back

by zug_zug·6y ago·view on hn ↗
I can't even bring myself to read this. After perusing and seeing it gets so emotional (bringing up eugenics, morality, etc), I have a hard time believing it's going to be objective.

"There is no significant statistical association between IQ and hard measures such as wealth."

So... you're saying that people with an IQ of 60 are equally wealthy as those with an IQ of 110?

I understand getting frustrated about misapplication of tests. But it's absolute that science must measure what is, not what we wish to be. Science is never unfair, never has an agenda, never backward -- it's simply a method to ascertain facts, entirely indifferent to our human affairs and feelings.

14 comments
Even as someone who agreed in large part with the conclusion before coming to this article, I find this virtually unreadable. It's just so full of vindictiveness and pettiness that it loses sight of its argument, and the images and graphs it uses aren't particularly coherent in their framing.

> you're saying that people with an IQ of 60 are equally wealthy as those with an IQ of 110?

Actually, if I understand his point correctly, he's arguing that the test "correctly" identifies the low tail of intelligence (because they would fail at any task), but is pure noise for people of average or above-average intelligence.

> ...but is pure noise for people of average or above-average intelligence.

Maybe not pure noise. He pointed out in many paragraphs that further to the right of the distribution you would find "nerds", salarymen, and other sorts of boring people that would never write such revolutionary medium articles as this one. People ill-adapted for "real life"; academics and their lot. Oh, and "suckers" who don't have the instinct, "why is he/she asking me that question?"

The impression I got from the article was not that Taleb believes that IQ is useless, but rather that all the ideal people are at the middle of the distribution. Too low and you're genuinely unintelligent. Too high and you're "in on it", one of those "lifeless bureaucrat" metrics-obsessed pocket-protector types.

My Google-fu is failing me, but I swear I've read before a quote from some business guy (I want to say Peter Drucker) that is similar to this: You don't want those with a 4.0 GPA because they will never think for themselves, and don't get those 2.0 GPA because they will just do enough to go by.
You have a point re: vindictiveness and pettiness. I clicked the link and immediately scrolled into the article failing to notice the author. After about 30 seconds I had the following reaction, “wait! Is this Taleb?” Despite the severe flaws in his prose, he makes some points that I think are worth considering. The chief one is that it’s kind of dumb to attempt a measure of total mental capacity. At best it will result in gross over generalization of the individual’s potential. So what’s the point if not to serve as a means for separating people into developmental tracks as early as possible. Put another way, IQ is simply another way people get evaluated not as individuals but as cogs in the machine. I rather like this critique of IQ and I’m appreciative for Taleb bringing it up (though I doubt he’s the first).
Two thoughts.

Firstly if that's his point, his conclusion should match. Saying there is "no correlation" is a reach.

Secondly, I'm not convinced I even believe that point. So is he saying that if I took an IQ test from nobel prize winners they would be equally distributed to a normal population sample (assuming a 100 minimum cutoff)?

Yes, in fact.

http://nautil.us/issue/65/in-plain-sight/your-iq-matters-les...

Sorry, I don't want to read that whole thing. Is there a linked journal article about IQs of nobel winners?
I agree with your criticism in that the form and tone really do get in the way of a clear communication of his point. But I don't think it negates that he does have a point, which is that IQ is "scientistic" rather than scientific.

And furthermore, not only is what you're saying ("that people with an IQ of 60 are equally wealthy as those with an IQ of 110") a disingenuous reading of his point (he actually says that IQ is closer to a step function where correlation exists below the median but not above), but so too is the tone of what you're saying. To state that science is "a method to ascertain facts, entirely indifferent to our human affairs and feelings" is not something he is even arguing, it's the other half of what you say ("science is never unfair, never has an agenda") where you indict yourself.

He says this:

```

The “best measure” charlatans: IQ is reminiscent of risk charlatans insisting on selling “value at risk”, VaR, and RiskMetrics saying “it’s the best measure”. That “best” measure, being unreliable blew them up many many times. Note the class of suckers for whom a bad measure is better than no measure across domains.

```

When you say "science is never unfair, never has an agenda, never backward", shouldn't you be saying "science SHOULD never be unfair, science SHOULD never have an agenda, science SHOULD never be backward"? History is littered with examples of scientism masquerading as science that were only later on unmasked (such as Lysenkoism, phrenology, the humors, VaR as he mentions and so on) as frauds by the purportedly "scientific" concept's inability to _work_, and only then after causing a significant amount of ignorance and damage. For what reason are we to believe IQ is not one of these? Taleb's argument is that assumptions about IQ which misinterpret it as a linear function when it is a clipped linear function injected with noise lead to erroneous conclusions. If you have any criticism of Taleb here, then by your own reasoning, shouldn't it be because his hypothesis is wrong? And shouldn't you have a compelling argument for that? What are you doing but proving his point?

