I don't exactly know how the pure DAG approach would deal with double spends (i.e., I sign a transaction on my local PC, the transaction begins propagating the network... 20 ms later I log into some machine on the other side of the world using the same output but sending somewhere else)... now you have the same output in two different branches in the DAG, and from what was presented, I didn't pick up a mechanism to 100% say "this is the right one". I mean, perhaps a timestamp could be used.
This.
I need to hear a bulletproof ELI5 explanation for how a node-based currency will work and prevent double spends and Sybil attacks before I can even begin to consider it as a real thing.
It almost feels like this is the crypto equivilant of the Five Room House Puzzle. At first glance you think that a solution might be difficult, but possible -- but in fact it is a completely impossible mathematical task.
what if we changed blocks to have a one tx limit? would that make the blockchain impossible?
just like with a blockchain having forks and reorgs, a txnet will also have the same property. The answer is to wait more than 20 milliseconds before accepting a transaction. Without blocks doesnt mean without confirmations. As time passes, the txnet reaches eventual consensus, just like the blockchain that has more than one tx per block.
there are more overall reorgs needed as the txnet propagates each transaction instead of a block of them, but this is nothing fundamentally different in a qualitative sense. One big problem is that there wont be anyway to debate about blocksizes when there are no blocks
James
I dont see how a one tx block vs normal blocks has any significant difference as far as sybil resistance goes. each tx will be signed by the one spending the output and in a txnet once that is broadcast to the network, then it will eventually propagate. it decentralizes the block
