Treat IQ as a blackbox. It doesn't matter what people think it measures or what it actually measures. Those things are psychological not psychometric.

So let's consider it psychometrically:

1. Does it measure something? Yes it does. This something is consistent across time as people's IQs do not vary dramatically. It is "stable". So whatever it measures is not spurious.

2. Ok so we measure an actual thing. Is the thing it measures distinct? Yes it is. It's been controlled for against every conceivable variable like education, socioeconomic background, and family health. Whatever it is measuring, it is definitely an independent thing, not a representation of something else.

3. Is the real independent thing we're measuring even important? Maybe it's just like fingerprints, real but essentially random and inconsequential. It turns out it correlates with all kinds of things. It acts as a predictor in all kinds of outcomes. It is the strongest predictor available in psychometry. If you could only administer one test to a subject, testing this thing is the most revealing test.

So, regardless of what IQ is, we know what it isn't. We know it isn't fake, unimportant, random, or a combination of other factors.

This is deeply unpopular and goes against many people's core beliefs about society and education, but the science is well-founded.

It can be refuted but it has to be refuted with science, not quotes, metaphors, and anecdotes.