Boson sampling _works_. It's now on the error-correction skeptics to provide a framework where that remains true, but stops being true when a particular combination of gates is applied, because the “naive” math says it should work fine. It would require a new fundamental physical principle to be true, for a quantum system to be sensitive to the nature of the computation involved in generating the distribution being sampled from.
At toy scale, yes, just like quantum computers. And since we haven't seen any news of working boson sampling with thousands of photons I assume that it stops working when you try to scale up, just like quantum computers.
New fundamental physical principles are at this stage required in any case, quantum mechanics and general relativity don't fit together, meaning that at least one of them need to be overhauled into something that explains the same observations but doesn't have quite the same maths. The lacklustre performance of quantum computers could be a hint to what needs overhauling.
