Hacker News Reader

Comment by voidUpdate | Hacker News Reader

IQ is calibrated so that the average IQ is 100. If the average IQ is going down, something has gone very wrong somewhere

View on HN

IQ is a relative measure not an absolute one. 100 IQ in 2026 can be massively different in absolute terms to 100 IQ in 1956.

If you look at old test papers - the lack of intelligence in modern times is very stark.

Yes, but the comment was about IQ, not absolute intelligence. The median intelligence is, by definition, 100 IQ, so the median IQ cannot be decreasing

There's always a lag, you calibrate the test based on previous results, not future ones. If intelligence is decreasing I'd expect every IQ test to keep scoring a median lower than 100 unless taken immediately after calibration. Once it stops decreasing they'd converge again.

The median IQ of successive generations may absolutely be falling

Look up the Flynn effect. You're right in the first paragraph, but wrong about the direction.

The Flynn effect ended several years ago and started reversing.

Even if that turns out to be true, rather than it starting to hit a ceiling where we can expect it to not consistently go in a single direction it does not alter anything I wrote, which addressed a comparison between 1956 and 2026.

The claim of a "lack of intelligence in modern times" is flat out contradicted by the data. It'd take a lot of regression to get back to 1956 levels.

IQ is a measure that tests how well people can do IQ tests, I don't think it's an objective measure at all.

People love to say this, but the whole point of IQ is that it happens to correlate well with many other, more practical measures of cognitive capability. That’s why it is used.

Other things also correlate well with many practical measures of cognitive ability, but you wouldn't be so quick to say that they're privileged the way IQ is, so there must be more to the explanation.

IQ is privileged because it's easily quantifiable by definition, testable with minimal setup and (assuming a modern test) more or less universal without cultural biases.

I mean it's certainly convenient! But that's a weak epistemic claim. Obviously, the cultural loading of g and of IQ tests is a complicated subject; I certainly wouldn't summarize it as "universal without cultural biases".

I wonder why is that, what the point of having a score that you can not compare them without external context (the day of taking exam)

IQ gets manually recalibrated downwards when people get stupider so it's mean 100 by construction.

Hmm, who does the calibration, and how scientific are they about it? Can we check their work?

I am not too deep into the science of it, but this video gives a good outline about IQ in general.

https://www.youtube.com/watch?v=FkKPsLxgpuY

I feel like this post unironically proves the point.

Reply on news.ycombinator.com