P.S. If anyone is wondering how this applies to Braidpool, we HAVE to do this, because weak blocks are shares, and the valid share requirement includes a valid block requirement, so we HAVE to send weak-compact blocks for every share, and validate an entire block with every share, which might occur as fast as once per second. (Kaspa is achieving sub-second block times)
However the overhead of this could be substantial, and ideas for optimizations are welcome. BIP152 is already pretty efficient, but we have the advantage that previous weak-blocks (beads) named as parents can be expected to already be known to the receiver, and all txs must have already been mined in an ancestor weak-block.
I don’t want to put this overhead on bitcoin though, and only miners/centralized pools should do it with beefier hardware and networking. Therefore we’re planning on a completely separate p2p network for this. If some variant of weak blocks ends up on Bitcoin though, we will use it.
Cheers, – Bob