I worked a bunch with BT closer to the physical layer, and I have to say I was always impressed by how well specified and implemented things were up to the link layer. I think it works fantastically well. But unfortunately it's an invisible achievement, kind of how IT folks don't get recognized when things work; instead when people say "Bluetooth sucks" it's usually due to poor app profile implementations/specs.

As a result, saying "let's invent a better Bluetooth" contains a hidden trap: it requires reinventing that really good core, which is probably really expensive, which then would not leave much to build better app profiles and we would end up with a similar problem.

We should not be inventing a better Bluetooth but fixing the app profile certification. I'm not sure what the right solution there is.