism). It was purely body measurements iirc.


>Skin color

I see we are playing with fire today


I mean, he is predicting whether someone is a terrorist... based on physical measurements. It's pretty much the definition of prejudice. What could he otherwise be doing?


I always thought all Pakistani peoples dark.

Then last week I realized two of my "Easter Europen" coworkers were in fact from Pakistan.



These are great. Imagine the kind of laws politicians would make up based on these facts.

Start by regulating the maximum length of spelling bee words, then let every japanese car be audited for motorcycle homicidal AI. And do something for the honey bee.


> the physical identification of terrorists i.e. by various body measurements

The guy was disciple of Cesare Lombroso, I guess? Ah well... that should put a swift end to the "good research reputation" of that institution.


> quite racist to say factually that a lot of charity money goes to fund terrorist attacks and the proof for this is nada

The more commonly accepted fact is that Muslim terrorist organizations have laundered most of their money through Muslim charities (or at least have done so in the past).

As you state, this doesn't necessarily imply the opposite: that most Muslim charities fund terrorism. But it doesn't not say that either. In any case, this seems more like a lapse in logic to me than overt racism.


So you claim that saying "there is no evidence for this, but I believe most Muslim charities fund terrorism" is just a simple logical error?


he is saying that 'most terrorist money comes from charities' does not imply 'most charity money goes to terrorists'


I appreciate you are pointing out the logical fallacy. Thou I’ll point out that their money is laundered through is different from their money comes from.


yeah I debated trying to bring that up, but felt it would make the point a little more complex and he's still wrong even if the money wasn't laundered


"Most X's are also Y's" does not imply "Most Y's are also X's." We would need additional information about the X and Y populations to make this stronger claim.

However, making this claim incorrectly is one of the more common logical fallacies. Given that, it seems more appropriate to attribute the error to logical fallacy than to specific racist intent, again, unless we have more information about the authors with which to make this stronger claim.


>Do you really think that people living in metros are bankrolling indiscriminate terrorist attacks against themselves?

That’s not necessarily contradictory. A different example is drug related gang violence. Money from drugs can be traced and consequential and resulting in gang violence, but obviously a given gangster isn’t going to fund a shootout targeting her or him, but obviously other targets, but in the end in aggregate can result in violence for all gangsters (and many bystanders).


The paper pretty robustly shows charitable giving is correlated with the price of silver.

That intuitively doesn’t contradict the idea that most people don’t change their giving based on the price of silver. If only the rich do they can still move the needle.


Correlation is not causality!

Also,...

Over-fitting data.

My god these things drive me crazy.

If it’s real, then he should be able to answer this question posted on Twitter by Gregory Allen Perry ( @gallenperry ):

What is the payout on a $1000 bet that a terrorist attack in will occur in Kashmir in the month of February with no less than 20 victims killed if the price of silver drops by 5% immediately prior to Pakistani tax assessment in the year before?

If he can’t answer a question at least somewhat in the same vein as that, then this is crap. If he can (within any reasonable degree of confidence), then I’m wrong.


That would be if the price of silver increases by 5% then.

According to the post if the price of silver is higher, the tax threshold increases and more people have extra money to support whatever organization they want.


Actually no, the lower the price of silver the more people are eligible to donate since the criteria to fall in that group is lower.

Despite the lower price of the silver, the "tax" is a static fiat-based % so basically more people are donating the same %. All the people that were donating 1.5% if holding a balance over $380 would still be donating it (and the same $amount) if the threshold were $360 (silver price drops), more people would be donating since now everyone between $360 and $380 would also be donating a 1.5% in addition.

If the donation was in silver then it might even itself out, but you'd have to look at a distribution curve of income to know for sure.


But if I understand correctly people donate when they don't have to pay tax.

From the twitter posts:

> Individuals who fall just below the threshold one year due the chance price of silver suddenly avoid a 2.5% gvt tax. Those people then increase their private donations to charity (since they have more disposable income + islam calls for alms-giving).

Or in clearer terms:

> ↑silver price → ↑tax threshold → ↓taxes → ↑charity → ↑terrorist financing → ↑terrorist attacks!


I stand corrected.


> Over-fitting data.

Worse: their data doesn't even fit! Just look at that https://mobile.twitter.com/benmarrow/status/1180141555044294...

How can anyone see this kind of plot and be like “Eureka!”…


I agree - I was just trying to be kind. LOL!


No one said in the twitter post that correlation is causality. There is a correlation and this is an attempt to explain the correlation.

Even if the explanation of the twitter post is far-fetched, the approach of understanding and finding periods of stress in some population around the world by trying to understand their cultures and economies is refreshing.



Damn you beat me at it :) makes you think about what unsupervised AI training can lead to... imagine not having insurance because you bought some Nutella on amazon during a full moon or any other crazy scenario



Link to the actual paper

https://t.co/s0NZ1PcpQf

Is it normal for econ papers to use the first person? Super weird to read. His Latex formatting is a mess too. Check out page 37 for a scatter plot of rock solid correlation, lol. Not going to read the full 56 page mess because that's a job for an unpaid reviewer, not an unpaid commentator.

