Given that it is impossible to measure heritability directly, various estimation methods are the best that can be done. You can estimate the heritability of any trait, with or without twins. Applying those methods to geographic features will uniformly indicate that they are heritable.
No it can't. Twins certainly do differ in intelligence, personality, income, level of education, patent count, adult SES, or pretty much everything which could plausibly be lumped into a term like 'giftedness'.
> Given that it is impossible to measure heritability directly, various estimation methods are the best that can be done.
??? Is it possible to measure anything 'directly', under your redefinitions of words?
> Applying those methods to geographic features will uniformly indicate that they are heritable.
They would indicate that there are between-family genetic differences, certainly. They would not indicate that they are heritable and it would be wrong or misleading to criticize heritabilities drawn from twin studies on this basis of gene-environment correlation. You can't simply lump together every kind of genetic pattern under your redefinition of the term 'heritability'.
This article is a reasonable summary of the issues with heritability:
https://academic.oup.com/ije/article-pdf/35/3/525/2195546/dy...
The tl;dr is that the people who trot out additive models of heritability as evidence of anything in particular are by and large those invested in the notion of genetic determinism.
No, it's not.
> Likewise, attendance to a given school is also heritable.
No, it's not.
Think about it for a moment, just in the twin study framework. You have monozygotic and dzygotic twins, and twice the difference in their correlation is the heritability. How correlated are the addresses of two dyzgotic twins? Well, they're going to live in... the same house. Just like monozygotic twins are. The exception will be cases of divorce, but the brokenness of their family will have nothing to do with their twinship & genetic relatedness and they will be about as likely to be split up either way (which is minimal). Since there is no difference, twice the difference is still zero, and therefore, the heritability will be zero. Or if that doesn't convince you, consider an adopted child; what address do they live at compared to their adoptive siblings, and what school do they go to?
There certainly will be between-family differences, which will be genetic in the parents, but that falls elsewhere into shared-environment (because all the kids in a family will share the same parents putting them into a better or worse zip code/school) or possibly passive gene-environment correlation.