An unasked question is how many of the identical twins in the original study were separated at birth and raised in very different environments? The answer is not many. A large number of the separated twins were separated at an older age, or were adopted by a close family member of a similar class background.

I don't see this study as any more muddled than the original study.

Other twin studies didn't look solely at twins, but compared them with non-twins. As poster civilized said https://news.ycombinator.com/item?id=31304805:

monozygotic twins reared apart show much more similar IQs than unrelated individuals raised together, indicating that genetics are the dominant driver of IQ. https://www.gwern.net/docs/iq/1996-bouchard-2.pdf

To be clear, one of these twins suffered several concussions while the other did not

Increasing the sample size is useful to get nice statistical properties out of noisy data. That doesn't mean you cannot learn anything from tiny samples, especially when the data isn't noisy (which is the case here, since they have the same set of genes, a lot of the noise is gone).

This is not an example of p-hacking. P-hacking is when you test so many different hypotheses that some of them randomly happen to have low p-values.

Exactly, this is ridiculous...