I didn't write the and test code myself so you can label that as appealing to authority all you like but I have looked into it and understand the benefits of moving the witness to a subtree of the merkle root so that it can be dropped when it is no longer needed as well as the other changes that are included in the update that allow further soft fork changes enough to see the benefits. I don't just accept someone saying it is good or bad.

Can you explain why such an obvious refactor that is backward and forward compatible isn't a clear benefit?

No it isn't sufficient and won't instantly increase the transaction throughput but the the arguments against it that I've heard are weak and almost all political and could also be argued as appeal to a different authority.

There are two visions of what people see as scaling and are pushing for, on-chain and off-chain. Segwit helps both so being against it can only be politically motivated.