That's not really the Sieve of Eratosthenes. The list data structure is not well suited for the algorithm. See [1] for a proper discussion. As someone who wants to write Haskell(and loves it) for code with mathematically well defined guarantees, the propagation of such examples, where the guarantees or behavior is not clear to the author, is not helpful.

[1]: https://www.cs.hmc.edu/~oneill/papers/Sieve-JFP.pdf