Technically it’s a hubless mesh network, not peer to peer, because there’s still a distinction between controller nodes (like the Hue bridge) and device nodes (like Hue bulbs).
It’s not that each node can both cue and perform requests as in true peer to peer. Rather in a mesh network each node can pass along information, which is not quite the same thing.
Hue bulb A can’t command Hue bulb B to dim. It can only pass along the dim request that originated with a controller node like the Hue bridge.
The difference in the ZLL standard is simply that each certified device is preauthorized to join the nearest network. The join network password comes installed on the device, and any node can verify it. So you don’t need a hub to authorize a new device.
LIMITED CHANGES
The only reason anyone thought this was a good idea is because from the beginning it was intended to be limited to residential lighting. On, off, dim, and color changes. That’s pretty much it. There’s very little damage a mockingbird device could do even if it was capable of joining the network and issuing commands.
CHANGE A BULB WITHOUT REQUIRING TECH SUPPORT
The use case was simple: to allow anyone to change a bulb without requiring router access or complicated steps. Put in the new bulb, push one button, done. You can keep the bridge in a locked room and still let your teenager put in a fresh bulb when one burns out.
In a true peer to peer network, each node has the ability to command other nodes. That’s not the intent here. You still have controller nodes that you can restrict access to if you want to. Your teenage son can’t use the bulb in his room to turn off the lights in his little brother’s room. But either boy could replace a burned out bulb in his own room with a one button process.
NETWORK CHOICES
So I do think hubless mesh networks make a lot of sense for the residential lighting use case, and I think Hue’s success demonstrates it.
A true peer to peer would likely open up too many hacker options, or just malicious mischief options, for most HA use cases. So I doubt if we’ll see that anytime soon. Maybe once we get biometric authentication and super smart devices.
Of course you only need this solution to begin with if the bulb itself is a node, since it’s the replaceable part. If the node ID is limited to the receptacle, then you don’t have the burned out bulb problem. Swapping one dumb bulb with another doesn’t require rejoining the network. So it’s not clear ZLL will end up as the dominant lighting protocol, it’s not the only way to address the use case.