I find it really naive to treat science like some kind of objective methodology that can output unbiased data. Its complete fantasy.

"Science" is much more of a set of institutions and social regard for those institutions than an objective methodology. We often just want to throw money and scientists at an issue, and get out good "science" but thats like managers hiring 50% more programmers for 50% better/more code, we know the details are what matters.

At the end of the day, "science" in the real world is a lot of professor vying for tenure, caught in publish or perish, funding coming from self-preserving institutions, and a bunch of overworked grad students. Its a lot of companies making a buck, too.

Everything is political! Everything. If you see something like science, religion, art, anything as apolitical, that probably means its just reinforcing prevailing politics.

Check out Anarchistic Epistemology if this idea is of any interest.

> Everything is political! Everything.

I think what you might be saying is something like, "politics is a fundamental part of human behavior, so all things human must at least be scrutinized through a political lens". Which is probably true. But I can also point out that gut flora are part of being human, so human interaction must be at least glanced at through a gastrointestinal lens. And to some extent, that's true as well; those microbes do influence behavior in surprising ways.

Even a though the political lens has non-zero utility in practically any case, it's still not sufficient to explain the physics of our world or even all of human behavior. There's a little red in any real-world colored object, but everything is not red.

>Everything is political! Everything. If you see something like science, religion, art, anything as apolitical, that probably means its just reinforcing prevailing politics.

Let's assume that everything is political. We could certainly argue that every concept and every thing created by human is in some way political.

But then what? If everything is, what does it mean for a thing to be "political"? That concept has simply become empty and useless.

I don't think math is political.

A scientific theory could be considered political only if you think Popper's definition is fundamentally flawed. I don't think his definition is wrong though and i have yet to see someone arguing over the basics since the 80s (nowadays its more bayesians vs frequentists).

> I find it really naive to treat science like some kind of objective methodology that can output unbiased data. Its complete fantasy.

No bayesian scientific will treat science like this.

> "Science" is much more of a set of institutions and social regard for those institutions than an objective methodology

Wrong, there is methodology. Scientific consensus is still the best approximation of the reality we have. It might be wrong, but it is wrong for the right reasons.

> We often just want to throw money and scientists at an issue, and get out good "science" but thats like managers hiring 50% more programmers for 50% better/more code, we know the details are what matters.

Pad argument, this is not comparable.

> At the end of the day, "science" in the real world is a lot of professor vying for tenure, caught in publish or perish, funding coming from self-preserving institutions, and a bunch of overworked grad students. Its a lot of companies making a buck, too.

Yes, good point.

> Everything is political! Everything. If you see something like science, religion, art, anything as apolitical, that probably means its just reinforcing prevailing politics.

I see math as apolitical. Tell me how i'm reinforcing capitalism and liberalism. If you can't, this argument is invalid.

> Check out Anarchistic Epistemology if this idea is of any interest.

Ok, now i see, why you're against the scientific method. Popper's falsification have been largely improved by epistemologist thanks or Feyerabend's critics, and he is right about the "consistency" criterion that was dumped in all fields except medecine and social science. However this epistemologic theory is old and its argument were already based on old discoveries (galileo, newton) or misrepresentation of newer ones (i think a dozen papers were written against his vision of Einstein, and i find his critics on Feynman really undeserved). If you have knowledge of any epistemologic book[0] on this theory updated with new knowledge i'd like to read it.

[0] Or any living philosopher since nowadays its more about short papers than books in the philosophical field.

Autism is related to high socioeconomic status, and has common factors with high IQ. Well, it's harder to say that autism is a "pseudoscientific swindle":

1. https://www.researchgate.net/publication/328734204_A_theory_...

2. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4927579/

The observation is that above roughly IQ 70 the test is not even correlated with SAT scores. And below 70 one does not need a test to see that.
Is there really any doubt that 80 IQ people are represented less than 110 IQ people in industries like medicine, law, or engineering?

If there's really no correlation in SAT scores, I wager that says more about that test than it does about IQ.

