[foobarian]

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.

[spicybright]

Granted, I know very little about this, but I wonder if you could wipe the slate above the link layer for a "new protocol".

Devices all have bluetooth chips already, which is probably the hardest part of adoption.

Then you can define a new system on top, and even swap out the bottom if this new "red tooth" catches on.

Ofc you still have a massive uphill battle, particularly getting manufacturers to agree to a more sane spec than what we have currently (which I wouldn't hold my breath on, that's for sure)

[stinkledinkle]

Also the android bluetooth stack has always been trash, with it varying per vendor on how bad it got. It's getting better, but still.