[Cryptography] Internet independent Blockchain

[Matthieu CHARETTE]

 > What do you think of rejecting the double-spend that has smaller
 > proof-of-work history, replaying any lost transactions in the next
 > block, and including record of the double-spend so everyone can
 > identify the key that misbehaved?

I really like your idea. It can really be helpful.

So to summary:
- The attacker has $15 on address A0.
- He signs a transaction T1 sending $10 from A0 to address A1.
- He signs a transaction T2 sending $10 from A0 to address A2.
- He sends T1 to miner M1 and T2 to miner M2 at the same time (or at 
least before M1 and M2 sync).
- When M1 and M2 sync, the least powerful one, let say M2, will have to 
discard all blocks after T2 (including T2 too).
- M1 and M2 keep a record of T2 (they might send it to other nodes too) 
so they will never allow transactions coming from A0 anymore.

Is it a good summary?

The problem is that the attacker can just have $10 on A0. And try to 
send $10 on A1 and $10 on A2. In this case the balance of A0 will be 
$0. So the attacker doesn't care about the fact that nodes will not 
accept transactions coming from A0.

Any suggestions on this?

BR,
Matthieu