Read the article. This is his point about circularity this is literally one of his examples. If IQ measures test taking acumen and there are fields that are gate kept with tests, you will see a correlation with IQ.
I doubt the skill of test taking is unrelated to aptitude in those fields, or any other field in which it's important to learn a great deal of information from books and to keep that information in mind for rapid recall during high pressure situations 'in the field.' Law or medical school is certainly harder for halfwits to get through, but that doesn't mean there's anything wrong (in that respect) with that schooling. Rather, it reflects the complexity and importance of those fields.

Edit for response:

I don't know anything about the MCAT in particular. However I'm confident in saying that the class of people unable to perform well in any sort of test that requires reading and recalling large amounts of information (as learning medicine surely requires, no matter how you structure the education system for it) should find a career outside medicine.

https://web.archive.org/web/20110713001038/http://www.iapsyc...

IQ and SAT scores are correlated.

The correlation comes from low scores. Moreover, above 70 IQ is not even well correlated with itself as variability between results for the same person is like 20%.
From wikipedia

> Psychometricians generally regard IQ tests as having high statistical reliability.[10][65] Reliability represents the measurement consistency of a test.[66] A reliable test produces similar scores upon repetition

https://en.wikipedia.org/wiki/Intelligence_quotient#Reliabil...

> The correlation comes from low scores.

Look at the pictures in this study, that is absolutely not the case, and it will take only a minute to verify.

Frey, M. C., & Detterman, D. K. (2004). Scholastic Assessment or g? Psychological Science, 15(6), 373–378. doi:10.1111/j.0956-7976.2004.00687.x

Well I'd say bringing up hard problems like morality and eugenics is part of being objective. It's not like they could be ignored.
I agree with this statement; it would be an omission to discuss the misapplication of IQ tests and leave out the fact that they have been misapplied in a way that has given ammunition to eugenicists and the "field" of so-called "scientific racism."

However, the polemic style of writing the author uses does make not only hard to read, but hard to take seriously, even though I ultimately agree with what is said.

IQ is one of the few actually, and maybe THE most robust and measurable dimension in humans that psychology has come up with.

Some people are smarter than others and they are expected to rise higher in a (necessarily imperfectly) meritocratic system. The big question is whether the bottom end should starve. I think they should not starve and should have certain minimal living standards met, but they are going to be lower than the top end. You risk dividing the population trying to convince the smart to settle for less than they can get. They will simply band together and manipulate the system until they achieve it.

> IQ is one of the few actually, and maybe THE most robust and measurable dimensions in humans that psychology has come up with.

IQ is the most robust psychometric (with the Big 5 personality being the only real contender), but that is largely because psychometrics in general are not robust. It's a one-eyed man in the land of the blind kind of scenario.

The first big issue you run into with IQ is that... as far as we understand intelligence in psychology, there isn't a single number that can be reflected as IQ. The closest we come is a "g factor" that influences all other types of intelligence, and is somewhat circularly defined as the correlation between all of the different intelligence metrics.

> in general are not robust

Sure, but they are robust enough to be relevant to everyday life.

> circularly defined

g is circularly defined, which is the nature of how a latent factor is defined. But g also does correlate with school and work performance (around r = .5), so it IS related to everyday life. Though you can only make extreme decisions based on single tests. E.g. you can say they are "likely not gifted" when they score low or "likely not an imbecile" when they score high.

IQ is probably the biggest success of human psychology as a field. Personalty tests are notoriously unreliable and can be faked (hell, that is what actors do).
"Actors can fake personalities" isn't a particularly rigorous argument in favor of IQ.
The solution to this is to teach clever kids from a young age to be compassionate towards the less clever. Right now our society does a terrible job of this; hating and mocking people for being stupid is far too common. Much of it may feel justified under reasoning like "Stupid people [bullied me|voted wrong|didn't follow instructions|did something wrong and broke things] and for that reason my negative feelings towards them are valid." But clever kids should be reminded that hating their peers for being stupid won't make them any less stupid. Compassion is constructive but disdain is not.
> The solution to this is to teach clever kids from a young age to be compassionate towards the less clever.

Given that high iq is highly correlated with depression and anxiety, maybe it should be the other way around.

Not saying the education system is perfect, but even though IQ is partly determined by education, science indicates there is not much one can achieve this way.
I don't propose that compassion towards the less clever will raise their IQ. When I say that compassion is constructive, I mean that compassion paves the road towards mutual understanding and social harmony.
> IQ is one of the few actually, and maybe THE most robust and measurable dimensions in humans that psychology has come up with.

How did you come to that conclusion? That feels like a claim that would be pretty difficult to defend.

