notes

questions:

Does b^d chain need to know b^{d+1} blockheaders?


If a reorg on b^d makes b^{d+1} invalid then we can to a first order guarantee
proper ordering of events.


* Mining

Always done with a full suite of blocks from from b^0 to b^d

"Boxed" difficulties: if PoW is valid for a given chain difficulty it must be
the best known block.

Enforce alternating pattern so that 51% attack resources *can't* be brought to
bear on an isolated part of the chain. Issue: How much worse are consensus
requirements? Orphan rate will go up a bit due to boxed difficulties. Or maybe
not an issue?


Note how all transactions are now two phase commits: mark coins as spent
followed by create new outputs.

preface with jason cronk's talk on legality and mining

make sure to get andrew miller's ideas on formalizing what kind of proof we get
out of all this