If you break down the necessary costs (storage, bandwidth, CPU cycles for verification), it's actually very low, at about $0.001 (a tenth of a cent) a transaction:
The only reason Bitcoin transactions cost so much is artificial scarcity of block space, which increases the proof of work generated per transaction. The absence of a static limit in Ethereum is one of its major advantages over Bitcoin.
That kind of napkin math is both dangerous and misleading. The block size limit is a security parameter, and the math above assumed that nobody was trying to attack the network. In attack situations, a larger block size is more expensive and more dangerous.
Also, the resource requirements for running a node are a lot higher than most people are willing to tolerate even at 1mb blocks. Huge portions of the incentives, security, and decentralization of the network depend on people running full validating nodes. A big misconception in cryptocurrency is that miners can set the rules of the network. That's only true if you aren't running a full node.
The dynamic block size in ethereum is an easy attack vector that can be exploited by a large miner. The miner simply increases the blocksize as much as possible, and fills blocks with autogenerated transactions to make verification more expensive for small nodes. Once you get it high enough, your competition will start dropping off the network. If home users stop running full nodes too, now it's a lot easier to change the rules of the network - instead of convincing everyone, you only need to convince everyone willing to pay for a full node. And if that's tens of thousands per month, it's not going to be many people.
>That kind of napkin math is both dangerous and misleading. The block size limit is a security parameter, and the math above assumed that nobody was trying to attack the network. In attack situations, a larger block size is more expensive and more dangerous.
The costs under an attack scenario were not what the parent comment was making a claim about. The defence costs per transaction actually decrease with each marginal increase in the number of transactions, because the number of parties worldwide with the resources capable of attacking a blockchain diminishes as the cost of a successful attack increases.
>Also, the resource requirements for running a node are a lot higher than most people are willing to tolerate even at 1mb blocks.
Again, not relevant to the issue at hand, which is the cost of the computing resources used per transaction in a distributed network.
>The dynamic block size in ethereum is an easy attack vector that can be exploited by a large miner.
https://bitcointalk.org/index.php?topic=3332.0
The only reason Bitcoin transactions cost so much is artificial scarcity of block space, which increases the proof of work generated per transaction. The absence of a static limit in Ethereum is one of its major advantages over Bitcoin.