[nextos]

I don't think this is entirely true.

Most heavy-tailed distributions make CLT convergence slow, and heavy-tailedness is relatively common.

Many methods in finance and pharmacology, just to give two examples, have had spectacular failures because they ignored that point.