I think you have far more of an uphill climb to justify that the wealthy/poor divide is largely correlated to intelligence, rather than social systems that advantage/disadvantage certain classes of people or one's access to certain privileges (prior wealth, access to education, etc.)

I'm not saying it is largely determined by it and not claiming there are no injustices, but I'm saying it is not a negligible factor.
>The big question is whether the bottom end should starve. I think they should not starve and should have certain minimal living standards met, but they are going to be lower than the top end. You risk dividing the population trying to convince the smart to settle for less than they can get. They will simply band together and manipulate the system until they achieve it.

This implies that poverty is largely a product of being on the wrong end of the intelligence scale, which ignores the substantial impact that social and economic structures play in determining one's economic success. When looking at factors that perpetuate poverty, IQ isn't even on the map.

Science is never unfair, but statistics and psychology as fields often misrepresent science for personal gain. The article is arguing that it's a useless (or harmful) metric for what most people use it for, which isn't science-denial.
Forget the polemical style; this is just genuinely hard to read. It’s structured like his personal notes, not like something that’s supposed to be read by anyone else.
> Science is never unfair, never has an agenda, never backward

That’s the ideal of course. But History shows otherwise. Personal biases often influence research. This isn’t a knock against science, as you mentioned it’s ultimately a tool to understand the world. Human judgements, prior beliefs and reasoning is the issue here.

I'm not here to defend a lazily-edited article (you're right, it is), but why respond to the premise with an ungenerous assumption about what you assume he meant, rather than reading the article? Either it's worth discovering if your conclusion is correct, or it's not worth responding to.
> So... you're saying that people with an IQ of 60 are equally wealthy as those with an IQ of 110?

Literally the first paragraph states that it can discover extreme unintelligence.

IQ is not science.
IQ is probably more well studied than any other phenomenon in the social sciences or psychology.

Also IQ is not science, it is something that is studied scientifically.

So was astrology at one point.
Your argument applies equally strongly to every scientific finding.

I think this means that either science is poor or the argument is.

IQ seems to be more predictive at the low end than the high end. Someone with an IQ of 60 is handicapped. Someone with an IQ of 100 is normal. Above 100 the differences start to vanish into noise due to loads of other confounding factors and random chance.

I have taken several tests in my life. They have all shown above average but the differences have varied wildly from 120 all the way up to 160 or something. That is a crazy variance and really makes me feel like the high numbers are hand wavey BS.

The same is true of lots of other biological metrics. Very high cholesterol can indicate a problem. From the normal range down it's not predictive of much unless it is absurdly low. Low IQ probably indicates a neurological problem. Normal and above is... normal.

The "IQ cult" is bullshit. It seems like compensatory narcissism to me, like people who are fairly intelligent trying to clutch IQ because they are not in fact super successful. They are trying to rationalize the fact that high IQ does not automatically translate to better outcomes and many other factors dominate once your IQ hits the normal range.

The IQ cult is also deeply anti-intellectual. It neglects the value of ideas entirely. Who would you bet on? A person with a normal 100 IQ with sane rational beliefs and habits of thought or a person with a really high score whose mind is full of loony garbage? Having a big engine in the car doesn't mean anything if the driver is bad. Some of the nuttiest people I have ever met would undoubtedly get a high IQ score.

You can see a good A/B experiment on IQ vs ideas by searching for photos of North and South Korea from space. There is basically no genetic difference at all there. The difference is 100% ideas.

My personal observation is that in the realm of neurological factors motivation and focus are more important in terms of outcome than IQ beyond the normal range. I would definitely be tempted to exchange some IQ points for the dopamine system that super successful people seem to have.

> The same is true of lots of other biological metrics. Very high cholesterol can indicate a problem. From the normal range down it's not predictive of much unless it is absurdly low. Low IQ probably indicates a neurological problem. Normal and above is... normal

Autism is caused by an excess of the same factors that increase IQ, i.e. general reduction of gene expression. Well, it's harder to say that autism is a "pseudoscientific swindle":

1. https://www.researchgate.net/publication/328734204_A_theory_...

2. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4927579/

> IQ seems to be more predictive at the low end than the high end.

More complex tasks actually become more g-loaded, i.e. among very sophisticated tasks, IQ and performance are more strongly correlated than at unsophisticated tasks. The reason is presumably because unsophisticated tasks have a quick satiation in what general thinking can help at all.

Though, true, a person who cannot solve simple cognitive tasks at all can very confidently be classified as someone who won't achieve much in the economy.