That being said, there's a shit ton of easy hidden variables here. Mountains of them. People are going to talk about spurious correlations, which is fine, but there' s a more subtle trick here, the hidden variable.

Sunscreen use is correlated with an increase in shark attacks and increase in hot dog sales. Are sharks attracted to sunscreen? Is there a secret ingredient in hot suncreen that makes people eat hot dogs? I can come up with pretty good explanations, that pass a qucik sniff test, all day long.

The answer is of course not. More people are simply at the beach. More people --> more collisions with sharks, more hot dogs shoved down gullets.

Not familiar enough with the culture of Pakistan to point out the specific variables, but that's the guess.


Got a good chuckle “that's a job for an unpaid reviewer, not an unpaid commentator.”


Right the, checks notes... University of Chicago is known for its lack of rigor in economics.


> predict the probabilities

Yeah that's all I needed to read to know that this is BS.

On a more serious note, if charities are indeed funding terrorism, you wouldn't have data on them. Especially in Pakistan where keeping data is something only the reputable orgs do.

This "paper" is not peer-reviewed and it will never get published. Just a usual case of a smartass thinking they can justify that their correlation means something when in fact it doesn't.


Fluctuations in stork births can predict human births. Sidebar: Did you know that's why storks carry babies in cartoons?

"Correlation is not causation."


612 grams of silver is $342 today. Those who have life-savings of $342 in bank or more, get auto-taxed. At this level, we are talking about the un-banked mostly, so they don't have accounts in the first place. In other words, nearly all bank accounts get auto-taxed, so silver price fluctuations do not really affect this.

Second, try to calculate the silver price fluctuations, and calculate the ACTUAL dollar amount that becomes disposable to spend. Its peanuts.

Finally, this category of people do not give charity. They take charity.


> 612 grams of silver is $342 today

Subtle, but this is spot price - typically the price negotiated by large institutions based on large transfers. Individuals in the $300 range of silver buying pay premiums on top of that which scales down with size; a single gram bar has a higher premium and costs a lot more than a single 10 ozt bar with a lower premium. This can be observed at any precious metals trading platform, whether it's corporate sited like APMEX or community sites like reddit or dedicated forums; it would be highly inefficient to buy gram bars to fund anything - but people with $300 in life savings probably can't afford more than 1 ozt at a time. Silver spot price is pretty volatile as well, the trade tariff war has driven it up $2 or more in the last year.


The paper is not arguing that the people that go from zero tax to some tax are driving the charitable giving tax. It’s arguing about the people in higher income brackets & it shows pretty robustly that charitable giving is linked to the price of silver.


He talks about "nisaab" which is the minimum amount you must own before you are auto-taxed, and that amount, based on today's rate is $342. See tweet 3 of 9: https://twitter.com/benmarrow/status/1180140774127742979?s=2...


And here is the actual author describing the link https://twitter.com/nicolalimodio/status/1181581317596745728...

You could also read the paper. The part linking silver prices to charitable giving is straightforward.



"This page isn't working"

Anyone else have a link to the paper/data?




As I look at the page (which I believe changes randomly), I especially appreciate the 0.81 correlation between "letters in winning word of Scripps National Spelling Bee" and "number of people killed by venomous spiders".

It's a good one in two ways. First, it's nicely obvious that these things are unrelated. Second, these are two variables that fluctuate (together) over time -- a lot of the spurious correlations turn out to be two things that basically just increase or decrease more or less constantly over the period in question.


>I especially appreciate the ...

And you missed this > 0.99 one (a bit dated, still ...):

Apple iPhone sales

correlates with

People who died by falling down the stairs

https://tylervigen.com/view_correlation?id=28669


Sure, but like all the 0.99+ correlations, that one is just two lines moving in the same direction.


Now I'm having fun imagining all the various hypothetical conspiracies that are causing these correlations.


This seems extremely far-fetched, see for example comment on Twitter pointing out some flaws.

Reminds me of that company that revived millions of dollars for their software for finding hidden messages on the internet and somehow predicting terror attacks.


No one sells silver in this story. It's a benchmark currency used to set the tax baseline.

The main flaw is that the graph doesn't show a correlation of fluctuations, it just shows two high-variance metrics increasing over time.

> https://www.tylervigen.com/spurious-correlations

> 2.5 million terrorist chat board messages from the dark web

Now that's fascinating.


Sorry, removed that part before I saw your comment :(

I agree with your comment.


Relevant: In the words of master data visualist Edward Tufte, “Correlation isn’t causation, but sure is a hint.”


Not anymore. They now know that you know.


I suspect that this was meant as a low value comment but it’s pretty important here that the paper is decidedly not saying that the people giving to charitable organizations are intentionally funding terrorist attacks.


Correlation, causation. So similar, so different.


So, Silver Causes Terrorism?


The paper suggests that cheaper silver causes more terrorism, yes. Hilarious how people try to explain spurious correlations with a far-fetched chain of causation.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: