We already have the technology to connect electronics to individual neurons and to collections of them - cochlear implants, for example. We also have the technology to keep organs alive outside of the body. We can simulate the world with enough fidelity to make CGI in movies seem realistic. None of these are advanced to the point where a BIV is feasible today, but it would take much more than I have seen so far to persuade me that they are not feasible in principle (on the other hand, if you are only arguing for the former, I agree.)

When you say an accurate simulation is indistinguishable from reality, we need to be careful about what we mean by 'reality' here. The reality here is that the BIV is a simulation, so unless the BIV knows this, its experiences are not reality.

> For the BIV to not know that it's connected to a simulation, the simulation must be physically accurate.

Why? It would be sufficient, but why would it be necessary? The BIV does not know what reality is like, and nor does it know what reality is like for those who put it in the vat. I imagine that its neural architecture has been shaped by evolution to preconfigure it to comprehend a world like ours, but we can tell that exact fidelity is not necessary because people with various congenital or neonatal sense-limiting pathologies still make very good sense of the world (Helen Keller is the go-to example.)

> If the simulation is accurate enough then the brain in the vat is extraneous.

For us, that is probably true, but there are philosophers and scientists who think there is something special about brains that makes them impossible to simulate, and so presumably would see a difference here.