for fee estimation
It seems to me if you have a latency sensitive transaction weak blocks could provide better fee estimation as it gives you a guarantee of transactions that are actively being hashed on by miners.
I concede your point for non latency sensitive transactions though.
the idea is you only keep one or two weak blocks,
(From @instagibbs in the OP)
How many blocks should we buffer?
How could you keep more than one weak block? Would these blocks be competing chain tips or blocks built on top of one another?
I think i conceptually understand how this would work if each weak block were competing chain tips.
I don’t understand how weak blocks could be built on top of each other - especially when receiving various weak blocks from a divergent set of miners across the p2p network?
Thanks in advance for clarification 