Hacker News Reader

Comment by s1artibartfast | Hacker News Reader

This whole anti-IQ meme is intellectually bankrupt. IQ is just a measure of relative intelligence on a specific scale. We know that some people are higher than others on this scale. We know that the scale falls on a normal distribution. We call one standard deviation 15 points.

By definition, 1 in 30,000 people will have an IQ of 160 or higher, which is simply 4 standard deviations above the mean.

View on HN

We can all easily recognize lower IQ-levels. But anything higher than our own is an abstraction and if used against me an insult.

Except in Software Engineering. One day you just realize, that some fucking dude can see the problem and its solution much better and faster. And you realize there is nothing you can do, except fake the comprehension and use fancy words.

This has nothing to do with IQ. IQ itself means nothing other than you manage to score higher on an IQ test relative to others who take the same test. The problem is (and as the article points out) that this doesn't actually measure anything about your intelligence or problem solving abilities.

It's in a sense the same way getting As in college tells you that they are a good student, but only correlates with other factors. Einstein was a B+ student in college but obviously wasn't a B+ physicist.

Feel free to use any two letter acronyms for those perceived qualities.

GD for General Dumbiness might be good, because it recognizes the subjectiveness of the measure.

Maybe he was a B+ physicist and we’ve just never heard from a A- physicist yet…

It’s literally a problem solving test. It doesn’t require existing knowledge so it’s the only test which actually is based just on intelligence. You can’t memorise quick problem solving.

It isn't true that it doesn't require existing knowledge. A lot of it is either pattern recognition, or group theory. Both can be learned about / trained for.

No, tests such as Raven's progressive matrices, which is the predominant method nowadays, don't require any more knowledge then what a normal baby is born with.

They were in fact designed to avoid the testing problems arising from varying cultures, prior knowledge, etc...

> IQ itself means nothing other than you manage to score higher on an IQ test relative to others who take the same test.

This has been proven again and again to not be true. IQ is highly correlated with many measures of success, it absolutely mean something other than just your ability to score well on the test.

IQ is a number that measures something. We do some retrofitting to make it normal. In theory, 1 in 30,000 people will have an IQ of 160 or higher as you say. But I think that's giving too much credit to the measure, and to g. Which is just working memory times application of logic or something like that. I personally think IQ is intellectually bankrupt.

You still have to admit that one person in 30,000 people that would be the highest, even if the method isn't perfect.

No - 1 in however many will take the test once, get lucky a few times, and test at 4 deviations above the mean.

In reality -- if they were to take the take the test a few more times, they'd quickly regress (closer) to the mean -- at probably more like 2 deviations above the mean.

Just because the notion of IQ seems to have some validity and can be roughly applicable in some situations -- doesn't mean that it hasn't also morphed (by overuse, overly credulous belief in its fundamentals) into -- just another bullshit psychometric.

And an especially wobbly indicator of anything, especially at higher octaves.

Why is it intellectually bankrupt? The IQ test was specifically designed to test for a _lack_ of intelligence, having a 160 IQ doesn't really mean much as the test wasn't setup to measure above average intelligence.

Do you have a specific criticism of anything in the article?

Yes, the author treats high IQ as some extraordinarily rare unicorn event, when in fact you would expect to if find reliably if you look at enough people. e.g. "Your IQ isn't 160. No one's is."

They think Einsteins IQ was around 120, which they think is "Indeed very high!". An IQ of 120 is 10% of the general population. If you sort people, an IQ of 120 might be on the low end for some groups/teams.

If you put 1000 people in a room, you would expect to find 4 "geniuses" with IQ over 140.

They point to data that the average IQ in one Phd program had an average IQ of 140 "doesn't make sense".

In fact it is almost exactly what you would expect. 1% of americans get PhDs, 1% of americans have IQ over 135, so 140 is about what you would expect.

The way they approach IQ suggests they dont appreciate how a normal distribution works.

You're assuming that IQ sees linear growth. Getting from 100 to 110 is much easier than 140-150. It's not just "10"

It is literally a normal distribution, not linear. This means that the higher you go, the number of people above a threshold drop off in a non-linear way.

I don't think you understand what IQ is.

This isn't really correct. How they decide what an IQ of 160 is, is by get thousands of people together, asking them questions and seeing what percentage of the population get which questions right or wrong.

Questions that are tagged 160 are irrelevant because they norming group is WAYYY to small to say 1 in 30,000 people will get this correct. It's simply not possible to come up with such a question. They questions that get you 160 on IQ score are almost certainly questions that no one in the norming group got correct.

So no, there's zero proof that 1 in 30,000 people have an IQ of 160. There simply haven't been that many IQ test administered to the general population to determine that.

Testing problems can make it hard to tell which specific people in a room of geniuses have 160 IQ, but by definition the smartest fraction of a percent of people in the world have 160 IQ. The only way that group doesn't exist is if that percent doesn't exist, as in IQ hits a brick wall at 140 or 150 and higher is impossible.

Reply on news.ycombinator.com